diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9974 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 4968,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002012882447665056,
+      "grad_norm": 2.297874630841598,
+      "learning_rate": 9.999984004554854e-06,
+      "loss": 0.867,
+      "num_input_tokens_seen": 292832,
+      "step": 5,
+      "train_runtime": 39.7021,
+      "train_tokens_per_second": 7375.738
+    },
+    {
+      "epoch": 0.004025764895330112,
+      "grad_norm": 1.5412972986291504,
+      "learning_rate": 9.999919023234347e-06,
+      "loss": 0.7277,
+      "num_input_tokens_seen": 564944,
+      "step": 10,
+      "train_runtime": 66.4419,
+      "train_tokens_per_second": 8502.83
+    },
+    {
+      "epoch": 0.006038647342995169,
+      "grad_norm": 1.3285416449380885,
+      "learning_rate": 9.999804056972284e-06,
+      "loss": 0.6583,
+      "num_input_tokens_seen": 816128,
+      "step": 15,
+      "train_runtime": 95.0256,
+      "train_tokens_per_second": 8588.504
+    },
+    {
+      "epoch": 0.008051529790660225,
+      "grad_norm": 0.9814892211337461,
+      "learning_rate": 9.999639106918003e-06,
+      "loss": 0.6616,
+      "num_input_tokens_seen": 1099328,
+      "step": 20,
+      "train_runtime": 127.2285,
+      "train_tokens_per_second": 8640.583
+    },
+    {
+      "epoch": 0.010064412238325281,
+      "grad_norm": 1.0896984128385303,
+      "learning_rate": 9.99942417472053e-06,
+      "loss": 0.5853,
+      "num_input_tokens_seen": 1374032,
+      "step": 25,
+      "train_runtime": 153.9678,
+      "train_tokens_per_second": 8924.152
+    },
+    {
+      "epoch": 0.012077294685990338,
+      "grad_norm": 1.0662278961026128,
+      "learning_rate": 9.999159262528582e-06,
+      "loss": 0.604,
+      "num_input_tokens_seen": 1636080,
+      "step": 30,
+      "train_runtime": 178.3494,
+      "train_tokens_per_second": 9173.454
+    },
+    {
+      "epoch": 0.014090177133655395,
+      "grad_norm": 0.9930923199663517,
+      "learning_rate": 9.99884437299052e-06,
+      "loss": 0.5905,
+      "num_input_tokens_seen": 1900688,
+      "step": 35,
+      "train_runtime": 201.2774,
+      "train_tokens_per_second": 9443.129
+    },
+    {
+      "epoch": 0.01610305958132045,
+      "grad_norm": 0.9690627403707294,
+      "learning_rate": 9.998479509254341e-06,
+      "loss": 0.575,
+      "num_input_tokens_seen": 2199616,
+      "step": 40,
+      "train_runtime": 235.7208,
+      "train_tokens_per_second": 9331.448
+    },
+    {
+      "epoch": 0.018115942028985508,
+      "grad_norm": 1.0729878367285486,
+      "learning_rate": 9.998064674967647e-06,
+      "loss": 0.5676,
+      "num_input_tokens_seen": 2476496,
+      "step": 45,
+      "train_runtime": 270.2691,
+      "train_tokens_per_second": 9163.074
+    },
+    {
+      "epoch": 0.020128824476650563,
+      "grad_norm": 1.0319800558502288,
+      "learning_rate": 9.997599874277596e-06,
+      "loss": 0.5935,
+      "num_input_tokens_seen": 2761112,
+      "step": 50,
+      "train_runtime": 303.6384,
+      "train_tokens_per_second": 9093.42
+    },
+    {
+      "epoch": 0.02214170692431562,
+      "grad_norm": 1.0330630133831564,
+      "learning_rate": 9.99708511183087e-06,
+      "loss": 0.5641,
+      "num_input_tokens_seen": 3062000,
+      "step": 55,
+      "train_runtime": 332.92,
+      "train_tokens_per_second": 9197.405
+    },
+    {
+      "epoch": 0.024154589371980676,
+      "grad_norm": 1.024063001375934,
+      "learning_rate": 9.99652039277363e-06,
+      "loss": 0.5104,
+      "num_input_tokens_seen": 3329408,
+      "step": 60,
+      "train_runtime": 364.1266,
+      "train_tokens_per_second": 9143.545
+    },
+    {
+      "epoch": 0.026167471819645734,
+      "grad_norm": 1.0403150394762068,
+      "learning_rate": 9.995905722751453e-06,
+      "loss": 0.554,
+      "num_input_tokens_seen": 3591352,
+      "step": 65,
+      "train_runtime": 396.3182,
+      "train_tokens_per_second": 9061.789
+    },
+    {
+      "epoch": 0.02818035426731079,
+      "grad_norm": 0.989881460344966,
+      "learning_rate": 9.99524110790929e-06,
+      "loss": 0.5112,
+      "num_input_tokens_seen": 3864640,
+      "step": 70,
+      "train_runtime": 422.9422,
+      "train_tokens_per_second": 9137.514
+    },
+    {
+      "epoch": 0.030193236714975844,
+      "grad_norm": 1.0121773824486404,
+      "learning_rate": 9.994526554891394e-06,
+      "loss": 0.5092,
+      "num_input_tokens_seen": 4118608,
+      "step": 75,
+      "train_runtime": 455.3539,
+      "train_tokens_per_second": 9044.85
+    },
+    {
+      "epoch": 0.0322061191626409,
+      "grad_norm": 1.0623009014957763,
+      "learning_rate": 9.99376207084126e-06,
+      "loss": 0.5149,
+      "num_input_tokens_seen": 4397920,
+      "step": 80,
+      "train_runtime": 480.756,
+      "train_tokens_per_second": 9147.926
+    },
+    {
+      "epoch": 0.03421900161030596,
+      "grad_norm": 0.9593445283598526,
+      "learning_rate": 9.992947663401548e-06,
+      "loss": 0.5463,
+      "num_input_tokens_seen": 4693768,
+      "step": 85,
+      "train_runtime": 512.7989,
+      "train_tokens_per_second": 9153.233
+    },
+    {
+      "epoch": 0.036231884057971016,
+      "grad_norm": 1.0508704978399308,
+      "learning_rate": 9.992083340714016e-06,
+      "loss": 0.5106,
+      "num_input_tokens_seen": 4961600,
+      "step": 90,
+      "train_runtime": 544.8363,
+      "train_tokens_per_second": 9106.588
+    },
+    {
+      "epoch": 0.038244766505636074,
+      "grad_norm": 1.1518906280289636,
+      "learning_rate": 9.991169111419422e-06,
+      "loss": 0.5364,
+      "num_input_tokens_seen": 5248632,
+      "step": 95,
+      "train_runtime": 582.7007,
+      "train_tokens_per_second": 9007.424
+    },
+    {
+      "epoch": 0.040257648953301126,
+      "grad_norm": 1.0455583857626112,
+      "learning_rate": 9.990204984657458e-06,
+      "loss": 0.5126,
+      "num_input_tokens_seen": 5535592,
+      "step": 100,
+      "train_runtime": 613.3673,
+      "train_tokens_per_second": 9024.922
+    },
+    {
+      "epoch": 0.042270531400966184,
+      "grad_norm": 0.9647063794409773,
+      "learning_rate": 9.989190970066643e-06,
+      "loss": 0.4907,
+      "num_input_tokens_seen": 5825616,
+      "step": 105,
+      "train_runtime": 648.4131,
+      "train_tokens_per_second": 8984.421
+    },
+    {
+      "epoch": 0.04428341384863124,
+      "grad_norm": 1.025853461617905,
+      "learning_rate": 9.98812707778423e-06,
+      "loss": 0.5066,
+      "num_input_tokens_seen": 6096520,
+      "step": 110,
+      "train_runtime": 677.1808,
+      "train_tokens_per_second": 9002.795
+    },
+    {
+      "epoch": 0.046296296296296294,
+      "grad_norm": 0.9941746624599364,
+      "learning_rate": 9.987013318446114e-06,
+      "loss": 0.5259,
+      "num_input_tokens_seen": 6388368,
+      "step": 115,
+      "train_runtime": 703.6922,
+      "train_tokens_per_second": 9078.355
+    },
+    {
+      "epoch": 0.04830917874396135,
+      "grad_norm": 1.005276106932477,
+      "learning_rate": 9.985849703186711e-06,
+      "loss": 0.5149,
+      "num_input_tokens_seen": 6666624,
+      "step": 120,
+      "train_runtime": 737.6899,
+      "train_tokens_per_second": 9037.163
+    },
+    {
+      "epoch": 0.05032206119162641,
+      "grad_norm": 1.0054534660231844,
+      "learning_rate": 9.984636243638864e-06,
+      "loss": 0.5026,
+      "num_input_tokens_seen": 6936568,
+      "step": 125,
+      "train_runtime": 765.4033,
+      "train_tokens_per_second": 9062.631
+    },
+    {
+      "epoch": 0.05233494363929147,
+      "grad_norm": 0.9649854296642881,
+      "learning_rate": 9.983372951933702e-06,
+      "loss": 0.4955,
+      "num_input_tokens_seen": 7206584,
+      "step": 130,
+      "train_runtime": 796.5003,
+      "train_tokens_per_second": 9047.811
+    },
+    {
+      "epoch": 0.05434782608695652,
+      "grad_norm": 1.0014564760412348,
+      "learning_rate": 9.982059840700546e-06,
+      "loss": 0.4968,
+      "num_input_tokens_seen": 7475776,
+      "step": 135,
+      "train_runtime": 828.7164,
+      "train_tokens_per_second": 9020.91
+    },
+    {
+      "epoch": 0.05636070853462158,
+      "grad_norm": 0.9385880473188393,
+      "learning_rate": 9.980696923066768e-06,
+      "loss": 0.4931,
+      "num_input_tokens_seen": 7731464,
+      "step": 140,
+      "train_runtime": 861.9344,
+      "train_tokens_per_second": 8969.898
+    },
+    {
+      "epoch": 0.05837359098228664,
+      "grad_norm": 1.029261356856895,
+      "learning_rate": 9.979284212657658e-06,
+      "loss": 0.5025,
+      "num_input_tokens_seen": 8017968,
+      "step": 145,
+      "train_runtime": 884.1322,
+      "train_tokens_per_second": 9068.743
+    },
+    {
+      "epoch": 0.06038647342995169,
+      "grad_norm": 0.9568689832650803,
+      "learning_rate": 9.977821723596293e-06,
+      "loss": 0.5134,
+      "num_input_tokens_seen": 8291456,
+      "step": 150,
+      "train_runtime": 908.7213,
+      "train_tokens_per_second": 9124.312
+    },
+    {
+      "epoch": 0.06239935587761675,
+      "grad_norm": 1.0073839022410702,
+      "learning_rate": 9.976309470503398e-06,
+      "loss": 0.4773,
+      "num_input_tokens_seen": 8578792,
+      "step": 155,
+      "train_runtime": 944.0011,
+      "train_tokens_per_second": 9087.693
+    },
+    {
+      "epoch": 0.0644122383252818,
+      "grad_norm": 0.9920386431386734,
+      "learning_rate": 9.974747468497194e-06,
+      "loss": 0.5019,
+      "num_input_tokens_seen": 8890568,
+      "step": 160,
+      "train_runtime": 971.9993,
+      "train_tokens_per_second": 9146.682
+    },
+    {
+      "epoch": 0.06642512077294686,
+      "grad_norm": 0.9884792357109632,
+      "learning_rate": 9.973135733193249e-06,
+      "loss": 0.4943,
+      "num_input_tokens_seen": 9179704,
+      "step": 165,
+      "train_runtime": 1001.0592,
+      "train_tokens_per_second": 9169.991
+    },
+    {
+      "epoch": 0.06843800322061191,
+      "grad_norm": 1.092841343494877,
+      "learning_rate": 9.971474280704325e-06,
+      "loss": 0.4921,
+      "num_input_tokens_seen": 9461368,
+      "step": 170,
+      "train_runtime": 1027.2,
+      "train_tokens_per_second": 9210.833
+    },
+    {
+      "epoch": 0.07045088566827697,
+      "grad_norm": 1.2286594209419102,
+      "learning_rate": 9.969763127640209e-06,
+      "loss": 0.4741,
+      "num_input_tokens_seen": 9713984,
+      "step": 175,
+      "train_runtime": 1056.8353,
+      "train_tokens_per_second": 9191.577
+    },
+    {
+      "epoch": 0.07246376811594203,
+      "grad_norm": 0.8373313585482862,
+      "learning_rate": 9.968002291107557e-06,
+      "loss": 0.4759,
+      "num_input_tokens_seen": 10012384,
+      "step": 180,
+      "train_runtime": 1086.8129,
+      "train_tokens_per_second": 9212.61
+    },
+    {
+      "epoch": 0.07447665056360708,
+      "grad_norm": 1.0232386757931464,
+      "learning_rate": 9.966191788709716e-06,
+      "loss": 0.4987,
+      "num_input_tokens_seen": 10309304,
+      "step": 185,
+      "train_runtime": 1124.6164,
+      "train_tokens_per_second": 9166.952
+    },
+    {
+      "epoch": 0.07648953301127215,
+      "grad_norm": 1.0192498037984576,
+      "learning_rate": 9.96433163854655e-06,
+      "loss": 0.5044,
+      "num_input_tokens_seen": 10601024,
+      "step": 190,
+      "train_runtime": 1154.457,
+      "train_tokens_per_second": 9182.693
+    },
+    {
+      "epoch": 0.0785024154589372,
+      "grad_norm": 0.8967750445238334,
+      "learning_rate": 9.96242185921426e-06,
+      "loss": 0.4963,
+      "num_input_tokens_seen": 10843288,
+      "step": 195,
+      "train_runtime": 1184.9067,
+      "train_tokens_per_second": 9151.174
+    },
+    {
+      "epoch": 0.08051529790660225,
+      "grad_norm": 0.913336553563134,
+      "learning_rate": 9.9604624698052e-06,
+      "loss": 0.4897,
+      "num_input_tokens_seen": 11122280,
+      "step": 200,
+      "train_runtime": 1215.3143,
+      "train_tokens_per_second": 9151.772
+    },
+    {
+      "epoch": 0.08252818035426732,
+      "grad_norm": 1.0051464363210705,
+      "learning_rate": 9.958453489907673e-06,
+      "loss": 0.4868,
+      "num_input_tokens_seen": 11398920,
+      "step": 205,
+      "train_runtime": 1247.2302,
+      "train_tokens_per_second": 9139.388
+    },
+    {
+      "epoch": 0.08454106280193237,
+      "grad_norm": 0.9520523183963829,
+      "learning_rate": 9.95639493960576e-06,
+      "loss": 0.4624,
+      "num_input_tokens_seen": 11687640,
+      "step": 210,
+      "train_runtime": 1281.9313,
+      "train_tokens_per_second": 9117.213
+    },
+    {
+      "epoch": 0.08655394524959742,
+      "grad_norm": 1.004052413112944,
+      "learning_rate": 9.954286839479093e-06,
+      "loss": 0.4623,
+      "num_input_tokens_seen": 11963376,
+      "step": 215,
+      "train_runtime": 1310.3705,
+      "train_tokens_per_second": 9129.766
+    },
+    {
+      "epoch": 0.08856682769726248,
+      "grad_norm": 0.9695102563911814,
+      "learning_rate": 9.952129210602667e-06,
+      "loss": 0.5101,
+      "num_input_tokens_seen": 12245024,
+      "step": 220,
+      "train_runtime": 1343.393,
+      "train_tokens_per_second": 9114.997
+    },
+    {
+      "epoch": 0.09057971014492754,
+      "grad_norm": 0.962229278452791,
+      "learning_rate": 9.949922074546622e-06,
+      "loss": 0.4567,
+      "num_input_tokens_seen": 12524040,
+      "step": 225,
+      "train_runtime": 1369.1511,
+      "train_tokens_per_second": 9147.303
+    },
+    {
+      "epoch": 0.09259259259259259,
+      "grad_norm": 0.9824369554443164,
+      "learning_rate": 9.947665453376032e-06,
+      "loss": 0.4815,
+      "num_input_tokens_seen": 12785280,
+      "step": 230,
+      "train_runtime": 1400.5983,
+      "train_tokens_per_second": 9128.442
+    },
+    {
+      "epoch": 0.09460547504025765,
+      "grad_norm": 0.85368460319332,
+      "learning_rate": 9.945359369650673e-06,
+      "loss": 0.4564,
+      "num_input_tokens_seen": 13058144,
+      "step": 235,
+      "train_runtime": 1434.6051,
+      "train_tokens_per_second": 9102.257
+    },
+    {
+      "epoch": 0.0966183574879227,
+      "grad_norm": 0.8997471128244209,
+      "learning_rate": 9.943003846424811e-06,
+      "loss": 0.4618,
+      "num_input_tokens_seen": 13344560,
+      "step": 240,
+      "train_runtime": 1465.1679,
+      "train_tokens_per_second": 9107.871
+    },
+    {
+      "epoch": 0.09863123993558776,
+      "grad_norm": 0.9464056047380616,
+      "learning_rate": 9.940598907246968e-06,
+      "loss": 0.48,
+      "num_input_tokens_seen": 13615896,
+      "step": 245,
+      "train_runtime": 1501.0603,
+      "train_tokens_per_second": 9070.852
+    },
+    {
+      "epoch": 0.10064412238325282,
+      "grad_norm": 0.8962483000661939,
+      "learning_rate": 9.93814457615968e-06,
+      "loss": 0.4692,
+      "num_input_tokens_seen": 13907312,
+      "step": 250,
+      "train_runtime": 1529.34,
+      "train_tokens_per_second": 9093.669
+    },
+    {
+      "epoch": 0.10265700483091787,
+      "grad_norm": 0.8970724655525292,
+      "learning_rate": 9.935640877699267e-06,
+      "loss": 0.4603,
+      "num_input_tokens_seen": 14164552,
+      "step": 255,
+      "train_runtime": 1560.4007,
+      "train_tokens_per_second": 9077.51
+    },
+    {
+      "epoch": 0.10466988727858294,
+      "grad_norm": 0.9285997033654927,
+      "learning_rate": 9.933087836895575e-06,
+      "loss": 0.4722,
+      "num_input_tokens_seen": 14443752,
+      "step": 260,
+      "train_runtime": 1587.6033,
+      "train_tokens_per_second": 9097.835
+    },
+    {
+      "epoch": 0.10668276972624799,
+      "grad_norm": 0.9859648420826093,
+      "learning_rate": 9.930485479271735e-06,
+      "loss": 0.4877,
+      "num_input_tokens_seen": 14736200,
+      "step": 265,
+      "train_runtime": 1620.9205,
+      "train_tokens_per_second": 9091.254
+    },
+    {
+      "epoch": 0.10869565217391304,
+      "grad_norm": 0.8933935555310302,
+      "learning_rate": 9.927833830843912e-06,
+      "loss": 0.466,
+      "num_input_tokens_seen": 15016808,
+      "step": 270,
+      "train_runtime": 1657.4694,
+      "train_tokens_per_second": 9060.081
+    },
+    {
+      "epoch": 0.1107085346215781,
+      "grad_norm": 0.901422608365093,
+      "learning_rate": 9.925132918121031e-06,
+      "loss": 0.4919,
+      "num_input_tokens_seen": 15308392,
+      "step": 275,
+      "train_runtime": 1687.0118,
+      "train_tokens_per_second": 9074.265
+    },
+    {
+      "epoch": 0.11272141706924316,
+      "grad_norm": 1.0917968989269704,
+      "learning_rate": 9.922382768104522e-06,
+      "loss": 0.4688,
+      "num_input_tokens_seen": 15559416,
+      "step": 280,
+      "train_runtime": 1714.5165,
+      "train_tokens_per_second": 9075.104
+    },
+    {
+      "epoch": 0.11473429951690821,
+      "grad_norm": 1.061812935108647,
+      "learning_rate": 9.919583408288049e-06,
+      "loss": 0.4684,
+      "num_input_tokens_seen": 15836040,
+      "step": 285,
+      "train_runtime": 1744.45,
+      "train_tokens_per_second": 9077.956
+    },
+    {
+      "epoch": 0.11674718196457327,
+      "grad_norm": 0.9537165872143243,
+      "learning_rate": 9.916734866657233e-06,
+      "loss": 0.4808,
+      "num_input_tokens_seen": 16110584,
+      "step": 290,
+      "train_runtime": 1775.4645,
+      "train_tokens_per_second": 9074.011
+    },
+    {
+      "epoch": 0.11876006441223833,
+      "grad_norm": 0.916650519501917,
+      "learning_rate": 9.913837171689371e-06,
+      "loss": 0.4565,
+      "num_input_tokens_seen": 16386896,
+      "step": 295,
+      "train_runtime": 1800.6521,
+      "train_tokens_per_second": 9100.534
+    },
+    {
+      "epoch": 0.12077294685990338,
+      "grad_norm": 0.8896413097983231,
+      "learning_rate": 9.910890352353154e-06,
+      "loss": 0.4622,
+      "num_input_tokens_seen": 16681768,
+      "step": 300,
+      "train_runtime": 1826.484,
+      "train_tokens_per_second": 9133.268
+    },
+    {
+      "epoch": 0.12278582930756844,
+      "grad_norm": 1.047536047835029,
+      "learning_rate": 9.90789443810838e-06,
+      "loss": 0.4496,
+      "num_input_tokens_seen": 16957488,
+      "step": 305,
+      "train_runtime": 1852.5616,
+      "train_tokens_per_second": 9153.535
+    },
+    {
+      "epoch": 0.1247987117552335,
+      "grad_norm": 0.8071488982140137,
+      "learning_rate": 9.904849458905654e-06,
+      "loss": 0.486,
+      "num_input_tokens_seen": 17260552,
+      "step": 310,
+      "train_runtime": 1890.3429,
+      "train_tokens_per_second": 9130.91
+    },
+    {
+      "epoch": 0.12681159420289856,
+      "grad_norm": 0.9224613988609329,
+      "learning_rate": 9.901755445186088e-06,
+      "loss": 0.4643,
+      "num_input_tokens_seen": 17526528,
+      "step": 315,
+      "train_runtime": 1919.1621,
+      "train_tokens_per_second": 9132.385
+    },
+    {
+      "epoch": 0.1288244766505636,
+      "grad_norm": 0.8816714224692683,
+      "learning_rate": 9.898612427881002e-06,
+      "loss": 0.4747,
+      "num_input_tokens_seen": 17801096,
+      "step": 320,
+      "train_runtime": 1951.7409,
+      "train_tokens_per_second": 9120.625
+    },
+    {
+      "epoch": 0.13083735909822866,
+      "grad_norm": 1.013524769504045,
+      "learning_rate": 9.895420438411616e-06,
+      "loss": 0.5203,
+      "num_input_tokens_seen": 18086600,
+      "step": 325,
+      "train_runtime": 1979.6703,
+      "train_tokens_per_second": 9136.168
+    },
+    {
+      "epoch": 0.13285024154589373,
+      "grad_norm": 0.9633152341610259,
+      "learning_rate": 9.892179508688724e-06,
+      "loss": 0.4754,
+      "num_input_tokens_seen": 18348912,
+      "step": 330,
+      "train_runtime": 2010.2191,
+      "train_tokens_per_second": 9127.817
+    },
+    {
+      "epoch": 0.13486312399355876,
+      "grad_norm": 1.0039963734989743,
+      "learning_rate": 9.888889671112388e-06,
+      "loss": 0.445,
+      "num_input_tokens_seen": 18619192,
+      "step": 335,
+      "train_runtime": 2036.4325,
+      "train_tokens_per_second": 9143.044
+    },
+    {
+      "epoch": 0.13687600644122383,
+      "grad_norm": 1.0832401853018647,
+      "learning_rate": 9.88555095857161e-06,
+      "loss": 0.4669,
+      "num_input_tokens_seen": 18905056,
+      "step": 340,
+      "train_runtime": 2068.8702,
+      "train_tokens_per_second": 9137.865
+    },
+    {
+      "epoch": 0.1388888888888889,
+      "grad_norm": 0.862382244963817,
+      "learning_rate": 9.882163404444001e-06,
+      "loss": 0.4782,
+      "num_input_tokens_seen": 19184608,
+      "step": 345,
+      "train_runtime": 2098.8762,
+      "train_tokens_per_second": 9140.419
+    },
+    {
+      "epoch": 0.14090177133655393,
+      "grad_norm": 0.864044298283736,
+      "learning_rate": 9.878727042595447e-06,
+      "loss": 0.4543,
+      "num_input_tokens_seen": 19461712,
+      "step": 350,
+      "train_runtime": 2128.7935,
+      "train_tokens_per_second": 9142.132
+    },
+    {
+      "epoch": 0.142914653784219,
+      "grad_norm": 0.9320696362077064,
+      "learning_rate": 9.875241907379774e-06,
+      "loss": 0.4441,
+      "num_input_tokens_seen": 19740152,
+      "step": 355,
+      "train_runtime": 2160.5367,
+      "train_tokens_per_second": 9136.689
+    },
+    {
+      "epoch": 0.14492753623188406,
+      "grad_norm": 0.8487837648535018,
+      "learning_rate": 9.871708033638403e-06,
+      "loss": 0.4712,
+      "num_input_tokens_seen": 20008048,
+      "step": 360,
+      "train_runtime": 2188.3335,
+      "train_tokens_per_second": 9143.053
+    },
+    {
+      "epoch": 0.1469404186795491,
+      "grad_norm": 0.8449813828387256,
+      "learning_rate": 9.8681254567e-06,
+      "loss": 0.461,
+      "num_input_tokens_seen": 20280608,
+      "step": 365,
+      "train_runtime": 2212.8748,
+      "train_tokens_per_second": 9164.824
+    },
+    {
+      "epoch": 0.14895330112721417,
+      "grad_norm": 0.9446992309876536,
+      "learning_rate": 9.864494212380125e-06,
+      "loss": 0.4195,
+      "num_input_tokens_seen": 20529280,
+      "step": 370,
+      "train_runtime": 2239.8901,
+      "train_tokens_per_second": 9165.307
+    },
+    {
+      "epoch": 0.15096618357487923,
+      "grad_norm": 0.8916069751957232,
+      "learning_rate": 9.860814336980873e-06,
+      "loss": 0.4436,
+      "num_input_tokens_seen": 20811944,
+      "step": 375,
+      "train_runtime": 2269.4774,
+      "train_tokens_per_second": 9170.369
+    },
+    {
+      "epoch": 0.1529790660225443,
+      "grad_norm": 0.902573279691924,
+      "learning_rate": 9.85708586729051e-06,
+      "loss": 0.4566,
+      "num_input_tokens_seen": 21099408,
+      "step": 380,
+      "train_runtime": 2303.2507,
+      "train_tokens_per_second": 9160.708
+    },
+    {
+      "epoch": 0.15499194847020933,
+      "grad_norm": 0.8867322843522712,
+      "learning_rate": 9.85330884058311e-06,
+      "loss": 0.4646,
+      "num_input_tokens_seen": 21383888,
+      "step": 385,
+      "train_runtime": 2332.8596,
+      "train_tokens_per_second": 9166.385
+    },
+    {
+      "epoch": 0.1570048309178744,
+      "grad_norm": 0.8675325974577762,
+      "learning_rate": 9.849483294618174e-06,
+      "loss": 0.4672,
+      "num_input_tokens_seen": 21649424,
+      "step": 390,
+      "train_runtime": 2357.814,
+      "train_tokens_per_second": 9181.99
+    },
+    {
+      "epoch": 0.15901771336553946,
+      "grad_norm": 0.8905402068232048,
+      "learning_rate": 9.84560926764026e-06,
+      "loss": 0.4698,
+      "num_input_tokens_seen": 21936800,
+      "step": 395,
+      "train_runtime": 2388.2232,
+      "train_tokens_per_second": 9185.406
+    },
+    {
+      "epoch": 0.1610305958132045,
+      "grad_norm": 1.7325432324949936,
+      "learning_rate": 9.8416867983786e-06,
+      "loss": 0.4787,
+      "num_input_tokens_seen": 22236592,
+      "step": 400,
+      "train_runtime": 2415.6416,
+      "train_tokens_per_second": 9205.253
+    },
+    {
+      "epoch": 0.16304347826086957,
+      "grad_norm": 0.8270328203472112,
+      "learning_rate": 9.837715926046705e-06,
+      "loss": 0.4563,
+      "num_input_tokens_seen": 22511304,
+      "step": 405,
+      "train_runtime": 2443.4163,
+      "train_tokens_per_second": 9213.045
+    },
+    {
+      "epoch": 0.16505636070853463,
+      "grad_norm": 0.9386742183347965,
+      "learning_rate": 9.833696690341985e-06,
+      "loss": 0.4427,
+      "num_input_tokens_seen": 22789600,
+      "step": 410,
+      "train_runtime": 2473.4378,
+      "train_tokens_per_second": 9213.735
+    },
+    {
+      "epoch": 0.16706924315619967,
+      "grad_norm": 0.823193403541499,
+      "learning_rate": 9.829629131445342e-06,
+      "loss": 0.4448,
+      "num_input_tokens_seen": 23057896,
+      "step": 415,
+      "train_runtime": 2499.8246,
+      "train_tokens_per_second": 9223.806
+    },
+    {
+      "epoch": 0.16908212560386474,
+      "grad_norm": 1.0759155560083942,
+      "learning_rate": 9.825513290020774e-06,
+      "loss": 0.4716,
+      "num_input_tokens_seen": 23322920,
+      "step": 420,
+      "train_runtime": 2524.8992,
+      "train_tokens_per_second": 9237.169
+    },
+    {
+      "epoch": 0.1710950080515298,
+      "grad_norm": 1.2895244102227148,
+      "learning_rate": 9.821349207214965e-06,
+      "loss": 0.4635,
+      "num_input_tokens_seen": 23607720,
+      "step": 425,
+      "train_runtime": 2555.2346,
+      "train_tokens_per_second": 9238.964
+    },
+    {
+      "epoch": 0.17310789049919484,
+      "grad_norm": 1.0077840856078137,
+      "learning_rate": 9.817136924656876e-06,
+      "loss": 0.4601,
+      "num_input_tokens_seen": 23878208,
+      "step": 430,
+      "train_runtime": 2582.5663,
+      "train_tokens_per_second": 9245.923
+    },
+    {
+      "epoch": 0.1751207729468599,
+      "grad_norm": 0.9407880632968855,
+      "learning_rate": 9.81287648445733e-06,
+      "loss": 0.4771,
+      "num_input_tokens_seen": 24132696,
+      "step": 435,
+      "train_runtime": 2610.4406,
+      "train_tokens_per_second": 9244.683
+    },
+    {
+      "epoch": 0.17713365539452497,
+      "grad_norm": 0.9053142341505027,
+      "learning_rate": 9.808567929208585e-06,
+      "loss": 0.5071,
+      "num_input_tokens_seen": 24370704,
+      "step": 440,
+      "train_runtime": 2635.5968,
+      "train_tokens_per_second": 9246.75
+    },
+    {
+      "epoch": 0.17914653784219,
+      "grad_norm": 0.9661909090080892,
+      "learning_rate": 9.804211301983919e-06,
+      "loss": 0.4677,
+      "num_input_tokens_seen": 24646976,
+      "step": 445,
+      "train_runtime": 2664.2255,
+      "train_tokens_per_second": 9251.085
+    },
+    {
+      "epoch": 0.18115942028985507,
+      "grad_norm": 0.8394811999152494,
+      "learning_rate": 9.799806646337186e-06,
+      "loss": 0.4412,
+      "num_input_tokens_seen": 24928696,
+      "step": 450,
+      "train_runtime": 2692.6558,
+      "train_tokens_per_second": 9258.033
+    },
+    {
+      "epoch": 0.18317230273752014,
+      "grad_norm": 0.8442351104000755,
+      "learning_rate": 9.795354006302391e-06,
+      "loss": 0.4778,
+      "num_input_tokens_seen": 25219216,
+      "step": 455,
+      "train_runtime": 2716.0677,
+      "train_tokens_per_second": 9285.194
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 0.8146128583546326,
+      "learning_rate": 9.790853426393246e-06,
+      "loss": 0.4418,
+      "num_input_tokens_seen": 25478368,
+      "step": 460,
+      "train_runtime": 2745.2859,
+      "train_tokens_per_second": 9280.77
+    },
+    {
+      "epoch": 0.18719806763285024,
+      "grad_norm": 0.8503603324829592,
+      "learning_rate": 9.786304951602721e-06,
+      "loss": 0.4505,
+      "num_input_tokens_seen": 25769232,
+      "step": 465,
+      "train_runtime": 2776.8307,
+      "train_tokens_per_second": 9280.088
+    },
+    {
+      "epoch": 0.1892109500805153,
+      "grad_norm": 0.8906558768292534,
+      "learning_rate": 9.781708627402607e-06,
+      "loss": 0.4682,
+      "num_input_tokens_seen": 26048576,
+      "step": 470,
+      "train_runtime": 2815.5419,
+      "train_tokens_per_second": 9251.71
+    },
+    {
+      "epoch": 0.19122383252818034,
+      "grad_norm": 0.8560769187270413,
+      "learning_rate": 9.777064499743044e-06,
+      "loss": 0.4755,
+      "num_input_tokens_seen": 26328000,
+      "step": 475,
+      "train_runtime": 2848.0769,
+      "train_tokens_per_second": 9244.133
+    },
+    {
+      "epoch": 0.1932367149758454,
+      "grad_norm": 0.9460611498214977,
+      "learning_rate": 9.772372615052073e-06,
+      "loss": 0.4521,
+      "num_input_tokens_seen": 26625520,
+      "step": 480,
+      "train_runtime": 2871.8286,
+      "train_tokens_per_second": 9271.278
+    },
+    {
+      "epoch": 0.19524959742351047,
+      "grad_norm": 0.9498293368473315,
+      "learning_rate": 9.767633020235175e-06,
+      "loss": 0.4664,
+      "num_input_tokens_seen": 26912912,
+      "step": 485,
+      "train_runtime": 2905.6311,
+      "train_tokens_per_second": 9262.329
+    },
+    {
+      "epoch": 0.1972624798711755,
+      "grad_norm": 0.880158192459652,
+      "learning_rate": 9.762845762674783e-06,
+      "loss": 0.4483,
+      "num_input_tokens_seen": 27186112,
+      "step": 490,
+      "train_runtime": 2933.3314,
+      "train_tokens_per_second": 9267.999
+    },
+    {
+      "epoch": 0.19927536231884058,
+      "grad_norm": 0.9087010500522344,
+      "learning_rate": 9.758010890229841e-06,
+      "loss": 0.4425,
+      "num_input_tokens_seen": 27458000,
+      "step": 495,
+      "train_runtime": 2958.1078,
+      "train_tokens_per_second": 9282.285
+    },
+    {
+      "epoch": 0.20128824476650564,
+      "grad_norm": 0.9534094173539439,
+      "learning_rate": 9.753128451235286e-06,
+      "loss": 0.4376,
+      "num_input_tokens_seen": 27728384,
+      "step": 500,
+      "train_runtime": 2984.0872,
+      "train_tokens_per_second": 9292.082
+    },
+    {
+      "epoch": 0.20330112721417068,
+      "grad_norm": 0.8811629303030996,
+      "learning_rate": 9.748198494501598e-06,
+      "loss": 0.4464,
+      "num_input_tokens_seen": 27995208,
+      "step": 505,
+      "train_runtime": 3012.725,
+      "train_tokens_per_second": 9292.321
+    },
+    {
+      "epoch": 0.20531400966183574,
+      "grad_norm": 0.8332138484162837,
+      "learning_rate": 9.743221069314293e-06,
+      "loss": 0.4636,
+      "num_input_tokens_seen": 28282768,
+      "step": 510,
+      "train_runtime": 3050.6012,
+      "train_tokens_per_second": 9271.211
+    },
+    {
+      "epoch": 0.2073268921095008,
+      "grad_norm": 1.0108628601807634,
+      "learning_rate": 9.73819622543344e-06,
+      "loss": 0.4625,
+      "num_input_tokens_seen": 28551128,
+      "step": 515,
+      "train_runtime": 3085.8372,
+      "train_tokens_per_second": 9252.312
+    },
+    {
+      "epoch": 0.20933977455716588,
+      "grad_norm": 0.8787448272227463,
+      "learning_rate": 9.733124013093158e-06,
+      "loss": 0.4933,
+      "num_input_tokens_seen": 28811192,
+      "step": 520,
+      "train_runtime": 3119.0151,
+      "train_tokens_per_second": 9237.272
+    },
+    {
+      "epoch": 0.2113526570048309,
+      "grad_norm": 0.9083794945005091,
+      "learning_rate": 9.728004483001117e-06,
+      "loss": 0.4467,
+      "num_input_tokens_seen": 29059944,
+      "step": 525,
+      "train_runtime": 3148.6589,
+      "train_tokens_per_second": 9229.308
+    },
+    {
+      "epoch": 0.21336553945249598,
+      "grad_norm": 0.9675936515894851,
+      "learning_rate": 9.722837686338025e-06,
+      "loss": 0.4537,
+      "num_input_tokens_seen": 29359232,
+      "step": 530,
+      "train_runtime": 3177.5706,
+      "train_tokens_per_second": 9239.522
+    },
+    {
+      "epoch": 0.21537842190016104,
+      "grad_norm": 0.8837171449614337,
+      "learning_rate": 9.717623674757126e-06,
+      "loss": 0.4708,
+      "num_input_tokens_seen": 29641328,
+      "step": 535,
+      "train_runtime": 3209.9208,
+      "train_tokens_per_second": 9234.286
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 0.7986281586917634,
+      "learning_rate": 9.712362500383678e-06,
+      "loss": 0.4612,
+      "num_input_tokens_seen": 29911064,
+      "step": 540,
+      "train_runtime": 3238.9396,
+      "train_tokens_per_second": 9234.832
+    },
+    {
+      "epoch": 0.21940418679549115,
+      "grad_norm": 0.8703945257775082,
+      "learning_rate": 9.707054215814428e-06,
+      "loss": 0.4375,
+      "num_input_tokens_seen": 30189032,
+      "step": 545,
+      "train_runtime": 3272.7725,
+      "train_tokens_per_second": 9224.299
+    },
+    {
+      "epoch": 0.2214170692431562,
+      "grad_norm": 0.8211355084617955,
+      "learning_rate": 9.7016988741171e-06,
+      "loss": 0.4562,
+      "num_input_tokens_seen": 30489720,
+      "step": 550,
+      "train_runtime": 3306.9035,
+      "train_tokens_per_second": 9220.021
+    },
+    {
+      "epoch": 0.22342995169082125,
+      "grad_norm": 0.8200822471994678,
+      "learning_rate": 9.696296528829847e-06,
+      "loss": 0.4629,
+      "num_input_tokens_seen": 30755376,
+      "step": 555,
+      "train_runtime": 3326.6834,
+      "train_tokens_per_second": 9245.057
+    },
+    {
+      "epoch": 0.22544283413848631,
+      "grad_norm": 0.8793668514223952,
+      "learning_rate": 9.690847233960723e-06,
+      "loss": 0.4438,
+      "num_input_tokens_seen": 31000552,
+      "step": 560,
+      "train_runtime": 3356.3594,
+      "train_tokens_per_second": 9236.362
+    },
+    {
+      "epoch": 0.22745571658615138,
+      "grad_norm": 0.8181596604472655,
+      "learning_rate": 9.685351043987151e-06,
+      "loss": 0.455,
+      "num_input_tokens_seen": 31284360,
+      "step": 565,
+      "train_runtime": 3385.453,
+      "train_tokens_per_second": 9240.819
+    },
+    {
+      "epoch": 0.22946859903381642,
+      "grad_norm": 0.8709313822510947,
+      "learning_rate": 9.679808013855367e-06,
+      "loss": 0.4755,
+      "num_input_tokens_seen": 31590160,
+      "step": 570,
+      "train_runtime": 3420.896,
+      "train_tokens_per_second": 9234.47
+    },
+    {
+      "epoch": 0.23148148148148148,
+      "grad_norm": 0.8219481315168611,
+      "learning_rate": 9.674218198979876e-06,
+      "loss": 0.4564,
+      "num_input_tokens_seen": 31891040,
+      "step": 575,
+      "train_runtime": 3453.0585,
+      "train_tokens_per_second": 9235.592
+    },
+    {
+      "epoch": 0.23349436392914655,
+      "grad_norm": 0.7890669251783218,
+      "learning_rate": 9.668581655242896e-06,
+      "loss": 0.4195,
+      "num_input_tokens_seen": 32172592,
+      "step": 580,
+      "train_runtime": 3478.2807,
+      "train_tokens_per_second": 9249.567
+    },
+    {
+      "epoch": 0.23550724637681159,
+      "grad_norm": 0.7996915699369995,
+      "learning_rate": 9.662898438993803e-06,
+      "loss": 0.4326,
+      "num_input_tokens_seen": 32453192,
+      "step": 585,
+      "train_runtime": 3513.1942,
+      "train_tokens_per_second": 9237.517
+    },
+    {
+      "epoch": 0.23752012882447665,
+      "grad_norm": 0.8486263117718159,
+      "learning_rate": 9.657168607048564e-06,
+      "loss": 0.4261,
+      "num_input_tokens_seen": 32734816,
+      "step": 590,
+      "train_runtime": 3535.6832,
+      "train_tokens_per_second": 9258.413
+    },
+    {
+      "epoch": 0.23953301127214172,
+      "grad_norm": 0.8253459827233957,
+      "learning_rate": 9.651392216689167e-06,
+      "loss": 0.4395,
+      "num_input_tokens_seen": 33026104,
+      "step": 595,
+      "train_runtime": 3564.8102,
+      "train_tokens_per_second": 9264.477
+    },
+    {
+      "epoch": 0.24154589371980675,
+      "grad_norm": 0.9293081796114288,
+      "learning_rate": 9.645569325663055e-06,
+      "loss": 0.4538,
+      "num_input_tokens_seen": 33305592,
+      "step": 600,
+      "train_runtime": 3591.1405,
+      "train_tokens_per_second": 9274.377
+    },
+    {
+      "epoch": 0.24355877616747182,
+      "grad_norm": 0.9986733124933477,
+      "learning_rate": 9.639699992182547e-06,
+      "loss": 0.4397,
+      "num_input_tokens_seen": 33562824,
+      "step": 605,
+      "train_runtime": 3622.7033,
+      "train_tokens_per_second": 9264.58
+    },
+    {
+      "epoch": 0.24557165861513688,
+      "grad_norm": 0.8505463413562964,
+      "learning_rate": 9.633784274924247e-06,
+      "loss": 0.4345,
+      "num_input_tokens_seen": 33826536,
+      "step": 610,
+      "train_runtime": 3655.0498,
+      "train_tokens_per_second": 9254.74
+    },
+    {
+      "epoch": 0.24758454106280192,
+      "grad_norm": 0.8221420870564581,
+      "learning_rate": 9.627822233028472e-06,
+      "loss": 0.4519,
+      "num_input_tokens_seen": 34097896,
+      "step": 615,
+      "train_runtime": 3682.0434,
+      "train_tokens_per_second": 9260.59
+    },
+    {
+      "epoch": 0.249597423510467,
+      "grad_norm": 0.9572344969236803,
+      "learning_rate": 9.62181392609865e-06,
+      "loss": 0.4348,
+      "num_input_tokens_seen": 34362496,
+      "step": 620,
+      "train_runtime": 3714.5196,
+      "train_tokens_per_second": 9250.859
+    },
+    {
+      "epoch": 0.25161030595813205,
+      "grad_norm": 0.8350794602782161,
+      "learning_rate": 9.615759414200729e-06,
+      "loss": 0.4379,
+      "num_input_tokens_seen": 34597456,
+      "step": 625,
+      "train_runtime": 3741.1155,
+      "train_tokens_per_second": 9247.898
+    },
+    {
+      "epoch": 0.2536231884057971,
+      "grad_norm": 0.8594070154909677,
+      "learning_rate": 9.609658757862572e-06,
+      "loss": 0.432,
+      "num_input_tokens_seen": 34884304,
+      "step": 630,
+      "train_runtime": 3774.4814,
+      "train_tokens_per_second": 9242.145
+    },
+    {
+      "epoch": 0.2556360708534622,
+      "grad_norm": 0.8797488931891037,
+      "learning_rate": 9.603512018073357e-06,
+      "loss": 0.4543,
+      "num_input_tokens_seen": 35153336,
+      "step": 635,
+      "train_runtime": 3800.1563,
+      "train_tokens_per_second": 9250.498
+    },
+    {
+      "epoch": 0.2576489533011272,
+      "grad_norm": 0.8459516696536561,
+      "learning_rate": 9.597319256282968e-06,
+      "loss": 0.4371,
+      "num_input_tokens_seen": 35430144,
+      "step": 640,
+      "train_runtime": 3826.8369,
+      "train_tokens_per_second": 9258.336
+    },
+    {
+      "epoch": 0.25966183574879226,
+      "grad_norm": 0.8090317585805571,
+      "learning_rate": 9.591080534401371e-06,
+      "loss": 0.4655,
+      "num_input_tokens_seen": 35718240,
+      "step": 645,
+      "train_runtime": 3862.5719,
+      "train_tokens_per_second": 9247.269
+    },
+    {
+      "epoch": 0.2616747181964573,
+      "grad_norm": 0.8140029646094938,
+      "learning_rate": 9.584795914798007e-06,
+      "loss": 0.4402,
+      "num_input_tokens_seen": 36023336,
+      "step": 650,
+      "train_runtime": 3900.0652,
+      "train_tokens_per_second": 9236.598
+    },
+    {
+      "epoch": 0.2636876006441224,
+      "grad_norm": 0.803608227344892,
+      "learning_rate": 9.578465460301162e-06,
+      "loss": 0.4209,
+      "num_input_tokens_seen": 36299232,
+      "step": 655,
+      "train_runtime": 3929.5733,
+      "train_tokens_per_second": 9237.449
+    },
+    {
+      "epoch": 0.26570048309178745,
+      "grad_norm": 0.8500201951380586,
+      "learning_rate": 9.572089234197343e-06,
+      "loss": 0.423,
+      "num_input_tokens_seen": 36576512,
+      "step": 660,
+      "train_runtime": 3958.9389,
+      "train_tokens_per_second": 9238.969
+    },
+    {
+      "epoch": 0.2677133655394525,
+      "grad_norm": 0.8002504364600952,
+      "learning_rate": 9.565667300230637e-06,
+      "loss": 0.444,
+      "num_input_tokens_seen": 36865456,
+      "step": 665,
+      "train_runtime": 3988.1235,
+      "train_tokens_per_second": 9243.81
+    },
+    {
+      "epoch": 0.26972624798711753,
+      "grad_norm": 0.8813722391574977,
+      "learning_rate": 9.559199722602087e-06,
+      "loss": 0.4292,
+      "num_input_tokens_seen": 37128216,
+      "step": 670,
+      "train_runtime": 4020.8644,
+      "train_tokens_per_second": 9233.889
+    },
+    {
+      "epoch": 0.2717391304347826,
+      "grad_norm": 0.741619158334734,
+      "learning_rate": 9.55268656596903e-06,
+      "loss": 0.4603,
+      "num_input_tokens_seen": 37410888,
+      "step": 675,
+      "train_runtime": 4048.3726,
+      "train_tokens_per_second": 9240.969
+    },
+    {
+      "epoch": 0.27375201288244766,
+      "grad_norm": 0.7837319147945744,
+      "learning_rate": 9.546127895444479e-06,
+      "loss": 0.4375,
+      "num_input_tokens_seen": 37665520,
+      "step": 680,
+      "train_runtime": 4080.0755,
+      "train_tokens_per_second": 9231.574
+    },
+    {
+      "epoch": 0.2757648953301127,
+      "grad_norm": 0.7781716538239603,
+      "learning_rate": 9.539523776596446e-06,
+      "loss": 0.4516,
+      "num_input_tokens_seen": 37934576,
+      "step": 685,
+      "train_runtime": 4112.8522,
+      "train_tokens_per_second": 9223.423
+    },
+    {
+      "epoch": 0.2777777777777778,
+      "grad_norm": 0.8876134513114659,
+      "learning_rate": 9.532874275447298e-06,
+      "loss": 0.44,
+      "num_input_tokens_seen": 38210848,
+      "step": 690,
+      "train_runtime": 4142.5888,
+      "train_tokens_per_second": 9223.906
+    },
+    {
+      "epoch": 0.27979066022544286,
+      "grad_norm": 1.0087666728714288,
+      "learning_rate": 9.526179458473098e-06,
+      "loss": 0.4742,
+      "num_input_tokens_seen": 38484672,
+      "step": 695,
+      "train_runtime": 4171.4542,
+      "train_tokens_per_second": 9225.721
+    },
+    {
+      "epoch": 0.28180354267310787,
+      "grad_norm": 0.9433638020332537,
+      "learning_rate": 9.519439392602938e-06,
+      "loss": 0.4228,
+      "num_input_tokens_seen": 38773360,
+      "step": 700,
+      "train_runtime": 4197.9499,
+      "train_tokens_per_second": 9236.261
+    },
+    {
+      "epoch": 0.28381642512077293,
+      "grad_norm": 0.7475047318957952,
+      "learning_rate": 9.512654145218265e-06,
+      "loss": 0.4582,
+      "num_input_tokens_seen": 39076664,
+      "step": 705,
+      "train_runtime": 4224.7127,
+      "train_tokens_per_second": 9249.543
+    },
+    {
+      "epoch": 0.285829307568438,
+      "grad_norm": 0.8275147754782305,
+      "learning_rate": 9.505823784152222e-06,
+      "loss": 0.4208,
+      "num_input_tokens_seen": 39376712,
+      "step": 710,
+      "train_runtime": 4257.6488,
+      "train_tokens_per_second": 9248.464
+    },
+    {
+      "epoch": 0.28784219001610306,
+      "grad_norm": 0.8728787984650637,
+      "learning_rate": 9.49894837768895e-06,
+      "loss": 0.4375,
+      "num_input_tokens_seen": 39671280,
+      "step": 715,
+      "train_runtime": 4285.7816,
+      "train_tokens_per_second": 9256.487
+    },
+    {
+      "epoch": 0.2898550724637681,
+      "grad_norm": 0.823635244187005,
+      "learning_rate": 9.492027994562922e-06,
+      "loss": 0.4424,
+      "num_input_tokens_seen": 39961104,
+      "step": 720,
+      "train_runtime": 4312.1558,
+      "train_tokens_per_second": 9267.083
+    },
+    {
+      "epoch": 0.2918679549114332,
+      "grad_norm": 0.7872971284102936,
+      "learning_rate": 9.485062703958248e-06,
+      "loss": 0.4444,
+      "num_input_tokens_seen": 40248688,
+      "step": 725,
+      "train_runtime": 4345.9208,
+      "train_tokens_per_second": 9261.257
+    },
+    {
+      "epoch": 0.2938808373590982,
+      "grad_norm": 0.7870209485119529,
+      "learning_rate": 9.478052575507983e-06,
+      "loss": 0.4435,
+      "num_input_tokens_seen": 40533448,
+      "step": 730,
+      "train_runtime": 4378.5157,
+      "train_tokens_per_second": 9257.349
+    },
+    {
+      "epoch": 0.29589371980676327,
+      "grad_norm": 0.772232048901054,
+      "learning_rate": 9.470997679293437e-06,
+      "loss": 0.4491,
+      "num_input_tokens_seen": 40808632,
+      "step": 735,
+      "train_runtime": 4414.73,
+      "train_tokens_per_second": 9243.743
+    },
+    {
+      "epoch": 0.29790660225442833,
+      "grad_norm": 0.7979562892210088,
+      "learning_rate": 9.463898085843466e-06,
+      "loss": 0.4357,
+      "num_input_tokens_seen": 41074136,
+      "step": 740,
+      "train_runtime": 4449.2024,
+      "train_tokens_per_second": 9231.798
+    },
+    {
+      "epoch": 0.2999194847020934,
+      "grad_norm": 0.7946168194703691,
+      "learning_rate": 9.45675386613377e-06,
+      "loss": 0.4616,
+      "num_input_tokens_seen": 41381384,
+      "step": 745,
+      "train_runtime": 4479.5656,
+      "train_tokens_per_second": 9237.812
+    },
+    {
+      "epoch": 0.30193236714975846,
+      "grad_norm": 0.8093103137211708,
+      "learning_rate": 9.449565091586189e-06,
+      "loss": 0.4129,
+      "num_input_tokens_seen": 41677360,
+      "step": 750,
+      "train_runtime": 4512.0793,
+      "train_tokens_per_second": 9236.841
+    },
+    {
+      "epoch": 0.30394524959742353,
+      "grad_norm": 0.7941033756238285,
+      "learning_rate": 9.442331834067982e-06,
+      "loss": 0.4439,
+      "num_input_tokens_seen": 41958456,
+      "step": 755,
+      "train_runtime": 4538.833,
+      "train_tokens_per_second": 9244.327
+    },
+    {
+      "epoch": 0.3059581320450886,
+      "grad_norm": 0.7468203438579443,
+      "learning_rate": 9.43505416589111e-06,
+      "loss": 0.4476,
+      "num_input_tokens_seen": 42241672,
+      "step": 760,
+      "train_runtime": 4568.1097,
+      "train_tokens_per_second": 9247.079
+    },
+    {
+      "epoch": 0.3079710144927536,
+      "grad_norm": 0.7781812073284736,
+      "learning_rate": 9.427732159811514e-06,
+      "loss": 0.4011,
+      "num_input_tokens_seen": 42520552,
+      "step": 765,
+      "train_runtime": 4597.5076,
+      "train_tokens_per_second": 9248.609
+    },
+    {
+      "epoch": 0.30998389694041867,
+      "grad_norm": 0.8257801146294816,
+      "learning_rate": 9.420365889028392e-06,
+      "loss": 0.4158,
+      "num_input_tokens_seen": 42785184,
+      "step": 770,
+      "train_runtime": 4625.0383,
+      "train_tokens_per_second": 9250.774
+    },
+    {
+      "epoch": 0.31199677938808373,
+      "grad_norm": 0.8335089907764719,
+      "learning_rate": 9.412955427183457e-06,
+      "loss": 0.4231,
+      "num_input_tokens_seen": 43036192,
+      "step": 775,
+      "train_runtime": 4651.0601,
+      "train_tokens_per_second": 9252.986
+    },
+    {
+      "epoch": 0.3140096618357488,
+      "grad_norm": 0.8536268199607578,
+      "learning_rate": 9.405500848360209e-06,
+      "loss": 0.4444,
+      "num_input_tokens_seen": 43319440,
+      "step": 780,
+      "train_runtime": 4686.2317,
+      "train_tokens_per_second": 9243.982
+    },
+    {
+      "epoch": 0.31602254428341386,
+      "grad_norm": 1.0122938642596453,
+      "learning_rate": 9.398002227083193e-06,
+      "loss": 0.4405,
+      "num_input_tokens_seen": 43622824,
+      "step": 785,
+      "train_runtime": 4718.2054,
+      "train_tokens_per_second": 9245.639
+    },
+    {
+      "epoch": 0.31803542673107893,
+      "grad_norm": 0.8613394924717964,
+      "learning_rate": 9.390459638317255e-06,
+      "loss": 0.4537,
+      "num_input_tokens_seen": 43895160,
+      "step": 790,
+      "train_runtime": 4757.2257,
+      "train_tokens_per_second": 9227.05
+    },
+    {
+      "epoch": 0.32004830917874394,
+      "grad_norm": 0.8721494209128039,
+      "learning_rate": 9.382873157466783e-06,
+      "loss": 0.4064,
+      "num_input_tokens_seen": 44151912,
+      "step": 795,
+      "train_runtime": 4784.7113,
+      "train_tokens_per_second": 9227.707
+    },
+    {
+      "epoch": 0.322061191626409,
+      "grad_norm": 0.8371103072855253,
+      "learning_rate": 9.37524286037497e-06,
+      "loss": 0.4505,
+      "num_input_tokens_seen": 44441312,
+      "step": 800,
+      "train_runtime": 4809.465,
+      "train_tokens_per_second": 9240.386
+    },
+    {
+      "epoch": 0.32407407407407407,
+      "grad_norm": 0.7797596868320031,
+      "learning_rate": 9.367568823323039e-06,
+      "loss": 0.4343,
+      "num_input_tokens_seen": 44735576,
+      "step": 805,
+      "train_runtime": 4837.9636,
+      "train_tokens_per_second": 9246.778
+    },
+    {
+      "epoch": 0.32608695652173914,
+      "grad_norm": 0.8306805617663875,
+      "learning_rate": 9.359851123029493e-06,
+      "loss": 0.4256,
+      "num_input_tokens_seen": 45033720,
+      "step": 810,
+      "train_runtime": 4864.7062,
+      "train_tokens_per_second": 9257.233
+    },
+    {
+      "epoch": 0.3280998389694042,
+      "grad_norm": 0.7931899410302765,
+      "learning_rate": 9.352089836649341e-06,
+      "loss": 0.4307,
+      "num_input_tokens_seen": 45290536,
+      "step": 815,
+      "train_runtime": 4890.393,
+      "train_tokens_per_second": 9261.124
+    },
+    {
+      "epoch": 0.33011272141706927,
+      "grad_norm": 0.8744341856838176,
+      "learning_rate": 9.344285041773329e-06,
+      "loss": 0.4425,
+      "num_input_tokens_seen": 45583520,
+      "step": 820,
+      "train_runtime": 4919.8658,
+      "train_tokens_per_second": 9265.196
+    },
+    {
+      "epoch": 0.3321256038647343,
+      "grad_norm": 0.8825076873370873,
+      "learning_rate": 9.336436816427161e-06,
+      "loss": 0.4223,
+      "num_input_tokens_seen": 45852248,
+      "step": 825,
+      "train_runtime": 4946.4443,
+      "train_tokens_per_second": 9269.739
+    },
+    {
+      "epoch": 0.33413848631239934,
+      "grad_norm": 1.017940693566255,
+      "learning_rate": 9.32854523907073e-06,
+      "loss": 0.4349,
+      "num_input_tokens_seen": 46107152,
+      "step": 830,
+      "train_runtime": 4978.8463,
+      "train_tokens_per_second": 9260.61
+    },
+    {
+      "epoch": 0.3361513687600644,
+      "grad_norm": 0.9924252831980989,
+      "learning_rate": 9.320610388597316e-06,
+      "loss": 0.4411,
+      "num_input_tokens_seen": 46378192,
+      "step": 835,
+      "train_runtime": 5004.7666,
+      "train_tokens_per_second": 9266.804
+    },
+    {
+      "epoch": 0.33816425120772947,
+      "grad_norm": 0.8377028821329031,
+      "learning_rate": 9.312632344332812e-06,
+      "loss": 0.4439,
+      "num_input_tokens_seen": 46648456,
+      "step": 840,
+      "train_runtime": 5035.0354,
+      "train_tokens_per_second": 9264.772
+    },
+    {
+      "epoch": 0.34017713365539454,
+      "grad_norm": 0.7941019674205831,
+      "learning_rate": 9.304611186034926e-06,
+      "loss": 0.4151,
+      "num_input_tokens_seen": 46925064,
+      "step": 845,
+      "train_runtime": 5064.5973,
+      "train_tokens_per_second": 9265.31
+    },
+    {
+      "epoch": 0.3421900161030596,
+      "grad_norm": 0.762996698357896,
+      "learning_rate": 9.296546993892383e-06,
+      "loss": 0.418,
+      "num_input_tokens_seen": 47200176,
+      "step": 850,
+      "train_runtime": 5095.2394,
+      "train_tokens_per_second": 9263.584
+    },
+    {
+      "epoch": 0.3442028985507246,
+      "grad_norm": 0.7773040963301553,
+      "learning_rate": 9.28843984852412e-06,
+      "loss": 0.4578,
+      "num_input_tokens_seen": 47480384,
+      "step": 855,
+      "train_runtime": 5129.1372,
+      "train_tokens_per_second": 9256.992
+    },
+    {
+      "epoch": 0.3462157809983897,
+      "grad_norm": 0.8176393942733152,
+      "learning_rate": 9.280289830978491e-06,
+      "loss": 0.4561,
+      "num_input_tokens_seen": 47755752,
+      "step": 860,
+      "train_runtime": 5163.028,
+      "train_tokens_per_second": 9249.563
+    },
+    {
+      "epoch": 0.34822866344605474,
+      "grad_norm": 0.7077728156100337,
+      "learning_rate": 9.272097022732444e-06,
+      "loss": 0.4489,
+      "num_input_tokens_seen": 48073920,
+      "step": 865,
+      "train_runtime": 5194.1612,
+      "train_tokens_per_second": 9255.377
+    },
+    {
+      "epoch": 0.3502415458937198,
+      "grad_norm": 0.8615853331028447,
+      "learning_rate": 9.263861505690712e-06,
+      "loss": 0.4155,
+      "num_input_tokens_seen": 48375272,
+      "step": 870,
+      "train_runtime": 5223.1997,
+      "train_tokens_per_second": 9261.616
+    },
+    {
+      "epoch": 0.3522544283413849,
+      "grad_norm": 0.7139922249684153,
+      "learning_rate": 9.255583362184998e-06,
+      "loss": 0.408,
+      "num_input_tokens_seen": 48658240,
+      "step": 875,
+      "train_runtime": 5247.3367,
+      "train_tokens_per_second": 9272.94
+    },
+    {
+      "epoch": 0.35426731078904994,
+      "grad_norm": 0.7637028710213635,
+      "learning_rate": 9.247262674973147e-06,
+      "loss": 0.4384,
+      "num_input_tokens_seen": 48929232,
+      "step": 880,
+      "train_runtime": 5278.9032,
+      "train_tokens_per_second": 9268.825
+    },
+    {
+      "epoch": 0.35628019323671495,
+      "grad_norm": 0.7841414950127197,
+      "learning_rate": 9.238899527238319e-06,
+      "loss": 0.4335,
+      "num_input_tokens_seen": 49221736,
+      "step": 885,
+      "train_runtime": 5305.2139,
+      "train_tokens_per_second": 9277.993
+    },
+    {
+      "epoch": 0.35829307568438,
+      "grad_norm": 0.7652918502907062,
+      "learning_rate": 9.230494002588154e-06,
+      "loss": 0.4227,
+      "num_input_tokens_seen": 49496208,
+      "step": 890,
+      "train_runtime": 5331.9527,
+      "train_tokens_per_second": 9282.942
+    },
+    {
+      "epoch": 0.3603059581320451,
+      "grad_norm": 0.7735867417804985,
+      "learning_rate": 9.22204618505395e-06,
+      "loss": 0.4222,
+      "num_input_tokens_seen": 49752400,
+      "step": 895,
+      "train_runtime": 5364.6362,
+      "train_tokens_per_second": 9274.142
+    },
+    {
+      "epoch": 0.36231884057971014,
+      "grad_norm": 0.8782098347012791,
+      "learning_rate": 9.213556159089805e-06,
+      "loss": 0.4317,
+      "num_input_tokens_seen": 50034312,
+      "step": 900,
+      "train_runtime": 5394.6337,
+      "train_tokens_per_second": 9274.83
+    },
+    {
+      "epoch": 0.3643317230273752,
+      "grad_norm": 0.7927812987321239,
+      "learning_rate": 9.205024009571784e-06,
+      "loss": 0.4577,
+      "num_input_tokens_seen": 50301344,
+      "step": 905,
+      "train_runtime": 5425.2018,
+      "train_tokens_per_second": 9271.792
+    },
+    {
+      "epoch": 0.3663446054750403,
+      "grad_norm": 0.7953441718203047,
+      "learning_rate": 9.196449821797067e-06,
+      "loss": 0.4242,
+      "num_input_tokens_seen": 50594360,
+      "step": 910,
+      "train_runtime": 5458.7984,
+      "train_tokens_per_second": 9268.406
+    },
+    {
+      "epoch": 0.36835748792270534,
+      "grad_norm": 0.8161470419870963,
+      "learning_rate": 9.187833681483098e-06,
+      "loss": 0.4389,
+      "num_input_tokens_seen": 50865872,
+      "step": 915,
+      "train_runtime": 5489.1951,
+      "train_tokens_per_second": 9266.545
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.7722563446619088,
+      "learning_rate": 9.179175674766727e-06,
+      "loss": 0.41,
+      "num_input_tokens_seen": 51140920,
+      "step": 920,
+      "train_runtime": 5516.1437,
+      "train_tokens_per_second": 9271.136
+    },
+    {
+      "epoch": 0.3723832528180354,
+      "grad_norm": 0.907677374195102,
+      "learning_rate": 9.170475888203348e-06,
+      "loss": 0.4284,
+      "num_input_tokens_seen": 51405568,
+      "step": 925,
+      "train_runtime": 5545.9719,
+      "train_tokens_per_second": 9268.992
+    },
+    {
+      "epoch": 0.3743961352657005,
+      "grad_norm": 0.8141115976341028,
+      "learning_rate": 9.16173440876603e-06,
+      "loss": 0.428,
+      "num_input_tokens_seen": 51699992,
+      "step": 930,
+      "train_runtime": 5571.5192,
+      "train_tokens_per_second": 9279.335
+    },
+    {
+      "epoch": 0.37640901771336555,
+      "grad_norm": 0.8839776819768248,
+      "learning_rate": 9.152951323844661e-06,
+      "loss": 0.4387,
+      "num_input_tokens_seen": 51988568,
+      "step": 935,
+      "train_runtime": 5605.6123,
+      "train_tokens_per_second": 9274.378
+    },
+    {
+      "epoch": 0.3784219001610306,
+      "grad_norm": 0.7717885720181126,
+      "learning_rate": 9.144126721245059e-06,
+      "loss": 0.439,
+      "num_input_tokens_seen": 52260104,
+      "step": 940,
+      "train_runtime": 5633.2138,
+      "train_tokens_per_second": 9277.138
+    },
+    {
+      "epoch": 0.3804347826086957,
+      "grad_norm": 0.815831817830111,
+      "learning_rate": 9.1352606891881e-06,
+      "loss": 0.4157,
+      "num_input_tokens_seen": 52549592,
+      "step": 945,
+      "train_runtime": 5661.7321,
+      "train_tokens_per_second": 9281.54
+    },
+    {
+      "epoch": 0.3824476650563607,
+      "grad_norm": 0.725655932083562,
+      "learning_rate": 9.126353316308838e-06,
+      "loss": 0.4293,
+      "num_input_tokens_seen": 52844200,
+      "step": 950,
+      "train_runtime": 5689.3221,
+      "train_tokens_per_second": 9288.312
+    },
+    {
+      "epoch": 0.38446054750402575,
+      "grad_norm": 0.7763255437116424,
+      "learning_rate": 9.11740469165562e-06,
+      "loss": 0.3919,
+      "num_input_tokens_seen": 53133880,
+      "step": 955,
+      "train_runtime": 5724.8989,
+      "train_tokens_per_second": 9281.191
+    },
+    {
+      "epoch": 0.3864734299516908,
+      "grad_norm": 0.7918081168391918,
+      "learning_rate": 9.108414904689192e-06,
+      "loss": 0.4312,
+      "num_input_tokens_seen": 53411536,
+      "step": 960,
+      "train_runtime": 5759.0785,
+      "train_tokens_per_second": 9274.32
+    },
+    {
+      "epoch": 0.3884863123993559,
+      "grad_norm": 0.7996370434684448,
+      "learning_rate": 9.099384045281801e-06,
+      "loss": 0.4185,
+      "num_input_tokens_seen": 53662792,
+      "step": 965,
+      "train_runtime": 5786.8852,
+      "train_tokens_per_second": 9273.174
+    },
+    {
+      "epoch": 0.39049919484702095,
+      "grad_norm": 0.8525653676568737,
+      "learning_rate": 9.090312203716307e-06,
+      "loss": 0.4295,
+      "num_input_tokens_seen": 53939184,
+      "step": 970,
+      "train_runtime": 5820.6568,
+      "train_tokens_per_second": 9266.855
+    },
+    {
+      "epoch": 0.392512077294686,
+      "grad_norm": 0.8523529936158146,
+      "learning_rate": 9.081199470685273e-06,
+      "loss": 0.402,
+      "num_input_tokens_seen": 54204248,
+      "step": 975,
+      "train_runtime": 5844.8732,
+      "train_tokens_per_second": 9273.811
+    },
+    {
+      "epoch": 0.394524959742351,
+      "grad_norm": 0.7939566350442157,
+      "learning_rate": 9.07204593729006e-06,
+      "loss": 0.413,
+      "num_input_tokens_seen": 54487640,
+      "step": 980,
+      "train_runtime": 5875.535,
+      "train_tokens_per_second": 9273.647
+    },
+    {
+      "epoch": 0.3965378421900161,
+      "grad_norm": 0.8350351016268522,
+      "learning_rate": 9.062851695039915e-06,
+      "loss": 0.4246,
+      "num_input_tokens_seen": 54787960,
+      "step": 985,
+      "train_runtime": 5903.0474,
+      "train_tokens_per_second": 9281.301
+    },
+    {
+      "epoch": 0.39855072463768115,
+      "grad_norm": 0.7583521918717786,
+      "learning_rate": 9.053616835851062e-06,
+      "loss": 0.4407,
+      "num_input_tokens_seen": 55067216,
+      "step": 990,
+      "train_runtime": 5932.0419,
+      "train_tokens_per_second": 9283.012
+    },
+    {
+      "epoch": 0.4005636070853462,
+      "grad_norm": 0.8199267479423458,
+      "learning_rate": 9.044341452045773e-06,
+      "loss": 0.4234,
+      "num_input_tokens_seen": 55366344,
+      "step": 995,
+      "train_runtime": 5965.1435,
+      "train_tokens_per_second": 9281.645
+    },
+    {
+      "epoch": 0.4025764895330113,
+      "grad_norm": 0.8786863350187458,
+      "learning_rate": 9.035025636351453e-06,
+      "loss": 0.4302,
+      "num_input_tokens_seen": 55620376,
+      "step": 1000,
+      "train_runtime": 5992.4648,
+      "train_tokens_per_second": 9281.719
+    },
+    {
+      "epoch": 0.40458937198067635,
+      "grad_norm": 0.8315975144034723,
+      "learning_rate": 9.025669481899713e-06,
+      "loss": 0.4356,
+      "num_input_tokens_seen": 55919896,
+      "step": 1005,
+      "train_runtime": 6025.3717,
+      "train_tokens_per_second": 9280.738
+    },
+    {
+      "epoch": 0.40660225442834136,
+      "grad_norm": 0.8725249931079624,
+      "learning_rate": 9.016273082225434e-06,
+      "loss": 0.455,
+      "num_input_tokens_seen": 56200704,
+      "step": 1010,
+      "train_runtime": 6056.9886,
+      "train_tokens_per_second": 9278.654
+    },
+    {
+      "epoch": 0.4086151368760064,
+      "grad_norm": 0.76095253721177,
+      "learning_rate": 9.006836531265832e-06,
+      "loss": 0.4021,
+      "num_input_tokens_seen": 56491768,
+      "step": 1015,
+      "train_runtime": 6093.7201,
+      "train_tokens_per_second": 9270.489
+    },
+    {
+      "epoch": 0.4106280193236715,
+      "grad_norm": 0.7617050031927668,
+      "learning_rate": 8.997359923359527e-06,
+      "loss": 0.4372,
+      "num_input_tokens_seen": 56776864,
+      "step": 1020,
+      "train_runtime": 6127.1029,
+      "train_tokens_per_second": 9266.511
+    },
+    {
+      "epoch": 0.41264090177133655,
+      "grad_norm": 0.8218965042489578,
+      "learning_rate": 8.987843353245594e-06,
+      "loss": 0.426,
+      "num_input_tokens_seen": 57014104,
+      "step": 1025,
+      "train_runtime": 6159.7331,
+      "train_tokens_per_second": 9255.937
+    },
+    {
+      "epoch": 0.4146537842190016,
+      "grad_norm": 0.7771908303834842,
+      "learning_rate": 8.978286916062609e-06,
+      "loss": 0.4181,
+      "num_input_tokens_seen": 57308432,
+      "step": 1030,
+      "train_runtime": 6189.3701,
+      "train_tokens_per_second": 9259.17
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.8487816109780456,
+      "learning_rate": 8.968690707347714e-06,
+      "loss": 0.4224,
+      "num_input_tokens_seen": 57579872,
+      "step": 1035,
+      "train_runtime": 6216.4033,
+      "train_tokens_per_second": 9262.57
+    },
+    {
+      "epoch": 0.41867954911433175,
+      "grad_norm": 0.9626243037957115,
+      "learning_rate": 8.95905482303565e-06,
+      "loss": 0.4212,
+      "num_input_tokens_seen": 57851320,
+      "step": 1040,
+      "train_runtime": 6244.9798,
+      "train_tokens_per_second": 9263.652
+    },
+    {
+      "epoch": 0.42069243156199676,
+      "grad_norm": 0.8758714391395066,
+      "learning_rate": 8.949379359457795e-06,
+      "loss": 0.4277,
+      "num_input_tokens_seen": 58127192,
+      "step": 1045,
+      "train_runtime": 6274.5791,
+      "train_tokens_per_second": 9263.919
+    },
+    {
+      "epoch": 0.4227053140096618,
+      "grad_norm": 0.7216937763107959,
+      "learning_rate": 8.939664413341214e-06,
+      "loss": 0.43,
+      "num_input_tokens_seen": 58411088,
+      "step": 1050,
+      "train_runtime": 6296.0503,
+      "train_tokens_per_second": 9277.418
+    },
+    {
+      "epoch": 0.4247181964573269,
+      "grad_norm": 0.8099941281191689,
+      "learning_rate": 8.929910081807685e-06,
+      "loss": 0.4499,
+      "num_input_tokens_seen": 58690328,
+      "step": 1055,
+      "train_runtime": 6329.7588,
+      "train_tokens_per_second": 9272.127
+    },
+    {
+      "epoch": 0.42673107890499196,
+      "grad_norm": 0.7739158247575855,
+      "learning_rate": 8.920116462372723e-06,
+      "loss": 0.4042,
+      "num_input_tokens_seen": 58981968,
+      "step": 1060,
+      "train_runtime": 6352.9975,
+      "train_tokens_per_second": 9284.116
+    },
+    {
+      "epoch": 0.428743961352657,
+      "grad_norm": 0.7307253582039621,
+      "learning_rate": 8.910283652944614e-06,
+      "loss": 0.4248,
+      "num_input_tokens_seen": 59257176,
+      "step": 1065,
+      "train_runtime": 6385.961,
+      "train_tokens_per_second": 9279.289
+    },
+    {
+      "epoch": 0.4307568438003221,
+      "grad_norm": 0.7974935934335282,
+      "learning_rate": 8.900411751823427e-06,
+      "loss": 0.4202,
+      "num_input_tokens_seen": 59551288,
+      "step": 1070,
+      "train_runtime": 6414.4543,
+      "train_tokens_per_second": 9283.921
+    },
+    {
+      "epoch": 0.4327697262479871,
+      "grad_norm": 0.8125325024109311,
+      "learning_rate": 8.890500857700049e-06,
+      "loss": 0.4336,
+      "num_input_tokens_seen": 59802968,
+      "step": 1075,
+      "train_runtime": 6444.0157,
+      "train_tokens_per_second": 9280.388
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 0.8804885219731723,
+      "learning_rate": 8.88055106965517e-06,
+      "loss": 0.4087,
+      "num_input_tokens_seen": 60081152,
+      "step": 1080,
+      "train_runtime": 6478.292,
+      "train_tokens_per_second": 9274.227
+    },
+    {
+      "epoch": 0.4367954911433172,
+      "grad_norm": 0.8261992474901164,
+      "learning_rate": 8.870562487158317e-06,
+      "loss": 0.4111,
+      "num_input_tokens_seen": 60347144,
+      "step": 1085,
+      "train_runtime": 6506.2419,
+      "train_tokens_per_second": 9275.269
+    },
+    {
+      "epoch": 0.4388083735909823,
+      "grad_norm": 0.7975129914891287,
+      "learning_rate": 8.860535210066851e-06,
+      "loss": 0.4005,
+      "num_input_tokens_seen": 60646176,
+      "step": 1090,
+      "train_runtime": 6535.4243,
+      "train_tokens_per_second": 9279.608
+    },
+    {
+      "epoch": 0.44082125603864736,
+      "grad_norm": 0.7145945918668941,
+      "learning_rate": 8.850469338624968e-06,
+      "loss": 0.42,
+      "num_input_tokens_seen": 60919904,
+      "step": 1095,
+      "train_runtime": 6567.0681,
+      "train_tokens_per_second": 9276.576
+    },
+    {
+      "epoch": 0.4428341384863124,
+      "grad_norm": 0.7983659127960336,
+      "learning_rate": 8.840364973462697e-06,
+      "loss": 0.4269,
+      "num_input_tokens_seen": 61203944,
+      "step": 1100,
+      "train_runtime": 6593.8991,
+      "train_tokens_per_second": 9281.905
+    },
+    {
+      "epoch": 0.44484702093397743,
+      "grad_norm": 0.7697037595379774,
+      "learning_rate": 8.83022221559489e-06,
+      "loss": 0.42,
+      "num_input_tokens_seen": 61500688,
+      "step": 1105,
+      "train_runtime": 6621.0075,
+      "train_tokens_per_second": 9288.721
+    },
+    {
+      "epoch": 0.4468599033816425,
+      "grad_norm": 0.8870610863205187,
+      "learning_rate": 8.820041166420228e-06,
+      "loss": 0.4155,
+      "num_input_tokens_seen": 61761752,
+      "step": 1110,
+      "train_runtime": 6645.5292,
+      "train_tokens_per_second": 9293.73
+    },
+    {
+      "epoch": 0.44887278582930756,
+      "grad_norm": 0.7672116186872291,
+      "learning_rate": 8.809821927720182e-06,
+      "loss": 0.4029,
+      "num_input_tokens_seen": 62034568,
+      "step": 1115,
+      "train_runtime": 6668.3025,
+      "train_tokens_per_second": 9302.902
+    },
+    {
+      "epoch": 0.45088566827697263,
+      "grad_norm": 0.8024175585430984,
+      "learning_rate": 8.799564601658022e-06,
+      "loss": 0.4241,
+      "num_input_tokens_seen": 62312536,
+      "step": 1120,
+      "train_runtime": 6697.9535,
+      "train_tokens_per_second": 9303.22
+    },
+    {
+      "epoch": 0.4528985507246377,
+      "grad_norm": 0.7825203639589086,
+      "learning_rate": 8.789269290777774e-06,
+      "loss": 0.4015,
+      "num_input_tokens_seen": 62604504,
+      "step": 1125,
+      "train_runtime": 6728.772,
+      "train_tokens_per_second": 9304.001
+    },
+    {
+      "epoch": 0.45491143317230276,
+      "grad_norm": 0.8512537082380527,
+      "learning_rate": 8.778936098003208e-06,
+      "loss": 0.4176,
+      "num_input_tokens_seen": 62870376,
+      "step": 1130,
+      "train_runtime": 6763.2308,
+      "train_tokens_per_second": 9295.909
+    },
+    {
+      "epoch": 0.45692431561996777,
+      "grad_norm": 0.7398038695028081,
+      "learning_rate": 8.768565126636806e-06,
+      "loss": 0.4079,
+      "num_input_tokens_seen": 63167576,
+      "step": 1135,
+      "train_runtime": 6789.6214,
+      "train_tokens_per_second": 9303.549
+    },
+    {
+      "epoch": 0.45893719806763283,
+      "grad_norm": 0.7678412894221827,
+      "learning_rate": 8.758156480358725e-06,
+      "loss": 0.4161,
+      "num_input_tokens_seen": 63421152,
+      "step": 1140,
+      "train_runtime": 6813.3373,
+      "train_tokens_per_second": 9308.383
+    },
+    {
+      "epoch": 0.4609500805152979,
+      "grad_norm": 0.8255432717200041,
+      "learning_rate": 8.747710263225768e-06,
+      "loss": 0.3905,
+      "num_input_tokens_seen": 63688680,
+      "step": 1145,
+      "train_runtime": 6840.9033,
+      "train_tokens_per_second": 9309.981
+    },
+    {
+      "epoch": 0.46296296296296297,
+      "grad_norm": 0.8192651871722412,
+      "learning_rate": 8.737226579670332e-06,
+      "loss": 0.4339,
+      "num_input_tokens_seen": 63959584,
+      "step": 1150,
+      "train_runtime": 6869.6767,
+      "train_tokens_per_second": 9310.421
+    },
+    {
+      "epoch": 0.46497584541062803,
+      "grad_norm": 0.8628664715751423,
+      "learning_rate": 8.726705534499382e-06,
+      "loss": 0.4133,
+      "num_input_tokens_seen": 64213416,
+      "step": 1155,
+      "train_runtime": 6895.9926,
+      "train_tokens_per_second": 9311.7
+    },
+    {
+      "epoch": 0.4669887278582931,
+      "grad_norm": 0.7543690529507546,
+      "learning_rate": 8.716147232893382e-06,
+      "loss": 0.4012,
+      "num_input_tokens_seen": 64497912,
+      "step": 1160,
+      "train_runtime": 6927.506,
+      "train_tokens_per_second": 9310.409
+    },
+    {
+      "epoch": 0.4690016103059581,
+      "grad_norm": 0.7059034953253939,
+      "learning_rate": 8.705551780405264e-06,
+      "loss": 0.4135,
+      "num_input_tokens_seen": 64780848,
+      "step": 1165,
+      "train_runtime": 6956.8259,
+      "train_tokens_per_second": 9311.84
+    },
+    {
+      "epoch": 0.47101449275362317,
+      "grad_norm": 0.876605346843507,
+      "learning_rate": 8.694919282959353e-06,
+      "loss": 0.421,
+      "num_input_tokens_seen": 65054040,
+      "step": 1170,
+      "train_runtime": 6982.9796,
+      "train_tokens_per_second": 9316.086
+    },
+    {
+      "epoch": 0.47302737520128824,
+      "grad_norm": 0.6761038282607027,
+      "learning_rate": 8.684249846850328e-06,
+      "loss": 0.4014,
+      "num_input_tokens_seen": 65345304,
+      "step": 1175,
+      "train_runtime": 7012.0805,
+      "train_tokens_per_second": 9318.961
+    },
+    {
+      "epoch": 0.4750402576489533,
+      "grad_norm": 0.7649467230911865,
+      "learning_rate": 8.673543578742141e-06,
+      "loss": 0.4398,
+      "num_input_tokens_seen": 65625352,
+      "step": 1180,
+      "train_runtime": 7040.0449,
+      "train_tokens_per_second": 9321.723
+    },
+    {
+      "epoch": 0.47705314009661837,
+      "grad_norm": 0.7195835840995003,
+      "learning_rate": 8.662800585666963e-06,
+      "loss": 0.417,
+      "num_input_tokens_seen": 65912288,
+      "step": 1185,
+      "train_runtime": 7076.146,
+      "train_tokens_per_second": 9314.716
+    },
+    {
+      "epoch": 0.47906602254428343,
+      "grad_norm": 0.7775110391022959,
+      "learning_rate": 8.652020975024114e-06,
+      "loss": 0.4081,
+      "num_input_tokens_seen": 66208080,
+      "step": 1190,
+      "train_runtime": 7111.0847,
+      "train_tokens_per_second": 9310.546
+    },
+    {
+      "epoch": 0.4810789049919485,
+      "grad_norm": 0.7360818939499673,
+      "learning_rate": 8.641204854578975e-06,
+      "loss": 0.3991,
+      "num_input_tokens_seen": 66494368,
+      "step": 1195,
+      "train_runtime": 7141.464,
+      "train_tokens_per_second": 9311.028
+    },
+    {
+      "epoch": 0.4830917874396135,
+      "grad_norm": 0.7774115398225843,
+      "learning_rate": 8.63035233246193e-06,
+      "loss": 0.4278,
+      "num_input_tokens_seen": 66760448,
+      "step": 1200,
+      "train_runtime": 7173.8755,
+      "train_tokens_per_second": 9306.051
+    },
+    {
+      "epoch": 0.4851046698872786,
+      "grad_norm": 0.9950808464160816,
+      "learning_rate": 8.61946351716727e-06,
+      "loss": 0.4183,
+      "num_input_tokens_seen": 66997576,
+      "step": 1205,
+      "train_runtime": 7201.2345,
+      "train_tokens_per_second": 9303.624
+    },
+    {
+      "epoch": 0.48711755233494364,
+      "grad_norm": 0.8292785156406555,
+      "learning_rate": 8.608538517552121e-06,
+      "loss": 0.3992,
+      "num_input_tokens_seen": 67271736,
+      "step": 1210,
+      "train_runtime": 7234.6368,
+      "train_tokens_per_second": 9298.564
+    },
+    {
+      "epoch": 0.4891304347826087,
+      "grad_norm": 0.7186650866409257,
+      "learning_rate": 8.597577442835343e-06,
+      "loss": 0.4436,
+      "num_input_tokens_seen": 67538936,
+      "step": 1215,
+      "train_runtime": 7261.1964,
+      "train_tokens_per_second": 9301.351
+    },
+    {
+      "epoch": 0.49114331723027377,
+      "grad_norm": 0.8950002960474922,
+      "learning_rate": 8.586580402596448e-06,
+      "loss": 0.3819,
+      "num_input_tokens_seen": 67806248,
+      "step": 1220,
+      "train_runtime": 7287.5961,
+      "train_tokens_per_second": 9304.337
+    },
+    {
+      "epoch": 0.49315619967793883,
+      "grad_norm": 0.8988044081615849,
+      "learning_rate": 8.575547506774498e-06,
+      "loss": 0.4269,
+      "num_input_tokens_seen": 68063000,
+      "step": 1225,
+      "train_runtime": 7316.4106,
+      "train_tokens_per_second": 9302.786
+    },
+    {
+      "epoch": 0.49516908212560384,
+      "grad_norm": 1.4499574487371216,
+      "learning_rate": 8.56447886566701e-06,
+      "loss": 0.4279,
+      "num_input_tokens_seen": 68330672,
+      "step": 1230,
+      "train_runtime": 7341.4495,
+      "train_tokens_per_second": 9307.518
+    },
+    {
+      "epoch": 0.4971819645732689,
+      "grad_norm": 0.79706105900836,
+      "learning_rate": 8.553374589928851e-06,
+      "loss": 0.4165,
+      "num_input_tokens_seen": 68610320,
+      "step": 1235,
+      "train_runtime": 7376.5667,
+      "train_tokens_per_second": 9301.118
+    },
+    {
+      "epoch": 0.499194847020934,
+      "grad_norm": 0.8033716429077462,
+      "learning_rate": 8.542234790571136e-06,
+      "loss": 0.4048,
+      "num_input_tokens_seen": 68887656,
+      "step": 1240,
+      "train_runtime": 7400.4789,
+      "train_tokens_per_second": 9308.54
+    },
+    {
+      "epoch": 0.501207729468599,
+      "grad_norm": 0.7698459872558903,
+      "learning_rate": 8.531059578960111e-06,
+      "loss": 0.4076,
+      "num_input_tokens_seen": 69170832,
+      "step": 1245,
+      "train_runtime": 7428.1267,
+      "train_tokens_per_second": 9312.016
+    },
+    {
+      "epoch": 0.5032206119162641,
+      "grad_norm": 0.7561813197187125,
+      "learning_rate": 8.519849066816046e-06,
+      "loss": 0.4199,
+      "num_input_tokens_seen": 69454320,
+      "step": 1250,
+      "train_runtime": 7448.8515,
+      "train_tokens_per_second": 9324.165
+    },
+    {
+      "epoch": 0.5052334943639292,
+      "grad_norm": 0.8724135923547159,
+      "learning_rate": 8.508603366212113e-06,
+      "loss": 0.4133,
+      "num_input_tokens_seen": 69706760,
+      "step": 1255,
+      "train_runtime": 7475.0445,
+      "train_tokens_per_second": 9325.264
+    },
+    {
+      "epoch": 0.5072463768115942,
+      "grad_norm": 0.7021208161084885,
+      "learning_rate": 8.497322589573272e-06,
+      "loss": 0.3747,
+      "num_input_tokens_seen": 69971696,
+      "step": 1260,
+      "train_runtime": 7507.0647,
+      "train_tokens_per_second": 9320.78
+    },
+    {
+      "epoch": 0.5092592592592593,
+      "grad_norm": 0.9245447692475792,
+      "learning_rate": 8.486006849675142e-06,
+      "loss": 0.4276,
+      "num_input_tokens_seen": 70233552,
+      "step": 1265,
+      "train_runtime": 7533.2107,
+      "train_tokens_per_second": 9323.19
+    },
+    {
+      "epoch": 0.5112721417069244,
+      "grad_norm": 0.8287865456418892,
+      "learning_rate": 8.474656259642874e-06,
+      "loss": 0.4103,
+      "num_input_tokens_seen": 70484672,
+      "step": 1270,
+      "train_runtime": 7564.9062,
+      "train_tokens_per_second": 9317.323
+    },
+    {
+      "epoch": 0.5132850241545893,
+      "grad_norm": 0.6454547720839349,
+      "learning_rate": 8.46327093295002e-06,
+      "loss": 0.4062,
+      "num_input_tokens_seen": 70777128,
+      "step": 1275,
+      "train_runtime": 7592.5195,
+      "train_tokens_per_second": 9321.955
+    },
+    {
+      "epoch": 0.5152979066022544,
+      "grad_norm": 0.8386391796322755,
+      "learning_rate": 8.451850983417404e-06,
+      "loss": 0.4385,
+      "num_input_tokens_seen": 71046616,
+      "step": 1280,
+      "train_runtime": 7618.5363,
+      "train_tokens_per_second": 9325.494
+    },
+    {
+      "epoch": 0.5173107890499195,
+      "grad_norm": 0.8042716303191277,
+      "learning_rate": 8.440396525211976e-06,
+      "loss": 0.401,
+      "num_input_tokens_seen": 71338800,
+      "step": 1285,
+      "train_runtime": 7641.3611,
+      "train_tokens_per_second": 9335.876
+    },
+    {
+      "epoch": 0.5193236714975845,
+      "grad_norm": 0.8564512745233012,
+      "learning_rate": 8.428907672845674e-06,
+      "loss": 0.4165,
+      "num_input_tokens_seen": 71582640,
+      "step": 1290,
+      "train_runtime": 7666.205,
+      "train_tokens_per_second": 9337.428
+    },
+    {
+      "epoch": 0.5213365539452496,
+      "grad_norm": 0.7948783252721567,
+      "learning_rate": 8.417384541174283e-06,
+      "loss": 0.4286,
+      "num_input_tokens_seen": 71872720,
+      "step": 1295,
+      "train_runtime": 7688.9796,
+      "train_tokens_per_second": 9347.498
+    },
+    {
+      "epoch": 0.5233494363929146,
+      "grad_norm": 0.811121493186393,
+      "learning_rate": 8.405827245396279e-06,
+      "loss": 0.4322,
+      "num_input_tokens_seen": 72174080,
+      "step": 1300,
+      "train_runtime": 7724.0508,
+      "train_tokens_per_second": 9344.071
+    },
+    {
+      "epoch": 0.5253623188405797,
+      "grad_norm": 0.762502090476539,
+      "learning_rate": 8.394235901051687e-06,
+      "loss": 0.4092,
+      "num_input_tokens_seen": 72421432,
+      "step": 1305,
+      "train_runtime": 7753.6075,
+      "train_tokens_per_second": 9340.353
+    },
+    {
+      "epoch": 0.5273752012882448,
+      "grad_norm": 0.7582519413781591,
+      "learning_rate": 8.382610624020913e-06,
+      "loss": 0.3822,
+      "num_input_tokens_seen": 72696312,
+      "step": 1310,
+      "train_runtime": 7785.3957,
+      "train_tokens_per_second": 9337.523
+    },
+    {
+      "epoch": 0.5293880837359098,
+      "grad_norm": 0.7990223884320373,
+      "learning_rate": 8.370951530523602e-06,
+      "loss": 0.4106,
+      "num_input_tokens_seen": 72970264,
+      "step": 1315,
+      "train_runtime": 7812.0981,
+      "train_tokens_per_second": 9340.674
+    },
+    {
+      "epoch": 0.5314009661835749,
+      "grad_norm": 0.7791346221612352,
+      "learning_rate": 8.359258737117463e-06,
+      "loss": 0.4172,
+      "num_input_tokens_seen": 73232672,
+      "step": 1320,
+      "train_runtime": 7843.852,
+      "train_tokens_per_second": 9336.315
+    },
+    {
+      "epoch": 0.53341384863124,
+      "grad_norm": 0.7899660696019841,
+      "learning_rate": 8.347532360697105e-06,
+      "loss": 0.4056,
+      "num_input_tokens_seen": 73518464,
+      "step": 1325,
+      "train_runtime": 7876.9916,
+      "train_tokens_per_second": 9333.318
+    },
+    {
+      "epoch": 0.535426731078905,
+      "grad_norm": 0.795792206607847,
+      "learning_rate": 8.335772518492878e-06,
+      "loss": 0.3938,
+      "num_input_tokens_seen": 73785808,
+      "step": 1330,
+      "train_runtime": 7910.2029,
+      "train_tokens_per_second": 9327.929
+    },
+    {
+      "epoch": 0.5374396135265701,
+      "grad_norm": 0.7775548882348562,
+      "learning_rate": 8.323979328069689e-06,
+      "loss": 0.4377,
+      "num_input_tokens_seen": 74059720,
+      "step": 1335,
+      "train_runtime": 7940.8486,
+      "train_tokens_per_second": 9326.424
+    },
+    {
+      "epoch": 0.5394524959742351,
+      "grad_norm": 0.7161132636674606,
+      "learning_rate": 8.312152907325837e-06,
+      "loss": 0.4267,
+      "num_input_tokens_seen": 74346536,
+      "step": 1340,
+      "train_runtime": 7975.8771,
+      "train_tokens_per_second": 9321.424
+    },
+    {
+      "epoch": 0.5414653784219001,
+      "grad_norm": 0.8451827006688751,
+      "learning_rate": 8.300293374491821e-06,
+      "loss": 0.4234,
+      "num_input_tokens_seen": 74626960,
+      "step": 1345,
+      "train_runtime": 8005.6906,
+      "train_tokens_per_second": 9321.739
+    },
+    {
+      "epoch": 0.5434782608695652,
+      "grad_norm": 0.7823366094755075,
+      "learning_rate": 8.288400848129179e-06,
+      "loss": 0.4116,
+      "num_input_tokens_seen": 74928504,
+      "step": 1350,
+      "train_runtime": 8038.4807,
+      "train_tokens_per_second": 9321.227
+    },
+    {
+      "epoch": 0.5454911433172303,
+      "grad_norm": 0.7437423974396629,
+      "learning_rate": 8.276475447129278e-06,
+      "loss": 0.3841,
+      "num_input_tokens_seen": 75228432,
+      "step": 1355,
+      "train_runtime": 8074.609,
+      "train_tokens_per_second": 9316.666
+    },
+    {
+      "epoch": 0.5475040257648953,
+      "grad_norm": 0.7556745121587225,
+      "learning_rate": 8.264517290712146e-06,
+      "loss": 0.4082,
+      "num_input_tokens_seen": 75509304,
+      "step": 1360,
+      "train_runtime": 8098.6441,
+      "train_tokens_per_second": 9323.697
+    },
+    {
+      "epoch": 0.5495169082125604,
+      "grad_norm": 0.706878077272447,
+      "learning_rate": 8.252526498425271e-06,
+      "loss": 0.4147,
+      "num_input_tokens_seen": 75798112,
+      "step": 1365,
+      "train_runtime": 8125.2554,
+      "train_tokens_per_second": 9328.705
+    },
+    {
+      "epoch": 0.5515297906602254,
+      "grad_norm": 0.7517771167022544,
+      "learning_rate": 8.2405031901424e-06,
+      "loss": 0.4163,
+      "num_input_tokens_seen": 76073616,
+      "step": 1370,
+      "train_runtime": 8153.0426,
+      "train_tokens_per_second": 9330.703
+    },
+    {
+      "epoch": 0.5535426731078905,
+      "grad_norm": 0.6918834269343338,
+      "learning_rate": 8.228447486062355e-06,
+      "loss": 0.4307,
+      "num_input_tokens_seen": 76342592,
+      "step": 1375,
+      "train_runtime": 8182.8543,
+      "train_tokens_per_second": 9329.58
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.7590110801395502,
+      "learning_rate": 8.216359506707819e-06,
+      "loss": 0.3974,
+      "num_input_tokens_seen": 76621592,
+      "step": 1380,
+      "train_runtime": 8214.1213,
+      "train_tokens_per_second": 9328.033
+    },
+    {
+      "epoch": 0.5575684380032206,
+      "grad_norm": 0.7822209492399311,
+      "learning_rate": 8.20423937292414e-06,
+      "loss": 0.4112,
+      "num_input_tokens_seen": 76907904,
+      "step": 1385,
+      "train_runtime": 8242.0754,
+      "train_tokens_per_second": 9331.133
+    },
+    {
+      "epoch": 0.5595813204508857,
+      "grad_norm": 0.7828236455605123,
+      "learning_rate": 8.192087205878112e-06,
+      "loss": 0.4179,
+      "num_input_tokens_seen": 77214200,
+      "step": 1390,
+      "train_runtime": 8277.4336,
+      "train_tokens_per_second": 9328.278
+    },
+    {
+      "epoch": 0.5615942028985508,
+      "grad_norm": 0.715514810639999,
+      "learning_rate": 8.17990312705678e-06,
+      "loss": 0.4108,
+      "num_input_tokens_seen": 77465448,
+      "step": 1395,
+      "train_runtime": 8311.6697,
+      "train_tokens_per_second": 9320.083
+    },
+    {
+      "epoch": 0.5636070853462157,
+      "grad_norm": 0.8056833261257931,
+      "learning_rate": 8.167687258266203e-06,
+      "loss": 0.4103,
+      "num_input_tokens_seen": 77741864,
+      "step": 1400,
+      "train_runtime": 8339.6511,
+      "train_tokens_per_second": 9321.956
+    },
+    {
+      "epoch": 0.5656199677938808,
+      "grad_norm": 0.7085581003165861,
+      "learning_rate": 8.155439721630265e-06,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 78039024,
+      "step": 1405,
+      "train_runtime": 8369.7823,
+      "train_tokens_per_second": 9323.901
+    },
+    {
+      "epoch": 0.5676328502415459,
+      "grad_norm": 0.7575190682347418,
+      "learning_rate": 8.143160639589424e-06,
+      "loss": 0.3872,
+      "num_input_tokens_seen": 78323352,
+      "step": 1410,
+      "train_runtime": 8402.8507,
+      "train_tokens_per_second": 9321.045
+    },
+    {
+      "epoch": 0.5696457326892109,
+      "grad_norm": 0.739616713739804,
+      "learning_rate": 8.130850134899508e-06,
+      "loss": 0.3951,
+      "num_input_tokens_seen": 78605544,
+      "step": 1415,
+      "train_runtime": 8429.5648,
+      "train_tokens_per_second": 9324.982
+    },
+    {
+      "epoch": 0.571658615136876,
+      "grad_norm": 0.8054504638632436,
+      "learning_rate": 8.118508330630484e-06,
+      "loss": 0.4369,
+      "num_input_tokens_seen": 78884656,
+      "step": 1420,
+      "train_runtime": 8461.3006,
+      "train_tokens_per_second": 9322.994
+    },
+    {
+      "epoch": 0.5736714975845411,
+      "grad_norm": 0.773554342824745,
+      "learning_rate": 8.106135350165219e-06,
+      "loss": 0.4105,
+      "num_input_tokens_seen": 79156248,
+      "step": 1425,
+      "train_runtime": 8494.0206,
+      "train_tokens_per_second": 9319.055
+    },
+    {
+      "epoch": 0.5756843800322061,
+      "grad_norm": 0.8126957140874953,
+      "learning_rate": 8.093731317198262e-06,
+      "loss": 0.4199,
+      "num_input_tokens_seen": 79440224,
+      "step": 1430,
+      "train_runtime": 8519.8467,
+      "train_tokens_per_second": 9324.138
+    },
+    {
+      "epoch": 0.5776972624798712,
+      "grad_norm": 0.7581883025220503,
+      "learning_rate": 8.08129635573459e-06,
+      "loss": 0.4022,
+      "num_input_tokens_seen": 79715160,
+      "step": 1435,
+      "train_runtime": 8556.3142,
+      "train_tokens_per_second": 9316.53
+    },
+    {
+      "epoch": 0.5797101449275363,
+      "grad_norm": 0.7497077845747706,
+      "learning_rate": 8.068830590088386e-06,
+      "loss": 0.4251,
+      "num_input_tokens_seen": 79969208,
+      "step": 1440,
+      "train_runtime": 8592.1119,
+      "train_tokens_per_second": 9307.282
+    },
+    {
+      "epoch": 0.5817230273752013,
+      "grad_norm": 0.782497742400787,
+      "learning_rate": 8.056334144881776e-06,
+      "loss": 0.4166,
+      "num_input_tokens_seen": 80274104,
+      "step": 1445,
+      "train_runtime": 8629.2464,
+      "train_tokens_per_second": 9302.563
+    },
+    {
+      "epoch": 0.5837359098228664,
+      "grad_norm": 0.7040320452210902,
+      "learning_rate": 8.043807145043604e-06,
+      "loss": 0.4187,
+      "num_input_tokens_seen": 80575952,
+      "step": 1450,
+      "train_runtime": 8660.1054,
+      "train_tokens_per_second": 9304.269
+    },
+    {
+      "epoch": 0.5857487922705314,
+      "grad_norm": 0.7772538084049987,
+      "learning_rate": 8.031249715808166e-06,
+      "loss": 0.4365,
+      "num_input_tokens_seen": 80875656,
+      "step": 1455,
+      "train_runtime": 8692.7988,
+      "train_tokens_per_second": 9303.753
+    },
+    {
+      "epoch": 0.5877616747181964,
+      "grad_norm": 0.7506376911421676,
+      "learning_rate": 8.018661982713971e-06,
+      "loss": 0.416,
+      "num_input_tokens_seen": 81141016,
+      "step": 1460,
+      "train_runtime": 8725.0541,
+      "train_tokens_per_second": 9299.772
+    },
+    {
+      "epoch": 0.5897745571658615,
+      "grad_norm": 0.7117021997539261,
+      "learning_rate": 8.006044071602476e-06,
+      "loss": 0.4446,
+      "num_input_tokens_seen": 81420136,
+      "step": 1465,
+      "train_runtime": 8755.3652,
+      "train_tokens_per_second": 9299.456
+    },
+    {
+      "epoch": 0.5917874396135265,
+      "grad_norm": 0.7187049533776776,
+      "learning_rate": 7.993396108616835e-06,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 81671240,
+      "step": 1470,
+      "train_runtime": 8783.2595,
+      "train_tokens_per_second": 9298.512
+    },
+    {
+      "epoch": 0.5938003220611916,
+      "grad_norm": 0.7853667948151974,
+      "learning_rate": 7.980718220200633e-06,
+      "loss": 0.4509,
+      "num_input_tokens_seen": 81924936,
+      "step": 1475,
+      "train_runtime": 8816.2974,
+      "train_tokens_per_second": 9292.442
+    },
+    {
+      "epoch": 0.5958132045088567,
+      "grad_norm": 0.7450821167641029,
+      "learning_rate": 7.968010533096626e-06,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 82205400,
+      "step": 1480,
+      "train_runtime": 8850.1629,
+      "train_tokens_per_second": 9288.575
+    },
+    {
+      "epoch": 0.5978260869565217,
+      "grad_norm": 0.8697385982049791,
+      "learning_rate": 7.955273174345469e-06,
+      "loss": 0.4269,
+      "num_input_tokens_seen": 82502664,
+      "step": 1485,
+      "train_runtime": 8885.7248,
+      "train_tokens_per_second": 9284.855
+    },
+    {
+      "epoch": 0.5998389694041868,
+      "grad_norm": 0.8163147034324832,
+      "learning_rate": 7.942506271284453e-06,
+      "loss": 0.423,
+      "num_input_tokens_seen": 82757296,
+      "step": 1490,
+      "train_runtime": 8914.7841,
+      "train_tokens_per_second": 9283.152
+    },
+    {
+      "epoch": 0.6018518518518519,
+      "grad_norm": 0.8193991292710503,
+      "learning_rate": 7.929709951546224e-06,
+      "loss": 0.4119,
+      "num_input_tokens_seen": 83024168,
+      "step": 1495,
+      "train_runtime": 8946.8388,
+      "train_tokens_per_second": 9279.721
+    },
+    {
+      "epoch": 0.6038647342995169,
+      "grad_norm": 0.8354348440646796,
+      "learning_rate": 7.916884343057514e-06,
+      "loss": 0.3975,
+      "num_input_tokens_seen": 83289440,
+      "step": 1500,
+      "train_runtime": 8977.7127,
+      "train_tokens_per_second": 9277.356
+    },
+    {
+      "epoch": 0.605877616747182,
+      "grad_norm": 0.7607012692990271,
+      "learning_rate": 7.904029574037856e-06,
+      "loss": 0.3957,
+      "num_input_tokens_seen": 83578104,
+      "step": 1505,
+      "train_runtime": 9005.3436,
+      "train_tokens_per_second": 9280.946
+    },
+    {
+      "epoch": 0.6078904991948471,
+      "grad_norm": 0.8764372525390703,
+      "learning_rate": 7.891145772998308e-06,
+      "loss": 0.4119,
+      "num_input_tokens_seen": 83852400,
+      "step": 1510,
+      "train_runtime": 9041.1196,
+      "train_tokens_per_second": 9274.559
+    },
+    {
+      "epoch": 0.6099033816425121,
+      "grad_norm": 0.7952168568176678,
+      "learning_rate": 7.878233068740167e-06,
+      "loss": 0.4114,
+      "num_input_tokens_seen": 84142088,
+      "step": 1515,
+      "train_runtime": 9068.2962,
+      "train_tokens_per_second": 9278.71
+    },
+    {
+      "epoch": 0.6119162640901772,
+      "grad_norm": 0.792949992765805,
+      "learning_rate": 7.865291590353673e-06,
+      "loss": 0.4299,
+      "num_input_tokens_seen": 84387064,
+      "step": 1520,
+      "train_runtime": 9097.1163,
+      "train_tokens_per_second": 9276.243
+    },
+    {
+      "epoch": 0.6139291465378421,
+      "grad_norm": 0.7358203819262567,
+      "learning_rate": 7.85232146721673e-06,
+      "loss": 0.414,
+      "num_input_tokens_seen": 84681760,
+      "step": 1525,
+      "train_runtime": 9119.6292,
+      "train_tokens_per_second": 9285.658
+    },
+    {
+      "epoch": 0.6159420289855072,
+      "grad_norm": 0.8199196632838759,
+      "learning_rate": 7.83932282899361e-06,
+      "loss": 0.3939,
+      "num_input_tokens_seen": 84970560,
+      "step": 1530,
+      "train_runtime": 9147.9479,
+      "train_tokens_per_second": 9288.483
+    },
+    {
+      "epoch": 0.6179549114331723,
+      "grad_norm": 0.7812276695878765,
+      "learning_rate": 7.826295805633647e-06,
+      "loss": 0.4165,
+      "num_input_tokens_seen": 85221368,
+      "step": 1535,
+      "train_runtime": 9178.5245,
+      "train_tokens_per_second": 9284.866
+    },
+    {
+      "epoch": 0.6199677938808373,
+      "grad_norm": 0.7783483059564613,
+      "learning_rate": 7.813240527369958e-06,
+      "loss": 0.4124,
+      "num_input_tokens_seen": 85527928,
+      "step": 1540,
+      "train_runtime": 9208.1988,
+      "train_tokens_per_second": 9288.236
+    },
+    {
+      "epoch": 0.6219806763285024,
+      "grad_norm": 0.7522800216311623,
+      "learning_rate": 7.800157124718115e-06,
+      "loss": 0.4255,
+      "num_input_tokens_seen": 85796712,
+      "step": 1545,
+      "train_runtime": 9238.9678,
+      "train_tokens_per_second": 9286.396
+    },
+    {
+      "epoch": 0.6239935587761675,
+      "grad_norm": 0.7822265349047505,
+      "learning_rate": 7.787045728474862e-06,
+      "loss": 0.4025,
+      "num_input_tokens_seen": 86058784,
+      "step": 1550,
+      "train_runtime": 9263.2742,
+      "train_tokens_per_second": 9290.32
+    },
+    {
+      "epoch": 0.6260064412238325,
+      "grad_norm": 0.8661952320332765,
+      "learning_rate": 7.773906469716795e-06,
+      "loss": 0.4256,
+      "num_input_tokens_seen": 86327264,
+      "step": 1555,
+      "train_runtime": 9297.4462,
+      "train_tokens_per_second": 9285.051
+    },
+    {
+      "epoch": 0.6280193236714976,
+      "grad_norm": 0.7345883465291086,
+      "learning_rate": 7.76073947979906e-06,
+      "loss": 0.3984,
+      "num_input_tokens_seen": 86608720,
+      "step": 1560,
+      "train_runtime": 9321.0558,
+      "train_tokens_per_second": 9291.729
+    },
+    {
+      "epoch": 0.6300322061191627,
+      "grad_norm": 0.8044894835591974,
+      "learning_rate": 7.747544890354031e-06,
+      "loss": 0.3923,
+      "num_input_tokens_seen": 86861072,
+      "step": 1565,
+      "train_runtime": 9347.042,
+      "train_tokens_per_second": 9292.894
+    },
+    {
+      "epoch": 0.6320450885668277,
+      "grad_norm": 0.8194405153201567,
+      "learning_rate": 7.734322833290002e-06,
+      "loss": 0.4188,
+      "num_input_tokens_seen": 87136000,
+      "step": 1570,
+      "train_runtime": 9376.5258,
+      "train_tokens_per_second": 9292.994
+    },
+    {
+      "epoch": 0.6340579710144928,
+      "grad_norm": 0.8190379059685756,
+      "learning_rate": 7.721073440789863e-06,
+      "loss": 0.4144,
+      "num_input_tokens_seen": 87420648,
+      "step": 1575,
+      "train_runtime": 9401.3475,
+      "train_tokens_per_second": 9298.736
+    },
+    {
+      "epoch": 0.6360708534621579,
+      "grad_norm": 0.7769071284034672,
+      "learning_rate": 7.707796845309782e-06,
+      "loss": 0.4197,
+      "num_input_tokens_seen": 87698832,
+      "step": 1580,
+      "train_runtime": 9431.8438,
+      "train_tokens_per_second": 9298.164
+    },
+    {
+      "epoch": 0.6380837359098228,
+      "grad_norm": 0.9118083880893668,
+      "learning_rate": 7.69449317957788e-06,
+      "loss": 0.4038,
+      "num_input_tokens_seen": 87983560,
+      "step": 1585,
+      "train_runtime": 9462.6486,
+      "train_tokens_per_second": 9297.985
+    },
+    {
+      "epoch": 0.6400966183574879,
+      "grad_norm": 0.7120844282259289,
+      "learning_rate": 7.681162576592896e-06,
+      "loss": 0.4143,
+      "num_input_tokens_seen": 88238112,
+      "step": 1590,
+      "train_runtime": 9487.721,
+      "train_tokens_per_second": 9300.243
+    },
+    {
+      "epoch": 0.6421095008051529,
+      "grad_norm": 1.1565344499264973,
+      "learning_rate": 7.667805169622874e-06,
+      "loss": 0.4131,
+      "num_input_tokens_seen": 88514136,
+      "step": 1595,
+      "train_runtime": 9516.8398,
+      "train_tokens_per_second": 9300.791
+    },
+    {
+      "epoch": 0.644122383252818,
+      "grad_norm": 1.020789517866519,
+      "learning_rate": 7.65442109220382e-06,
+      "loss": 0.4108,
+      "num_input_tokens_seen": 88801416,
+      "step": 1600,
+      "train_runtime": 9548.1558,
+      "train_tokens_per_second": 9300.374
+    },
+    {
+      "epoch": 0.6461352657004831,
+      "grad_norm": 0.6569246845466671,
+      "learning_rate": 7.641010478138365e-06,
+      "loss": 0.3984,
+      "num_input_tokens_seen": 89095008,
+      "step": 1605,
+      "train_runtime": 9580.0323,
+      "train_tokens_per_second": 9300.074
+    },
+    {
+      "epoch": 0.6481481481481481,
+      "grad_norm": 0.8117701181662421,
+      "learning_rate": 7.627573461494431e-06,
+      "loss": 0.3996,
+      "num_input_tokens_seen": 89371968,
+      "step": 1610,
+      "train_runtime": 9609.4486,
+      "train_tokens_per_second": 9300.426
+    },
+    {
+      "epoch": 0.6501610305958132,
+      "grad_norm": 0.802627170404663,
+      "learning_rate": 7.614110176603893e-06,
+      "loss": 0.4212,
+      "num_input_tokens_seen": 89652280,
+      "step": 1615,
+      "train_runtime": 9636.3858,
+      "train_tokens_per_second": 9303.517
+    },
+    {
+      "epoch": 0.6521739130434783,
+      "grad_norm": 0.778857767803496,
+      "learning_rate": 7.600620758061233e-06,
+      "loss": 0.4112,
+      "num_input_tokens_seen": 89934216,
+      "step": 1620,
+      "train_runtime": 9664.3841,
+      "train_tokens_per_second": 9305.737
+    },
+    {
+      "epoch": 0.6541867954911433,
+      "grad_norm": 0.8528084820284613,
+      "learning_rate": 7.587105340722194e-06,
+      "loss": 0.4044,
+      "num_input_tokens_seen": 90201848,
+      "step": 1625,
+      "train_runtime": 9694.8443,
+      "train_tokens_per_second": 9304.105
+    },
+    {
+      "epoch": 0.6561996779388084,
+      "grad_norm": 0.8384144695794434,
+      "learning_rate": 7.573564059702432e-06,
+      "loss": 0.4221,
+      "num_input_tokens_seen": 90476912,
+      "step": 1630,
+      "train_runtime": 9729.4824,
+      "train_tokens_per_second": 9299.252
+    },
+    {
+      "epoch": 0.6582125603864735,
+      "grad_norm": 0.7164788919800471,
+      "learning_rate": 7.55999705037617e-06,
+      "loss": 0.3935,
+      "num_input_tokens_seen": 90757344,
+      "step": 1635,
+      "train_runtime": 9761.927,
+      "train_tokens_per_second": 9297.073
+    },
+    {
+      "epoch": 0.6602254428341385,
+      "grad_norm": 0.6969999879121849,
+      "learning_rate": 7.546404448374835e-06,
+      "loss": 0.4108,
+      "num_input_tokens_seen": 91022032,
+      "step": 1640,
+      "train_runtime": 9785.9052,
+      "train_tokens_per_second": 9301.34
+    },
+    {
+      "epoch": 0.6622383252818036,
+      "grad_norm": 0.7640210704381736,
+      "learning_rate": 7.532786389585715e-06,
+      "loss": 0.3911,
+      "num_input_tokens_seen": 91292160,
+      "step": 1645,
+      "train_runtime": 9817.0135,
+      "train_tokens_per_second": 9299.382
+    },
+    {
+      "epoch": 0.6642512077294686,
+      "grad_norm": 0.8806554782695155,
+      "learning_rate": 7.5191430101505894e-06,
+      "loss": 0.4167,
+      "num_input_tokens_seen": 91547568,
+      "step": 1650,
+      "train_runtime": 9848.1072,
+      "train_tokens_per_second": 9295.956
+    },
+    {
+      "epoch": 0.6662640901771336,
+      "grad_norm": 0.7565596746445855,
+      "learning_rate": 7.5054744464643666e-06,
+      "loss": 0.4038,
+      "num_input_tokens_seen": 91809704,
+      "step": 1655,
+      "train_runtime": 9879.4862,
+      "train_tokens_per_second": 9292.963
+    },
+    {
+      "epoch": 0.6682769726247987,
+      "grad_norm": 0.7743504130877569,
+      "learning_rate": 7.491780835173735e-06,
+      "loss": 0.3984,
+      "num_input_tokens_seen": 92110000,
+      "step": 1660,
+      "train_runtime": 9903.829,
+      "train_tokens_per_second": 9300.443
+    },
+    {
+      "epoch": 0.6702898550724637,
+      "grad_norm": 0.830081498149515,
+      "learning_rate": 7.478062313175784e-06,
+      "loss": 0.3998,
+      "num_input_tokens_seen": 92362424,
+      "step": 1665,
+      "train_runtime": 9930.6532,
+      "train_tokens_per_second": 9300.74
+    },
+    {
+      "epoch": 0.6723027375201288,
+      "grad_norm": 0.7629380180843693,
+      "learning_rate": 7.464319017616638e-06,
+      "loss": 0.4099,
+      "num_input_tokens_seen": 92638656,
+      "step": 1670,
+      "train_runtime": 9964.0876,
+      "train_tokens_per_second": 9297.254
+    },
+    {
+      "epoch": 0.6743156199677939,
+      "grad_norm": 0.6821129536408884,
+      "learning_rate": 7.450551085890087e-06,
+      "loss": 0.4278,
+      "num_input_tokens_seen": 92921648,
+      "step": 1675,
+      "train_runtime": 9997.6409,
+      "train_tokens_per_second": 9294.357
+    },
+    {
+      "epoch": 0.6763285024154589,
+      "grad_norm": 0.7109160905237335,
+      "learning_rate": 7.4367586556362125e-06,
+      "loss": 0.3996,
+      "num_input_tokens_seen": 93188816,
+      "step": 1680,
+      "train_runtime": 10026.9369,
+      "train_tokens_per_second": 9293.847
+    },
+    {
+      "epoch": 0.678341384863124,
+      "grad_norm": 0.7314806086348711,
+      "learning_rate": 7.422941864740012e-06,
+      "loss": 0.4223,
+      "num_input_tokens_seen": 93469840,
+      "step": 1685,
+      "train_runtime": 10049.8583,
+      "train_tokens_per_second": 9300.613
+    },
+    {
+      "epoch": 0.6803542673107891,
+      "grad_norm": 0.8153676167745848,
+      "learning_rate": 7.40910085133002e-06,
+      "loss": 0.4353,
+      "num_input_tokens_seen": 93728136,
+      "step": 1690,
+      "train_runtime": 10085.7831,
+      "train_tokens_per_second": 9293.095
+    },
+    {
+      "epoch": 0.6823671497584541,
+      "grad_norm": 0.7624434169500551,
+      "learning_rate": 7.395235753776925e-06,
+      "loss": 0.4033,
+      "num_input_tokens_seen": 93992976,
+      "step": 1695,
+      "train_runtime": 10114.9703,
+      "train_tokens_per_second": 9292.462
+    },
+    {
+      "epoch": 0.6843800322061192,
+      "grad_norm": 0.9293179551465384,
+      "learning_rate": 7.381346710692192e-06,
+      "loss": 0.4061,
+      "num_input_tokens_seen": 94261376,
+      "step": 1700,
+      "train_runtime": 10134.1424,
+      "train_tokens_per_second": 9301.367
+    },
+    {
+      "epoch": 0.6863929146537843,
+      "grad_norm": 0.7614981667279354,
+      "learning_rate": 7.3674338609266705e-06,
+      "loss": 0.415,
+      "num_input_tokens_seen": 94533288,
+      "step": 1705,
+      "train_runtime": 10163.1213,
+      "train_tokens_per_second": 9301.6
+    },
+    {
+      "epoch": 0.6884057971014492,
+      "grad_norm": 0.8945677637099625,
+      "learning_rate": 7.3534973435692106e-06,
+      "loss": 0.4091,
+      "num_input_tokens_seen": 94780728,
+      "step": 1710,
+      "train_runtime": 10187.2798,
+      "train_tokens_per_second": 9303.831
+    },
+    {
+      "epoch": 0.6904186795491143,
+      "grad_norm": 0.8209860520811668,
+      "learning_rate": 7.3395372979452675e-06,
+      "loss": 0.4162,
+      "num_input_tokens_seen": 95052216,
+      "step": 1715,
+      "train_runtime": 10218.1152,
+      "train_tokens_per_second": 9302.324
+    },
+    {
+      "epoch": 0.6924315619967794,
+      "grad_norm": 0.7625878548231061,
+      "learning_rate": 7.325553863615516e-06,
+      "loss": 0.39,
+      "num_input_tokens_seen": 95317512,
+      "step": 1720,
+      "train_runtime": 10246.2772,
+      "train_tokens_per_second": 9302.648
+    },
+    {
+      "epoch": 0.6944444444444444,
+      "grad_norm": 0.7242204103504999,
+      "learning_rate": 7.311547180374448e-06,
+      "loss": 0.3915,
+      "num_input_tokens_seen": 95595816,
+      "step": 1725,
+      "train_runtime": 10269.0328,
+      "train_tokens_per_second": 9309.135
+    },
+    {
+      "epoch": 0.6964573268921095,
+      "grad_norm": 0.8312747967250838,
+      "learning_rate": 7.297517388248983e-06,
+      "loss": 0.4359,
+      "num_input_tokens_seen": 95857792,
+      "step": 1730,
+      "train_runtime": 10299.3139,
+      "train_tokens_per_second": 9307.202
+    },
+    {
+      "epoch": 0.6984702093397746,
+      "grad_norm": 0.7813375145597212,
+      "learning_rate": 7.283464627497056e-06,
+      "loss": 0.4279,
+      "num_input_tokens_seen": 96111304,
+      "step": 1735,
+      "train_runtime": 10331.5993,
+      "train_tokens_per_second": 9302.655
+    },
+    {
+      "epoch": 0.7004830917874396,
+      "grad_norm": 0.787882518766384,
+      "learning_rate": 7.269389038606231e-06,
+      "loss": 0.4117,
+      "num_input_tokens_seen": 96375072,
+      "step": 1740,
+      "train_runtime": 10354.8105,
+      "train_tokens_per_second": 9307.275
+    },
+    {
+      "epoch": 0.7024959742351047,
+      "grad_norm": 0.750610549048134,
+      "learning_rate": 7.255290762292286e-06,
+      "loss": 0.3852,
+      "num_input_tokens_seen": 96665096,
+      "step": 1745,
+      "train_runtime": 10384.3089,
+      "train_tokens_per_second": 9308.765
+    },
+    {
+      "epoch": 0.7045088566827697,
+      "grad_norm": 0.7300867273807855,
+      "learning_rate": 7.241169939497801e-06,
+      "loss": 0.4035,
+      "num_input_tokens_seen": 96944624,
+      "step": 1750,
+      "train_runtime": 10412.9763,
+      "train_tokens_per_second": 9309.982
+    },
+    {
+      "epoch": 0.7065217391304348,
+      "grad_norm": 0.7713030902557706,
+      "learning_rate": 7.227026711390766e-06,
+      "loss": 0.415,
+      "num_input_tokens_seen": 97202880,
+      "step": 1755,
+      "train_runtime": 10441.2599,
+      "train_tokens_per_second": 9309.497
+    },
+    {
+      "epoch": 0.7085346215780999,
+      "grad_norm": 0.8907574599657645,
+      "learning_rate": 7.2128612193631595e-06,
+      "loss": 0.4244,
+      "num_input_tokens_seen": 97511016,
+      "step": 1760,
+      "train_runtime": 10468.4289,
+      "train_tokens_per_second": 9314.771
+    },
+    {
+      "epoch": 0.7105475040257649,
+      "grad_norm": 0.712248488256245,
+      "learning_rate": 7.198673605029529e-06,
+      "loss": 0.4286,
+      "num_input_tokens_seen": 97781072,
+      "step": 1765,
+      "train_runtime": 10499.2572,
+      "train_tokens_per_second": 9313.142
+    },
+    {
+      "epoch": 0.7125603864734299,
+      "grad_norm": 0.7735802152450216,
+      "learning_rate": 7.18446401022559e-06,
+      "loss": 0.409,
+      "num_input_tokens_seen": 98050776,
+      "step": 1770,
+      "train_runtime": 10535.6233,
+      "train_tokens_per_second": 9306.595
+    },
+    {
+      "epoch": 0.714573268921095,
+      "grad_norm": 0.7152101465331846,
+      "learning_rate": 7.1702325770067935e-06,
+      "loss": 0.4178,
+      "num_input_tokens_seen": 98331192,
+      "step": 1775,
+      "train_runtime": 10568.2036,
+      "train_tokens_per_second": 9304.438
+    },
+    {
+      "epoch": 0.71658615136876,
+      "grad_norm": 0.7496795613405771,
+      "learning_rate": 7.155979447646916e-06,
+      "loss": 0.3932,
+      "num_input_tokens_seen": 98587896,
+      "step": 1780,
+      "train_runtime": 10592.683,
+      "train_tokens_per_second": 9307.169
+    },
+    {
+      "epoch": 0.7185990338164251,
+      "grad_norm": 0.828720621744438,
+      "learning_rate": 7.141704764636632e-06,
+      "loss": 0.4151,
+      "num_input_tokens_seen": 98849320,
+      "step": 1785,
+      "train_runtime": 10622.6204,
+      "train_tokens_per_second": 9305.549
+    },
+    {
+      "epoch": 0.7206119162640902,
+      "grad_norm": 0.719135405670325,
+      "learning_rate": 7.127408670682094e-06,
+      "loss": 0.4117,
+      "num_input_tokens_seen": 99119776,
+      "step": 1790,
+      "train_runtime": 10656.0193,
+      "train_tokens_per_second": 9301.764
+    },
+    {
+      "epoch": 0.7226247987117552,
+      "grad_norm": 0.7449978998893,
+      "learning_rate": 7.113091308703498e-06,
+      "loss": 0.3865,
+      "num_input_tokens_seen": 99397616,
+      "step": 1795,
+      "train_runtime": 10687.4255,
+      "train_tokens_per_second": 9300.427
+    },
+    {
+      "epoch": 0.7246376811594203,
+      "grad_norm": 0.6953501717545923,
+      "learning_rate": 7.0987528218336655e-06,
+      "loss": 0.4035,
+      "num_input_tokens_seen": 99675792,
+      "step": 1800,
+      "train_runtime": 10718.733,
+      "train_tokens_per_second": 9299.214
+    },
+    {
+      "epoch": 0.7266505636070854,
+      "grad_norm": 0.7388591099682715,
+      "learning_rate": 7.084393353416601e-06,
+      "loss": 0.3899,
+      "num_input_tokens_seen": 99951568,
+      "step": 1805,
+      "train_runtime": 10749.1466,
+      "train_tokens_per_second": 9298.558
+    },
+    {
+      "epoch": 0.7286634460547504,
+      "grad_norm": 0.7097960326496885,
+      "learning_rate": 7.070013047006068e-06,
+      "loss": 0.4243,
+      "num_input_tokens_seen": 100236096,
+      "step": 1810,
+      "train_runtime": 10782.2828,
+      "train_tokens_per_second": 9296.371
+    },
+    {
+      "epoch": 0.7306763285024155,
+      "grad_norm": 0.6864574084119663,
+      "learning_rate": 7.055612046364151e-06,
+      "loss": 0.4052,
+      "num_input_tokens_seen": 100505616,
+      "step": 1815,
+      "train_runtime": 10818.2258,
+      "train_tokens_per_second": 9290.397
+    },
+    {
+      "epoch": 0.7326892109500805,
+      "grad_norm": 0.7565530144066822,
+      "learning_rate": 7.041190495459816e-06,
+      "loss": 0.4107,
+      "num_input_tokens_seen": 100798848,
+      "step": 1820,
+      "train_runtime": 10852.7866,
+      "train_tokens_per_second": 9287.831
+    },
+    {
+      "epoch": 0.7347020933977456,
+      "grad_norm": 0.7410753349128333,
+      "learning_rate": 7.026748538467474e-06,
+      "loss": 0.4015,
+      "num_input_tokens_seen": 101070488,
+      "step": 1825,
+      "train_runtime": 10881.7282,
+      "train_tokens_per_second": 9288.092
+    },
+    {
+      "epoch": 0.7367149758454107,
+      "grad_norm": 0.7275446753992727,
+      "learning_rate": 7.012286319765538e-06,
+      "loss": 0.4101,
+      "num_input_tokens_seen": 101358128,
+      "step": 1830,
+      "train_runtime": 10913.1729,
+      "train_tokens_per_second": 9287.686
+    },
+    {
+      "epoch": 0.7387278582930756,
+      "grad_norm": 0.8916444782007619,
+      "learning_rate": 6.997803983934982e-06,
+      "loss": 0.4067,
+      "num_input_tokens_seen": 101643344,
+      "step": 1835,
+      "train_runtime": 10944.8125,
+      "train_tokens_per_second": 9286.897
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.7096495766706037,
+      "learning_rate": 6.983301675757892e-06,
+      "loss": 0.3939,
+      "num_input_tokens_seen": 101920000,
+      "step": 1840,
+      "train_runtime": 10973.4636,
+      "train_tokens_per_second": 9287.861
+    },
+    {
+      "epoch": 0.7427536231884058,
+      "grad_norm": 0.7874518992370024,
+      "learning_rate": 6.96877954021602e-06,
+      "loss": 0.4289,
+      "num_input_tokens_seen": 102208776,
+      "step": 1845,
+      "train_runtime": 11000.5187,
+      "train_tokens_per_second": 9291.269
+    },
+    {
+      "epoch": 0.7447665056360708,
+      "grad_norm": 0.7174361452345764,
+      "learning_rate": 6.9542377224893365e-06,
+      "loss": 0.397,
+      "num_input_tokens_seen": 102480896,
+      "step": 1850,
+      "train_runtime": 11032.9038,
+      "train_tokens_per_second": 9288.66
+    },
+    {
+      "epoch": 0.7467793880837359,
+      "grad_norm": 0.8857917776407266,
+      "learning_rate": 6.939676367954582e-06,
+      "loss": 0.4126,
+      "num_input_tokens_seen": 102761648,
+      "step": 1855,
+      "train_runtime": 11069.7423,
+      "train_tokens_per_second": 9283.111
+    },
+    {
+      "epoch": 0.748792270531401,
+      "grad_norm": 0.7544489839976317,
+      "learning_rate": 6.925095622183801e-06,
+      "loss": 0.3929,
+      "num_input_tokens_seen": 103017568,
+      "step": 1860,
+      "train_runtime": 11096.9418,
+      "train_tokens_per_second": 9283.42
+    },
+    {
+      "epoch": 0.750805152979066,
+      "grad_norm": 0.7140945958546089,
+      "learning_rate": 6.910495630942899e-06,
+      "loss": 0.3881,
+      "num_input_tokens_seen": 103298384,
+      "step": 1865,
+      "train_runtime": 11125.4672,
+      "train_tokens_per_second": 9284.858
+    },
+    {
+      "epoch": 0.7528180354267311,
+      "grad_norm": 0.8215969191604309,
+      "learning_rate": 6.895876540190185e-06,
+      "loss": 0.4048,
+      "num_input_tokens_seen": 103585408,
+      "step": 1870,
+      "train_runtime": 11162.6384,
+      "train_tokens_per_second": 9279.653
+    },
+    {
+      "epoch": 0.7548309178743962,
+      "grad_norm": 0.7600964332954867,
+      "learning_rate": 6.8812384960749024e-06,
+      "loss": 0.39,
+      "num_input_tokens_seen": 103868056,
+      "step": 1875,
+      "train_runtime": 11193.4162,
+      "train_tokens_per_second": 9279.388
+    },
+    {
+      "epoch": 0.7568438003220612,
+      "grad_norm": 0.6967150873736934,
+      "learning_rate": 6.8665816449357795e-06,
+      "loss": 0.4114,
+      "num_input_tokens_seen": 104123080,
+      "step": 1880,
+      "train_runtime": 11221.084,
+      "train_tokens_per_second": 9279.235
+    },
+    {
+      "epoch": 0.7588566827697263,
+      "grad_norm": 0.7531677844341517,
+      "learning_rate": 6.851906133299556e-06,
+      "loss": 0.4074,
+      "num_input_tokens_seen": 104389184,
+      "step": 1885,
+      "train_runtime": 11250.4979,
+      "train_tokens_per_second": 9278.628
+    },
+    {
+      "epoch": 0.7608695652173914,
+      "grad_norm": 0.739490274275597,
+      "learning_rate": 6.837212107879528e-06,
+      "loss": 0.3969,
+      "num_input_tokens_seen": 104685376,
+      "step": 1890,
+      "train_runtime": 11284.4634,
+      "train_tokens_per_second": 9276.948
+    },
+    {
+      "epoch": 0.7628824476650563,
+      "grad_norm": 0.836743288435197,
+      "learning_rate": 6.822499715574076e-06,
+      "loss": 0.4031,
+      "num_input_tokens_seen": 104970752,
+      "step": 1895,
+      "train_runtime": 11305.5373,
+      "train_tokens_per_second": 9284.897
+    },
+    {
+      "epoch": 0.7648953301127214,
+      "grad_norm": 0.7572572003700926,
+      "learning_rate": 6.8077691034651936e-06,
+      "loss": 0.3981,
+      "num_input_tokens_seen": 105245720,
+      "step": 1900,
+      "train_runtime": 11337.2114,
+      "train_tokens_per_second": 9283.211
+    },
+    {
+      "epoch": 0.7669082125603864,
+      "grad_norm": 0.6924824509813626,
+      "learning_rate": 6.7930204188170244e-06,
+      "loss": 0.4206,
+      "num_input_tokens_seen": 105532392,
+      "step": 1905,
+      "train_runtime": 11370.3789,
+      "train_tokens_per_second": 9281.343
+    },
+    {
+      "epoch": 0.7689210950080515,
+      "grad_norm": 0.7950851186977921,
+      "learning_rate": 6.778253809074384e-06,
+      "loss": 0.377,
+      "num_input_tokens_seen": 105816496,
+      "step": 1910,
+      "train_runtime": 11398.4739,
+      "train_tokens_per_second": 9283.392
+    },
+    {
+      "epoch": 0.7709339774557166,
+      "grad_norm": 0.9021912986436811,
+      "learning_rate": 6.7634694218612865e-06,
+      "loss": 0.4185,
+      "num_input_tokens_seen": 106061088,
+      "step": 1915,
+      "train_runtime": 11426.928,
+      "train_tokens_per_second": 9281.68
+    },
+    {
+      "epoch": 0.7729468599033816,
+      "grad_norm": 0.7504065463597139,
+      "learning_rate": 6.748667404979474e-06,
+      "loss": 0.3815,
+      "num_input_tokens_seen": 106348248,
+      "step": 1920,
+      "train_runtime": 11460.4051,
+      "train_tokens_per_second": 9279.624
+    },
+    {
+      "epoch": 0.7749597423510467,
+      "grad_norm": 0.9924061249878591,
+      "learning_rate": 6.733847906406933e-06,
+      "loss": 0.3946,
+      "num_input_tokens_seen": 106608736,
+      "step": 1925,
+      "train_runtime": 11483.195,
+      "train_tokens_per_second": 9283.891
+    },
+    {
+      "epoch": 0.7769726247987118,
+      "grad_norm": 0.747104973056609,
+      "learning_rate": 6.7190110742964155e-06,
+      "loss": 0.3968,
+      "num_input_tokens_seen": 106886400,
+      "step": 1930,
+      "train_runtime": 11512.1448,
+      "train_tokens_per_second": 9284.664
+    },
+    {
+      "epoch": 0.7789855072463768,
+      "grad_norm": 0.7843145322449393,
+      "learning_rate": 6.704157056973961e-06,
+      "loss": 0.391,
+      "num_input_tokens_seen": 107175088,
+      "step": 1935,
+      "train_runtime": 11546.096,
+      "train_tokens_per_second": 9282.366
+    },
+    {
+      "epoch": 0.7809983896940419,
+      "grad_norm": 0.7393308380484042,
+      "learning_rate": 6.6892860029374125e-06,
+      "loss": 0.4004,
+      "num_input_tokens_seen": 107448632,
+      "step": 1940,
+      "train_runtime": 11578.5199,
+      "train_tokens_per_second": 9279.997
+    },
+    {
+      "epoch": 0.783011272141707,
+      "grad_norm": 0.7099656749714811,
+      "learning_rate": 6.674398060854931e-06,
+      "loss": 0.3858,
+      "num_input_tokens_seen": 107718736,
+      "step": 1945,
+      "train_runtime": 11608.3027,
+      "train_tokens_per_second": 9279.456
+    },
+    {
+      "epoch": 0.785024154589372,
+      "grad_norm": 0.8096969009624011,
+      "learning_rate": 6.659493379563511e-06,
+      "loss": 0.3965,
+      "num_input_tokens_seen": 108021064,
+      "step": 1950,
+      "train_runtime": 11640.2335,
+      "train_tokens_per_second": 9279.974
+    },
+    {
+      "epoch": 0.7870370370370371,
+      "grad_norm": 0.6893544295938017,
+      "learning_rate": 6.644572108067487e-06,
+      "loss": 0.4064,
+      "num_input_tokens_seen": 108296792,
+      "step": 1955,
+      "train_runtime": 11672.4171,
+      "train_tokens_per_second": 9278.009
+    },
+    {
+      "epoch": 0.789049919484702,
+      "grad_norm": 0.9051143616508919,
+      "learning_rate": 6.6296343955370565e-06,
+      "loss": 0.3888,
+      "num_input_tokens_seen": 108557568,
+      "step": 1960,
+      "train_runtime": 11704.9466,
+      "train_tokens_per_second": 9274.503
+    },
+    {
+      "epoch": 0.7910628019323671,
+      "grad_norm": 0.688671545189268,
+      "learning_rate": 6.614680391306772e-06,
+      "loss": 0.3902,
+      "num_input_tokens_seen": 108817104,
+      "step": 1965,
+      "train_runtime": 11734.913,
+      "train_tokens_per_second": 9272.937
+    },
+    {
+      "epoch": 0.7930756843800322,
+      "grad_norm": 0.7331311344757184,
+      "learning_rate": 6.59971024487406e-06,
+      "loss": 0.3893,
+      "num_input_tokens_seen": 109093256,
+      "step": 1970,
+      "train_runtime": 11764.5957,
+      "train_tokens_per_second": 9273.014
+    },
+    {
+      "epoch": 0.7950885668276972,
+      "grad_norm": 0.7165818734455979,
+      "learning_rate": 6.5847241058977265e-06,
+      "loss": 0.3993,
+      "num_input_tokens_seen": 109383728,
+      "step": 1975,
+      "train_runtime": 11797.704,
+      "train_tokens_per_second": 9271.612
+    },
+    {
+      "epoch": 0.7971014492753623,
+      "grad_norm": 0.7977027081583492,
+      "learning_rate": 6.569722124196449e-06,
+      "loss": 0.3951,
+      "num_input_tokens_seen": 109659440,
+      "step": 1980,
+      "train_runtime": 11824.8966,
+      "train_tokens_per_second": 9273.607
+    },
+    {
+      "epoch": 0.7991143317230274,
+      "grad_norm": 0.7666924246188899,
+      "learning_rate": 6.554704449747295e-06,
+      "loss": 0.3876,
+      "num_input_tokens_seen": 109955232,
+      "step": 1985,
+      "train_runtime": 11856.9293,
+      "train_tokens_per_second": 9273.5
+    },
+    {
+      "epoch": 0.8011272141706924,
+      "grad_norm": 0.692098708811024,
+      "learning_rate": 6.539671232684215e-06,
+      "loss": 0.3825,
+      "num_input_tokens_seen": 110223952,
+      "step": 1990,
+      "train_runtime": 11881.7316,
+      "train_tokens_per_second": 9276.758
+    },
+    {
+      "epoch": 0.8031400966183575,
+      "grad_norm": 0.7528244866011002,
+      "learning_rate": 6.5246226232965324e-06,
+      "loss": 0.4025,
+      "num_input_tokens_seen": 110524704,
+      "step": 1995,
+      "train_runtime": 11913.6696,
+      "train_tokens_per_second": 9277.134
+    },
+    {
+      "epoch": 0.8051529790660226,
+      "grad_norm": 0.7051788232024218,
+      "learning_rate": 6.5095587720274585e-06,
+      "loss": 0.3925,
+      "num_input_tokens_seen": 110793688,
+      "step": 2000,
+      "train_runtime": 11940.2421,
+      "train_tokens_per_second": 9279.015
+    },
+    {
+      "epoch": 0.8071658615136876,
+      "grad_norm": 0.7085415031143671,
+      "learning_rate": 6.49447982947258e-06,
+      "loss": 0.3961,
+      "num_input_tokens_seen": 111090512,
+      "step": 2005,
+      "train_runtime": 11973.2767,
+      "train_tokens_per_second": 9278.205
+    },
+    {
+      "epoch": 0.8091787439613527,
+      "grad_norm": 0.7165852444456339,
+      "learning_rate": 6.479385946378345e-06,
+      "loss": 0.3922,
+      "num_input_tokens_seen": 111378176,
+      "step": 2010,
+      "train_runtime": 12004.5642,
+      "train_tokens_per_second": 9277.986
+    },
+    {
+      "epoch": 0.8111916264090178,
+      "grad_norm": 0.7368612295158381,
+      "learning_rate": 6.4642772736405756e-06,
+      "loss": 0.3964,
+      "num_input_tokens_seen": 111635624,
+      "step": 2015,
+      "train_runtime": 12030.3883,
+      "train_tokens_per_second": 9279.47
+    },
+    {
+      "epoch": 0.8132045088566827,
+      "grad_norm": 0.6807649596796009,
+      "learning_rate": 6.44915396230294e-06,
+      "loss": 0.4081,
+      "num_input_tokens_seen": 111931424,
+      "step": 2020,
+      "train_runtime": 12064.4531,
+      "train_tokens_per_second": 9277.787
+    },
+    {
+      "epoch": 0.8152173913043478,
+      "grad_norm": 0.6941522364672225,
+      "learning_rate": 6.434016163555452e-06,
+      "loss": 0.41,
+      "num_input_tokens_seen": 112222240,
+      "step": 2025,
+      "train_runtime": 12087.8376,
+      "train_tokens_per_second": 9283.897
+    },
+    {
+      "epoch": 0.8172302737520128,
+      "grad_norm": 0.7027289982332953,
+      "learning_rate": 6.418864028732963e-06,
+      "loss": 0.3759,
+      "num_input_tokens_seen": 112510040,
+      "step": 2030,
+      "train_runtime": 12117.3903,
+      "train_tokens_per_second": 9285.006
+    },
+    {
+      "epoch": 0.8192431561996779,
+      "grad_norm": 0.6452596611591622,
+      "learning_rate": 6.403697709313639e-06,
+      "loss": 0.3732,
+      "num_input_tokens_seen": 112798112,
+      "step": 2035,
+      "train_runtime": 12147.0537,
+      "train_tokens_per_second": 9286.047
+    },
+    {
+      "epoch": 0.821256038647343,
+      "grad_norm": 0.7861663607777234,
+      "learning_rate": 6.388517356917454e-06,
+      "loss": 0.4093,
+      "num_input_tokens_seen": 113070456,
+      "step": 2040,
+      "train_runtime": 12173.1405,
+      "train_tokens_per_second": 9288.52
+    },
+    {
+      "epoch": 0.823268921095008,
+      "grad_norm": 0.7486831445721319,
+      "learning_rate": 6.373323123304671e-06,
+      "loss": 0.3904,
+      "num_input_tokens_seen": 113343376,
+      "step": 2045,
+      "train_runtime": 12203.1997,
+      "train_tokens_per_second": 9288.005
+    },
+    {
+      "epoch": 0.8252818035426731,
+      "grad_norm": 0.7250130841822909,
+      "learning_rate": 6.358115160374326e-06,
+      "loss": 0.3883,
+      "num_input_tokens_seen": 113614944,
+      "step": 2050,
+      "train_runtime": 12237.7983,
+      "train_tokens_per_second": 9283.937
+    },
+    {
+      "epoch": 0.8272946859903382,
+      "grad_norm": 0.7592651019588287,
+      "learning_rate": 6.342893620162709e-06,
+      "loss": 0.4066,
+      "num_input_tokens_seen": 113914464,
+      "step": 2055,
+      "train_runtime": 12269.0497,
+      "train_tokens_per_second": 9284.701
+    },
+    {
+      "epoch": 0.8293075684380032,
+      "grad_norm": 0.8973442268194562,
+      "learning_rate": 6.3276586548418454e-06,
+      "loss": 0.3999,
+      "num_input_tokens_seen": 114157984,
+      "step": 2060,
+      "train_runtime": 12293.2014,
+      "train_tokens_per_second": 9286.27
+    },
+    {
+      "epoch": 0.8313204508856683,
+      "grad_norm": 0.750213524348002,
+      "learning_rate": 6.312410416717969e-06,
+      "loss": 0.4058,
+      "num_input_tokens_seen": 114456080,
+      "step": 2065,
+      "train_runtime": 12319.7952,
+      "train_tokens_per_second": 9290.421
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 1.1010047204334936,
+      "learning_rate": 6.29714905823001e-06,
+      "loss": 0.4164,
+      "num_input_tokens_seen": 114731320,
+      "step": 2070,
+      "train_runtime": 12347.9132,
+      "train_tokens_per_second": 9291.555
+    },
+    {
+      "epoch": 0.8353462157809984,
+      "grad_norm": 0.6684830320271861,
+      "learning_rate": 6.281874731948059e-06,
+      "loss": 0.4316,
+      "num_input_tokens_seen": 115029840,
+      "step": 2075,
+      "train_runtime": 12376.7565,
+      "train_tokens_per_second": 9294.021
+    },
+    {
+      "epoch": 0.8373590982286635,
+      "grad_norm": 0.9962569348783838,
+      "learning_rate": 6.266587590571852e-06,
+      "loss": 0.4102,
+      "num_input_tokens_seen": 115299712,
+      "step": 2080,
+      "train_runtime": 12405.6584,
+      "train_tokens_per_second": 9294.123
+    },
+    {
+      "epoch": 0.8393719806763285,
+      "grad_norm": 0.7468273606895267,
+      "learning_rate": 6.25128778692924e-06,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 115573048,
+      "step": 2085,
+      "train_runtime": 12434.8852,
+      "train_tokens_per_second": 9294.259
+    },
+    {
+      "epoch": 0.8413848631239935,
+      "grad_norm": 0.7535476931068493,
+      "learning_rate": 6.235975473974652e-06,
+      "loss": 0.4237,
+      "num_input_tokens_seen": 115860800,
+      "step": 2090,
+      "train_runtime": 12470.5401,
+      "train_tokens_per_second": 9290.76
+    },
+    {
+      "epoch": 0.8433977455716586,
+      "grad_norm": 0.7415339871426038,
+      "learning_rate": 6.220650804787587e-06,
+      "loss": 0.4177,
+      "num_input_tokens_seen": 116151880,
+      "step": 2095,
+      "train_runtime": 12498.1208,
+      "train_tokens_per_second": 9293.548
+    },
+    {
+      "epoch": 0.8454106280193237,
+      "grad_norm": 0.8253083225510711,
+      "learning_rate": 6.205313932571062e-06,
+      "loss": 0.3926,
+      "num_input_tokens_seen": 116398848,
+      "step": 2100,
+      "train_runtime": 12530.263,
+      "train_tokens_per_second": 9289.418
+    },
+    {
+      "epoch": 0.8474235104669887,
+      "grad_norm": 0.7576322131411791,
+      "learning_rate": 6.189965010650092e-06,
+      "loss": 0.4029,
+      "num_input_tokens_seen": 116688240,
+      "step": 2105,
+      "train_runtime": 12557.3549,
+      "train_tokens_per_second": 9292.422
+    },
+    {
+      "epoch": 0.8494363929146538,
+      "grad_norm": 0.7322750671925461,
+      "learning_rate": 6.174604192470156e-06,
+      "loss": 0.3806,
+      "num_input_tokens_seen": 116981576,
+      "step": 2110,
+      "train_runtime": 12589.1697,
+      "train_tokens_per_second": 9292.239
+    },
+    {
+      "epoch": 0.8514492753623188,
+      "grad_norm": 0.790254307835608,
+      "learning_rate": 6.159231631595663e-06,
+      "loss": 0.3976,
+      "num_input_tokens_seen": 117277320,
+      "step": 2115,
+      "train_runtime": 12615.0413,
+      "train_tokens_per_second": 9296.626
+    },
+    {
+      "epoch": 0.8534621578099839,
+      "grad_norm": 0.829071067604564,
+      "learning_rate": 6.1438474817084085e-06,
+      "loss": 0.4073,
+      "num_input_tokens_seen": 117539776,
+      "step": 2120,
+      "train_runtime": 12638.9997,
+      "train_tokens_per_second": 9299.769
+    },
+    {
+      "epoch": 0.855475040257649,
+      "grad_norm": 0.804926020396515,
+      "learning_rate": 6.128451896606054e-06,
+      "loss": 0.3865,
+      "num_input_tokens_seen": 117812360,
+      "step": 2125,
+      "train_runtime": 12659.248,
+      "train_tokens_per_second": 9306.426
+    },
+    {
+      "epoch": 0.857487922705314,
+      "grad_norm": 0.8859588309881129,
+      "learning_rate": 6.113045030200574e-06,
+      "loss": 0.3891,
+      "num_input_tokens_seen": 118081880,
+      "step": 2130,
+      "train_runtime": 12682.6147,
+      "train_tokens_per_second": 9310.531
+    },
+    {
+      "epoch": 0.8595008051529791,
+      "grad_norm": 0.8179054800389454,
+      "learning_rate": 6.0976270365167275e-06,
+      "loss": 0.3859,
+      "num_input_tokens_seen": 118333920,
+      "step": 2135,
+      "train_runtime": 12714.1661,
+      "train_tokens_per_second": 9307.25
+    },
+    {
+      "epoch": 0.8615136876006442,
+      "grad_norm": 0.7744298666299604,
+      "learning_rate": 6.0821980696905145e-06,
+      "loss": 0.4171,
+      "num_input_tokens_seen": 118630048,
+      "step": 2140,
+      "train_runtime": 12742.0117,
+      "train_tokens_per_second": 9310.151
+    },
+    {
+      "epoch": 0.8635265700483091,
+      "grad_norm": 0.8122944522241792,
+      "learning_rate": 6.066758283967633e-06,
+      "loss": 0.3983,
+      "num_input_tokens_seen": 118910024,
+      "step": 2145,
+      "train_runtime": 12768.569,
+      "train_tokens_per_second": 9312.713
+    },
+    {
+      "epoch": 0.8655394524959742,
+      "grad_norm": 0.7334131895666584,
+      "learning_rate": 6.0513078337019395e-06,
+      "loss": 0.378,
+      "num_input_tokens_seen": 119162688,
+      "step": 2150,
+      "train_runtime": 12794.47,
+      "train_tokens_per_second": 9313.609
+    },
+    {
+      "epoch": 0.8675523349436393,
+      "grad_norm": 0.8331239832134099,
+      "learning_rate": 6.0358468733539064e-06,
+      "loss": 0.4455,
+      "num_input_tokens_seen": 119436712,
+      "step": 2155,
+      "train_runtime": 12825.0349,
+      "train_tokens_per_second": 9312.779
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 0.7279560824537302,
+      "learning_rate": 6.020375557489078e-06,
+      "loss": 0.3911,
+      "num_input_tokens_seen": 119698112,
+      "step": 2160,
+      "train_runtime": 12848.8097,
+      "train_tokens_per_second": 9315.891
+    },
+    {
+      "epoch": 0.8715780998389694,
+      "grad_norm": 0.7001608470853314,
+      "learning_rate": 6.004894040776521e-06,
+      "loss": 0.3934,
+      "num_input_tokens_seen": 119987320,
+      "step": 2165,
+      "train_runtime": 12873.262,
+      "train_tokens_per_second": 9320.662
+    },
+    {
+      "epoch": 0.8735909822866345,
+      "grad_norm": 0.7127827055471014,
+      "learning_rate": 5.9894024779872865e-06,
+      "loss": 0.3839,
+      "num_input_tokens_seen": 120270896,
+      "step": 2170,
+      "train_runtime": 12903.3187,
+      "train_tokens_per_second": 9320.927
+    },
+    {
+      "epoch": 0.8756038647342995,
+      "grad_norm": 0.7096465124206889,
+      "learning_rate": 5.973901023992852e-06,
+      "loss": 0.4015,
+      "num_input_tokens_seen": 120577336,
+      "step": 2175,
+      "train_runtime": 12936.1114,
+      "train_tokens_per_second": 9320.988
+    },
+    {
+      "epoch": 0.8776167471819646,
+      "grad_norm": 0.6900365082143992,
+      "learning_rate": 5.9583898337635835e-06,
+      "loss": 0.4049,
+      "num_input_tokens_seen": 120839696,
+      "step": 2180,
+      "train_runtime": 12961.6154,
+      "train_tokens_per_second": 9322.889
+    },
+    {
+      "epoch": 0.8796296296296297,
+      "grad_norm": 0.6758900910753914,
+      "learning_rate": 5.9428690623671796e-06,
+      "loss": 0.3835,
+      "num_input_tokens_seen": 121130312,
+      "step": 2185,
+      "train_runtime": 12985.6953,
+      "train_tokens_per_second": 9327.98
+    },
+    {
+      "epoch": 0.8816425120772947,
+      "grad_norm": 0.7944161456158574,
+      "learning_rate": 5.927338864967121e-06,
+      "loss": 0.3999,
+      "num_input_tokens_seen": 121377376,
+      "step": 2190,
+      "train_runtime": 13012.0449,
+      "train_tokens_per_second": 9328.078
+    },
+    {
+      "epoch": 0.8836553945249598,
+      "grad_norm": 0.6883192612916793,
+      "learning_rate": 5.911799396821129e-06,
+      "loss": 0.384,
+      "num_input_tokens_seen": 121653072,
+      "step": 2195,
+      "train_runtime": 13042.6913,
+      "train_tokens_per_second": 9327.298
+    },
+    {
+      "epoch": 0.8856682769726248,
+      "grad_norm": 0.7761063530672764,
+      "learning_rate": 5.896250813279596e-06,
+      "loss": 0.3912,
+      "num_input_tokens_seen": 121941520,
+      "step": 2200,
+      "train_runtime": 13067.3891,
+      "train_tokens_per_second": 9331.743
+    },
+    {
+      "epoch": 0.8876811594202898,
+      "grad_norm": 0.7477373858141535,
+      "learning_rate": 5.8806932697840505e-06,
+      "loss": 0.4283,
+      "num_input_tokens_seen": 122218824,
+      "step": 2205,
+      "train_runtime": 13103.0069,
+      "train_tokens_per_second": 9327.54
+    },
+    {
+      "epoch": 0.8896940418679549,
+      "grad_norm": 0.7485343371246372,
+      "learning_rate": 5.86512692186559e-06,
+      "loss": 0.3821,
+      "num_input_tokens_seen": 122475232,
+      "step": 2210,
+      "train_runtime": 13129.2083,
+      "train_tokens_per_second": 9328.455
+    },
+    {
+      "epoch": 0.8917069243156199,
+      "grad_norm": 0.8607581419171527,
+      "learning_rate": 5.849551925143334e-06,
+      "loss": 0.4041,
+      "num_input_tokens_seen": 122735752,
+      "step": 2215,
+      "train_runtime": 13157.7834,
+      "train_tokens_per_second": 9327.996
+    },
+    {
+      "epoch": 0.893719806763285,
+      "grad_norm": 0.6939926252598053,
+      "learning_rate": 5.833968435322864e-06,
+      "loss": 0.3769,
+      "num_input_tokens_seen": 122996200,
+      "step": 2220,
+      "train_runtime": 13184.1228,
+      "train_tokens_per_second": 9329.115
+    },
+    {
+      "epoch": 0.8957326892109501,
+      "grad_norm": 0.7035129701939121,
+      "learning_rate": 5.818376608194669e-06,
+      "loss": 0.4118,
+      "num_input_tokens_seen": 123270096,
+      "step": 2225,
+      "train_runtime": 13219.3619,
+      "train_tokens_per_second": 9324.966
+    },
+    {
+      "epoch": 0.8977455716586151,
+      "grad_norm": 0.7470741723639145,
+      "learning_rate": 5.802776599632587e-06,
+      "loss": 0.3997,
+      "num_input_tokens_seen": 123552816,
+      "step": 2230,
+      "train_runtime": 13251.9959,
+      "train_tokens_per_second": 9323.336
+    },
+    {
+      "epoch": 0.8997584541062802,
+      "grad_norm": 0.7194781157776801,
+      "learning_rate": 5.787168565592246e-06,
+      "loss": 0.421,
+      "num_input_tokens_seen": 123844336,
+      "step": 2235,
+      "train_runtime": 13288.139,
+      "train_tokens_per_second": 9319.916
+    },
+    {
+      "epoch": 0.9017713365539453,
+      "grad_norm": 0.7364619886663366,
+      "learning_rate": 5.771552662109507e-06,
+      "loss": 0.3883,
+      "num_input_tokens_seen": 124126656,
+      "step": 2240,
+      "train_runtime": 13323.318,
+      "train_tokens_per_second": 9316.497
+    },
+    {
+      "epoch": 0.9037842190016103,
+      "grad_norm": 0.7495044251488605,
+      "learning_rate": 5.755929045298905e-06,
+      "loss": 0.4077,
+      "num_input_tokens_seen": 124377824,
+      "step": 2245,
+      "train_runtime": 13347.0688,
+      "train_tokens_per_second": 9318.737
+    },
+    {
+      "epoch": 0.9057971014492754,
+      "grad_norm": 0.7387542383777727,
+      "learning_rate": 5.740297871352083e-06,
+      "loss": 0.3801,
+      "num_input_tokens_seen": 124650968,
+      "step": 2250,
+      "train_runtime": 13377.8152,
+      "train_tokens_per_second": 9317.737
+    },
+    {
+      "epoch": 0.9078099838969405,
+      "grad_norm": 0.8284818162647343,
+      "learning_rate": 5.724659296536234e-06,
+      "loss": 0.4048,
+      "num_input_tokens_seen": 124882376,
+      "step": 2255,
+      "train_runtime": 13406.9024,
+      "train_tokens_per_second": 9314.782
+    },
+    {
+      "epoch": 0.9098228663446055,
+      "grad_norm": 0.7836495211389037,
+      "learning_rate": 5.709013477192541e-06,
+      "loss": 0.3879,
+      "num_input_tokens_seen": 125158504,
+      "step": 2260,
+      "train_runtime": 13440.4687,
+      "train_tokens_per_second": 9312.064
+    },
+    {
+      "epoch": 0.9118357487922706,
+      "grad_norm": 0.8230355820332363,
+      "learning_rate": 5.69336056973461e-06,
+      "loss": 0.407,
+      "num_input_tokens_seen": 125452616,
+      "step": 2265,
+      "train_runtime": 13468.1725,
+      "train_tokens_per_second": 9314.747
+    },
+    {
+      "epoch": 0.9138486312399355,
+      "grad_norm": 0.7174570231767459,
+      "learning_rate": 5.677700730646907e-06,
+      "loss": 0.383,
+      "num_input_tokens_seen": 125742352,
+      "step": 2270,
+      "train_runtime": 13501.2051,
+      "train_tokens_per_second": 9313.417
+    },
+    {
+      "epoch": 0.9158615136876006,
+      "grad_norm": 0.7445879268409392,
+      "learning_rate": 5.662034116483197e-06,
+      "loss": 0.4005,
+      "num_input_tokens_seen": 126054944,
+      "step": 2275,
+      "train_runtime": 13527.5332,
+      "train_tokens_per_second": 9318.398
+    },
+    {
+      "epoch": 0.9178743961352657,
+      "grad_norm": 0.6888028084778587,
+      "learning_rate": 5.646360883864975e-06,
+      "loss": 0.3846,
+      "num_input_tokens_seen": 126344720,
+      "step": 2280,
+      "train_runtime": 13554.316,
+      "train_tokens_per_second": 9321.365
+    },
+    {
+      "epoch": 0.9198872785829307,
+      "grad_norm": 0.8351061886236093,
+      "learning_rate": 5.630681189479902e-06,
+      "loss": 0.3868,
+      "num_input_tokens_seen": 126626232,
+      "step": 2285,
+      "train_runtime": 13589.2167,
+      "train_tokens_per_second": 9318.141
+    },
+    {
+      "epoch": 0.9219001610305958,
+      "grad_norm": 0.8864949989921621,
+      "learning_rate": 5.614995190080238e-06,
+      "loss": 0.4099,
+      "num_input_tokens_seen": 126878488,
+      "step": 2290,
+      "train_runtime": 13619.9302,
+      "train_tokens_per_second": 9315.649
+    },
+    {
+      "epoch": 0.9239130434782609,
+      "grad_norm": 0.7289414833615558,
+      "learning_rate": 5.5993030424812745e-06,
+      "loss": 0.4091,
+      "num_input_tokens_seen": 127194216,
+      "step": 2295,
+      "train_runtime": 13648.1735,
+      "train_tokens_per_second": 9319.505
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 0.998636117180562,
+      "learning_rate": 5.583604903559768e-06,
+      "loss": 0.3948,
+      "num_input_tokens_seen": 127477856,
+      "step": 2300,
+      "train_runtime": 13679.4947,
+      "train_tokens_per_second": 9318.901
+    },
+    {
+      "epoch": 0.927938808373591,
+      "grad_norm": 0.7540618643107654,
+      "learning_rate": 5.567900930252375e-06,
+      "loss": 0.4093,
+      "num_input_tokens_seen": 127767664,
+      "step": 2305,
+      "train_runtime": 13717.0826,
+      "train_tokens_per_second": 9314.493
+    },
+    {
+      "epoch": 0.9299516908212561,
+      "grad_norm": 0.7571411874322634,
+      "learning_rate": 5.552191279554071e-06,
+      "loss": 0.4048,
+      "num_input_tokens_seen": 128070808,
+      "step": 2310,
+      "train_runtime": 13745.1253,
+      "train_tokens_per_second": 9317.544
+    },
+    {
+      "epoch": 0.9319645732689211,
+      "grad_norm": 0.6582092832879837,
+      "learning_rate": 5.536476108516598e-06,
+      "loss": 0.3879,
+      "num_input_tokens_seen": 128353544,
+      "step": 2315,
+      "train_runtime": 13775.6743,
+      "train_tokens_per_second": 9317.406
+    },
+    {
+      "epoch": 0.9339774557165862,
+      "grad_norm": 0.7751924697578754,
+      "learning_rate": 5.5207555742468785e-06,
+      "loss": 0.3954,
+      "num_input_tokens_seen": 128641088,
+      "step": 2320,
+      "train_runtime": 13807.1001,
+      "train_tokens_per_second": 9317.024
+    },
+    {
+      "epoch": 0.9359903381642513,
+      "grad_norm": 0.8572280455593408,
+      "learning_rate": 5.505029833905456e-06,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 128926168,
+      "step": 2325,
+      "train_runtime": 13838.9584,
+      "train_tokens_per_second": 9316.176
+    },
+    {
+      "epoch": 0.9380032206119162,
+      "grad_norm": 0.9150846041594187,
+      "learning_rate": 5.4892990447049225e-06,
+      "loss": 0.3998,
+      "num_input_tokens_seen": 129219784,
+      "step": 2330,
+      "train_runtime": 13868.9306,
+      "train_tokens_per_second": 9317.213
+    },
+    {
+      "epoch": 0.9400161030595813,
+      "grad_norm": 0.7159606215745719,
+      "learning_rate": 5.473563363908333e-06,
+      "loss": 0.4001,
+      "num_input_tokens_seen": 129517336,
+      "step": 2335,
+      "train_runtime": 13900.685,
+      "train_tokens_per_second": 9317.335
+    },
+    {
+      "epoch": 0.9420289855072463,
+      "grad_norm": 0.7193685065146267,
+      "learning_rate": 5.457822948827658e-06,
+      "loss": 0.4007,
+      "num_input_tokens_seen": 129810320,
+      "step": 2340,
+      "train_runtime": 13931.625,
+      "train_tokens_per_second": 9317.673
+    },
+    {
+      "epoch": 0.9440418679549114,
+      "grad_norm": 0.7455042583512728,
+      "learning_rate": 5.442077956822191e-06,
+      "loss": 0.3956,
+      "num_input_tokens_seen": 130079760,
+      "step": 2345,
+      "train_runtime": 13963.6821,
+      "train_tokens_per_second": 9315.577
+    },
+    {
+      "epoch": 0.9460547504025765,
+      "grad_norm": 0.7655868059613371,
+      "learning_rate": 5.4263285452969805e-06,
+      "loss": 0.3793,
+      "num_input_tokens_seen": 130345000,
+      "step": 2350,
+      "train_runtime": 13995.5985,
+      "train_tokens_per_second": 9313.285
+    },
+    {
+      "epoch": 0.9480676328502415,
+      "grad_norm": 0.8112268217323834,
+      "learning_rate": 5.410574871701261e-06,
+      "loss": 0.3712,
+      "num_input_tokens_seen": 130614688,
+      "step": 2355,
+      "train_runtime": 14024.0555,
+      "train_tokens_per_second": 9313.617
+    },
+    {
+      "epoch": 0.9500805152979066,
+      "grad_norm": 0.7230830015558871,
+      "learning_rate": 5.394817093526873e-06,
+      "loss": 0.4034,
+      "num_input_tokens_seen": 130891080,
+      "step": 2360,
+      "train_runtime": 14058.0892,
+      "train_tokens_per_second": 9310.73
+    },
+    {
+      "epoch": 0.9520933977455717,
+      "grad_norm": 0.8145047149288794,
+      "learning_rate": 5.379055368306693e-06,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 131169720,
+      "step": 2365,
+      "train_runtime": 14081.7041,
+      "train_tokens_per_second": 9314.904
+    },
+    {
+      "epoch": 0.9541062801932367,
+      "grad_norm": 0.7440279991711057,
+      "learning_rate": 5.363289853613054e-06,
+      "loss": 0.3993,
+      "num_input_tokens_seen": 131464072,
+      "step": 2370,
+      "train_runtime": 14104.6792,
+      "train_tokens_per_second": 9320.6
+    },
+    {
+      "epoch": 0.9561191626409018,
+      "grad_norm": 0.769388688340615,
+      "learning_rate": 5.347520707056175e-06,
+      "loss": 0.4135,
+      "num_input_tokens_seen": 131732456,
+      "step": 2375,
+      "train_runtime": 14136.6076,
+      "train_tokens_per_second": 9318.534
+    },
+    {
+      "epoch": 0.9581320450885669,
+      "grad_norm": 0.7847493681215424,
+      "learning_rate": 5.331748086282584e-06,
+      "loss": 0.3655,
+      "num_input_tokens_seen": 132005232,
+      "step": 2380,
+      "train_runtime": 14163.0502,
+      "train_tokens_per_second": 9320.396
+    },
+    {
+      "epoch": 0.9601449275362319,
+      "grad_norm": 0.7632768515657365,
+      "learning_rate": 5.315972148973541e-06,
+      "loss": 0.4308,
+      "num_input_tokens_seen": 132291704,
+      "step": 2385,
+      "train_runtime": 14196.3319,
+      "train_tokens_per_second": 9318.724
+    },
+    {
+      "epoch": 0.962157809983897,
+      "grad_norm": 0.8016025093700802,
+      "learning_rate": 5.30019305284346e-06,
+      "loss": 0.3818,
+      "num_input_tokens_seen": 132561512,
+      "step": 2390,
+      "train_runtime": 14225.5835,
+      "train_tokens_per_second": 9318.529
+    },
+    {
+      "epoch": 0.964170692431562,
+      "grad_norm": 0.6853666243604544,
+      "learning_rate": 5.284410955638335e-06,
+      "loss": 0.4175,
+      "num_input_tokens_seen": 132840880,
+      "step": 2395,
+      "train_runtime": 14261.4508,
+      "train_tokens_per_second": 9314.682
+    },
+    {
+      "epoch": 0.966183574879227,
+      "grad_norm": 0.641665165324985,
+      "learning_rate": 5.268626015134163e-06,
+      "loss": 0.3679,
+      "num_input_tokens_seen": 133110416,
+      "step": 2400,
+      "train_runtime": 14287.474,
+      "train_tokens_per_second": 9316.581
+    },
+    {
+      "epoch": 0.9681964573268921,
+      "grad_norm": 0.690160582864306,
+      "learning_rate": 5.252838389135364e-06,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 133406792,
+      "step": 2405,
+      "train_runtime": 14318.6614,
+      "train_tokens_per_second": 9316.988
+    },
+    {
+      "epoch": 0.9702093397745571,
+      "grad_norm": 0.7726620712796,
+      "learning_rate": 5.237048235473207e-06,
+      "loss": 0.3964,
+      "num_input_tokens_seen": 133699960,
+      "step": 2410,
+      "train_runtime": 14349.797,
+      "train_tokens_per_second": 9317.202
+    },
+    {
+      "epoch": 0.9722222222222222,
+      "grad_norm": 0.8043653857263554,
+      "learning_rate": 5.221255712004233e-06,
+      "loss": 0.3927,
+      "num_input_tokens_seen": 133944592,
+      "step": 2415,
+      "train_runtime": 14375.3757,
+      "train_tokens_per_second": 9317.641
+    },
+    {
+      "epoch": 0.9742351046698873,
+      "grad_norm": 0.7292760265362234,
+      "learning_rate": 5.205460976608668e-06,
+      "loss": 0.3715,
+      "num_input_tokens_seen": 134236176,
+      "step": 2420,
+      "train_runtime": 14403.6986,
+      "train_tokens_per_second": 9319.563
+    },
+    {
+      "epoch": 0.9762479871175523,
+      "grad_norm": 0.7834060555314656,
+      "learning_rate": 5.189664187188857e-06,
+      "loss": 0.3808,
+      "num_input_tokens_seen": 134534560,
+      "step": 2425,
+      "train_runtime": 14431.4074,
+      "train_tokens_per_second": 9322.345
+    },
+    {
+      "epoch": 0.9782608695652174,
+      "grad_norm": 0.7526189852580767,
+      "learning_rate": 5.173865501667675e-06,
+      "loss": 0.3986,
+      "num_input_tokens_seen": 134799696,
+      "step": 2430,
+      "train_runtime": 14461.3195,
+      "train_tokens_per_second": 9321.397
+    },
+    {
+      "epoch": 0.9802737520128825,
+      "grad_norm": 0.7325874182041106,
+      "learning_rate": 5.158065077986956e-06,
+      "loss": 0.3952,
+      "num_input_tokens_seen": 135070128,
+      "step": 2435,
+      "train_runtime": 14486.1343,
+      "train_tokens_per_second": 9324.097
+    },
+    {
+      "epoch": 0.9822866344605475,
+      "grad_norm": 0.8297597418673935,
+      "learning_rate": 5.1422630741059075e-06,
+      "loss": 0.3916,
+      "num_input_tokens_seen": 135356280,
+      "step": 2440,
+      "train_runtime": 14520.857,
+      "train_tokens_per_second": 9321.508
+    },
+    {
+      "epoch": 0.9842995169082126,
+      "grad_norm": 0.6636715979011287,
+      "learning_rate": 5.126459647999535e-06,
+      "loss": 0.3911,
+      "num_input_tokens_seen": 135652664,
+      "step": 2445,
+      "train_runtime": 14551.1127,
+      "train_tokens_per_second": 9322.494
+    },
+    {
+      "epoch": 0.9863123993558777,
+      "grad_norm": 0.8156228369759273,
+      "learning_rate": 5.110654957657067e-06,
+      "loss": 0.4172,
+      "num_input_tokens_seen": 135930152,
+      "step": 2450,
+      "train_runtime": 14576.7269,
+      "train_tokens_per_second": 9325.149
+    },
+    {
+      "epoch": 0.9883252818035426,
+      "grad_norm": 0.709320797763845,
+      "learning_rate": 5.094849161080366e-06,
+      "loss": 0.389,
+      "num_input_tokens_seen": 136235360,
+      "step": 2455,
+      "train_runtime": 14601.2021,
+      "train_tokens_per_second": 9330.421
+    },
+    {
+      "epoch": 0.9903381642512077,
+      "grad_norm": 0.7907954734279484,
+      "learning_rate": 5.079042416282351e-06,
+      "loss": 0.4016,
+      "num_input_tokens_seen": 136501672,
+      "step": 2460,
+      "train_runtime": 14627.744,
+      "train_tokens_per_second": 9331.697
+    },
+    {
+      "epoch": 0.9923510466988728,
+      "grad_norm": 0.7382866900932901,
+      "learning_rate": 5.063234881285428e-06,
+      "loss": 0.3992,
+      "num_input_tokens_seen": 136798896,
+      "step": 2465,
+      "train_runtime": 14651.3528,
+      "train_tokens_per_second": 9336.946
+    },
+    {
+      "epoch": 0.9943639291465378,
+      "grad_norm": 0.7578591719070971,
+      "learning_rate": 5.047426714119897e-06,
+      "loss": 0.4029,
+      "num_input_tokens_seen": 137054328,
+      "step": 2470,
+      "train_runtime": 14673.8644,
+      "train_tokens_per_second": 9340.03
+    },
+    {
+      "epoch": 0.9963768115942029,
+      "grad_norm": 0.771734358475569,
+      "learning_rate": 5.0316180728223815e-06,
+      "loss": 0.3606,
+      "num_input_tokens_seen": 137342352,
+      "step": 2475,
+      "train_runtime": 14709.5613,
+      "train_tokens_per_second": 9336.944
+    },
+    {
+      "epoch": 0.998389694041868,
+      "grad_norm": 0.6780559159351053,
+      "learning_rate": 5.015809115434243e-06,
+      "loss": 0.3912,
+      "num_input_tokens_seen": 137622184,
+      "step": 2480,
+      "train_runtime": 14738.4306,
+      "train_tokens_per_second": 9337.642
+    },
+    {
+      "epoch": 1.000402576489533,
+      "grad_norm": 0.7243239434549581,
+      "learning_rate": 5e-06,
+      "loss": 0.3799,
+      "num_input_tokens_seen": 137898728,
+      "step": 2485,
+      "train_runtime": 14772.1169,
+      "train_tokens_per_second": 9335.069
+    },
+    {
+      "epoch": 1.002415458937198,
+      "grad_norm": 0.7098230223240772,
+      "learning_rate": 4.984190884565759e-06,
+      "loss": 0.3449,
+      "num_input_tokens_seen": 138185760,
+      "step": 2490,
+      "train_runtime": 14799.7595,
+      "train_tokens_per_second": 9337.027
+    },
+    {
+      "epoch": 1.0044283413848631,
+      "grad_norm": 0.7340641188132235,
+      "learning_rate": 4.96838192717762e-06,
+      "loss": 0.3613,
+      "num_input_tokens_seen": 138471720,
+      "step": 2495,
+      "train_runtime": 14830.2074,
+      "train_tokens_per_second": 9337.14
+    },
+    {
+      "epoch": 1.0064412238325282,
+      "grad_norm": 0.7395237621470443,
+      "learning_rate": 4.952573285880104e-06,
+      "loss": 0.3402,
+      "num_input_tokens_seen": 138734928,
+      "step": 2500,
+      "train_runtime": 14857.2749,
+      "train_tokens_per_second": 9337.845
+    },
+    {
+      "epoch": 1.0084541062801933,
+      "grad_norm": 0.749921849606483,
+      "learning_rate": 4.936765118714574e-06,
+      "loss": 0.3342,
+      "num_input_tokens_seen": 139021744,
+      "step": 2505,
+      "train_runtime": 14886.7496,
+      "train_tokens_per_second": 9338.623
+    },
+    {
+      "epoch": 1.0104669887278583,
+      "grad_norm": 0.7403069807136297,
+      "learning_rate": 4.9209575837176495e-06,
+      "loss": 0.3403,
+      "num_input_tokens_seen": 139265688,
+      "step": 2510,
+      "train_runtime": 14910.1625,
+      "train_tokens_per_second": 9340.32
+    },
+    {
+      "epoch": 1.0124798711755234,
+      "grad_norm": 0.8058981249171341,
+      "learning_rate": 4.905150838919635e-06,
+      "loss": 0.3671,
+      "num_input_tokens_seen": 139551768,
+      "step": 2515,
+      "train_runtime": 14938.5661,
+      "train_tokens_per_second": 9341.711
+    },
+    {
+      "epoch": 1.0144927536231885,
+      "grad_norm": 0.7585626177738353,
+      "learning_rate": 4.889345042342934e-06,
+      "loss": 0.3411,
+      "num_input_tokens_seen": 139831936,
+      "step": 2520,
+      "train_runtime": 14964.3183,
+      "train_tokens_per_second": 9344.357
+    },
+    {
+      "epoch": 1.0165056360708535,
+      "grad_norm": 0.6203505711488339,
+      "learning_rate": 4.873540352000464e-06,
+      "loss": 0.326,
+      "num_input_tokens_seen": 140130504,
+      "step": 2525,
+      "train_runtime": 14997.402,
+      "train_tokens_per_second": 9343.652
+    },
+    {
+      "epoch": 1.0185185185185186,
+      "grad_norm": 0.8300699910582235,
+      "learning_rate": 4.857736925894093e-06,
+      "loss": 0.3333,
+      "num_input_tokens_seen": 140386784,
+      "step": 2530,
+      "train_runtime": 15030.2297,
+      "train_tokens_per_second": 9340.295
+    },
+    {
+      "epoch": 1.0205314009661837,
+      "grad_norm": 0.7084117723918122,
+      "learning_rate": 4.841934922013045e-06,
+      "loss": 0.3316,
+      "num_input_tokens_seen": 140694184,
+      "step": 2535,
+      "train_runtime": 15055.6046,
+      "train_tokens_per_second": 9344.971
+    },
+    {
+      "epoch": 1.0225442834138487,
+      "grad_norm": 0.8145857514386612,
+      "learning_rate": 4.826134498332325e-06,
+      "loss": 0.3276,
+      "num_input_tokens_seen": 140932536,
+      "step": 2540,
+      "train_runtime": 15083.2902,
+      "train_tokens_per_second": 9343.62
+    },
+    {
+      "epoch": 1.0245571658615138,
+      "grad_norm": 0.8918613651030668,
+      "learning_rate": 4.8103358128111435e-06,
+      "loss": 0.3419,
+      "num_input_tokens_seen": 141181184,
+      "step": 2545,
+      "train_runtime": 15107.675,
+      "train_tokens_per_second": 9344.997
+    },
+    {
+      "epoch": 1.0265700483091786,
+      "grad_norm": 0.8736739082706182,
+      "learning_rate": 4.7945390233913345e-06,
+      "loss": 0.3366,
+      "num_input_tokens_seen": 141459424,
+      "step": 2550,
+      "train_runtime": 15133.3016,
+      "train_tokens_per_second": 9347.559
+    },
+    {
+      "epoch": 1.0285829307568437,
+      "grad_norm": 0.8401776229004124,
+      "learning_rate": 4.778744287995769e-06,
+      "loss": 0.3385,
+      "num_input_tokens_seen": 141737336,
+      "step": 2555,
+      "train_runtime": 15162.4982,
+      "train_tokens_per_second": 9347.888
+    },
+    {
+      "epoch": 1.0305958132045088,
+      "grad_norm": 0.7195894895978254,
+      "learning_rate": 4.762951764526794e-06,
+      "loss": 0.3445,
+      "num_input_tokens_seen": 142018120,
+      "step": 2560,
+      "train_runtime": 15191.1122,
+      "train_tokens_per_second": 9348.764
+    },
+    {
+      "epoch": 1.0326086956521738,
+      "grad_norm": 0.7307630347512138,
+      "learning_rate": 4.747161610864639e-06,
+      "loss": 0.3273,
+      "num_input_tokens_seen": 142287152,
+      "step": 2565,
+      "train_runtime": 15216.9461,
+      "train_tokens_per_second": 9350.572
+    },
+    {
+      "epoch": 1.034621578099839,
+      "grad_norm": 0.7345381304042853,
+      "learning_rate": 4.73137398486584e-06,
+      "loss": 0.3421,
+      "num_input_tokens_seen": 142568088,
+      "step": 2570,
+      "train_runtime": 15245.7839,
+      "train_tokens_per_second": 9351.312
+    },
+    {
+      "epoch": 1.036634460547504,
+      "grad_norm": 0.7028635588655188,
+      "learning_rate": 4.715589044361667e-06,
+      "loss": 0.3474,
+      "num_input_tokens_seen": 142838040,
+      "step": 2575,
+      "train_runtime": 15275.5171,
+      "train_tokens_per_second": 9350.783
+    },
+    {
+      "epoch": 1.038647342995169,
+      "grad_norm": 0.7650230908850711,
+      "learning_rate": 4.699806947156542e-06,
+      "loss": 0.3458,
+      "num_input_tokens_seen": 143123216,
+      "step": 2580,
+      "train_runtime": 15304.6563,
+      "train_tokens_per_second": 9351.613
+    },
+    {
+      "epoch": 1.040660225442834,
+      "grad_norm": 0.7479819978817114,
+      "learning_rate": 4.6840278510264595e-06,
+      "loss": 0.3652,
+      "num_input_tokens_seen": 143393424,
+      "step": 2585,
+      "train_runtime": 15332.5801,
+      "train_tokens_per_second": 9352.204
+    },
+    {
+      "epoch": 1.0426731078904992,
+      "grad_norm": 0.7008479183127246,
+      "learning_rate": 4.668251913717417e-06,
+      "loss": 0.3167,
+      "num_input_tokens_seen": 143652632,
+      "step": 2590,
+      "train_runtime": 15364.5823,
+      "train_tokens_per_second": 9349.596
+    },
+    {
+      "epoch": 1.0446859903381642,
+      "grad_norm": 0.6871744152580208,
+      "learning_rate": 4.652479292943826e-06,
+      "loss": 0.3408,
+      "num_input_tokens_seen": 143888208,
+      "step": 2595,
+      "train_runtime": 15395.4105,
+      "train_tokens_per_second": 9346.175
+    },
+    {
+      "epoch": 1.0466988727858293,
+      "grad_norm": 0.8296477414613053,
+      "learning_rate": 4.636710146386948e-06,
+      "loss": 0.3552,
+      "num_input_tokens_seen": 144150200,
+      "step": 2600,
+      "train_runtime": 15424.2999,
+      "train_tokens_per_second": 9345.656
+    },
+    {
+      "epoch": 1.0487117552334944,
+      "grad_norm": 0.7832823623856088,
+      "learning_rate": 4.620944631693309e-06,
+      "loss": 0.3389,
+      "num_input_tokens_seen": 144447672,
+      "step": 2605,
+      "train_runtime": 15453.905,
+      "train_tokens_per_second": 9347.001
+    },
+    {
+      "epoch": 1.0507246376811594,
+      "grad_norm": 0.8293799666537527,
+      "learning_rate": 4.605182906473128e-06,
+      "loss": 0.3368,
+      "num_input_tokens_seen": 144722136,
+      "step": 2610,
+      "train_runtime": 15482.4665,
+      "train_tokens_per_second": 9347.486
+    },
+    {
+      "epoch": 1.0527375201288245,
+      "grad_norm": 0.6905321241298036,
+      "learning_rate": 4.589425128298741e-06,
+      "loss": 0.3323,
+      "num_input_tokens_seen": 145013032,
+      "step": 2615,
+      "train_runtime": 15507.1834,
+      "train_tokens_per_second": 9351.346
+    },
+    {
+      "epoch": 1.0547504025764896,
+      "grad_norm": 0.8469692354433622,
+      "learning_rate": 4.57367145470302e-06,
+      "loss": 0.3327,
+      "num_input_tokens_seen": 145285128,
+      "step": 2620,
+      "train_runtime": 15539.7374,
+      "train_tokens_per_second": 9349.265
+    },
+    {
+      "epoch": 1.0567632850241546,
+      "grad_norm": 0.9018270625761752,
+      "learning_rate": 4.55792204317781e-06,
+      "loss": 0.3425,
+      "num_input_tokens_seen": 145568072,
+      "step": 2625,
+      "train_runtime": 15565.3908,
+      "train_tokens_per_second": 9352.035
+    },
+    {
+      "epoch": 1.0587761674718197,
+      "grad_norm": 0.7453021872864758,
+      "learning_rate": 4.5421770511723434e-06,
+      "loss": 0.3339,
+      "num_input_tokens_seen": 145837472,
+      "step": 2630,
+      "train_runtime": 15600.2438,
+      "train_tokens_per_second": 9348.41
+    },
+    {
+      "epoch": 1.0607890499194848,
+      "grad_norm": 0.7611448483350662,
+      "learning_rate": 4.5264366360916675e-06,
+      "loss": 0.3356,
+      "num_input_tokens_seen": 146128112,
+      "step": 2635,
+      "train_runtime": 15629.9638,
+      "train_tokens_per_second": 9349.229
+    },
+    {
+      "epoch": 1.0628019323671498,
+      "grad_norm": 0.7553438321706385,
+      "learning_rate": 4.51070095529508e-06,
+      "loss": 0.3404,
+      "num_input_tokens_seen": 146416416,
+      "step": 2640,
+      "train_runtime": 15666.7794,
+      "train_tokens_per_second": 9345.661
+    },
+    {
+      "epoch": 1.0648148148148149,
+      "grad_norm": 0.7298717537886026,
+      "learning_rate": 4.494970166094545e-06,
+      "loss": 0.3486,
+      "num_input_tokens_seen": 146697176,
+      "step": 2645,
+      "train_runtime": 15696.5993,
+      "train_tokens_per_second": 9345.794
+    },
+    {
+      "epoch": 1.06682769726248,
+      "grad_norm": 0.7174222447507045,
+      "learning_rate": 4.4792444257531215e-06,
+      "loss": 0.3443,
+      "num_input_tokens_seen": 146967184,
+      "step": 2650,
+      "train_runtime": 15717.4601,
+      "train_tokens_per_second": 9350.568
+    },
+    {
+      "epoch": 1.068840579710145,
+      "grad_norm": 0.7046934326285237,
+      "learning_rate": 4.4635238914834035e-06,
+      "loss": 0.3269,
+      "num_input_tokens_seen": 147245352,
+      "step": 2655,
+      "train_runtime": 15743.4171,
+      "train_tokens_per_second": 9352.82
+    },
+    {
+      "epoch": 1.07085346215781,
+      "grad_norm": 0.8055045816879963,
+      "learning_rate": 4.447808720445928e-06,
+      "loss": 0.343,
+      "num_input_tokens_seen": 147527912,
+      "step": 2660,
+      "train_runtime": 15774.4036,
+      "train_tokens_per_second": 9352.361
+    },
+    {
+      "epoch": 1.0728663446054751,
+      "grad_norm": 0.7958337133937959,
+      "learning_rate": 4.432099069747625e-06,
+      "loss": 0.3454,
+      "num_input_tokens_seen": 147805376,
+      "step": 2665,
+      "train_runtime": 15797.9431,
+      "train_tokens_per_second": 9355.989
+    },
+    {
+      "epoch": 1.07487922705314,
+      "grad_norm": 0.728393068834379,
+      "learning_rate": 4.416395096440232e-06,
+      "loss": 0.3414,
+      "num_input_tokens_seen": 148096704,
+      "step": 2670,
+      "train_runtime": 15824.7861,
+      "train_tokens_per_second": 9358.528
+    },
+    {
+      "epoch": 1.076892109500805,
+      "grad_norm": 0.7124620362127702,
+      "learning_rate": 4.400696957518729e-06,
+      "loss": 0.3226,
+      "num_input_tokens_seen": 148374240,
+      "step": 2675,
+      "train_runtime": 15854.1039,
+      "train_tokens_per_second": 9358.728
+    },
+    {
+      "epoch": 1.0789049919484701,
+      "grad_norm": 0.7824789936295722,
+      "learning_rate": 4.3850048099197655e-06,
+      "loss": 0.3456,
+      "num_input_tokens_seen": 148636264,
+      "step": 2680,
+      "train_runtime": 15881.1548,
+      "train_tokens_per_second": 9359.286
+    },
+    {
+      "epoch": 1.0809178743961352,
+      "grad_norm": 0.9546274823670803,
+      "learning_rate": 4.369318810520101e-06,
+      "loss": 0.3499,
+      "num_input_tokens_seen": 148904160,
+      "step": 2685,
+      "train_runtime": 15916.6142,
+      "train_tokens_per_second": 9355.266
+    },
+    {
+      "epoch": 1.0829307568438002,
+      "grad_norm": 0.8214590231595719,
+      "learning_rate": 4.353639116135027e-06,
+      "loss": 0.3499,
+      "num_input_tokens_seen": 149184824,
+      "step": 2690,
+      "train_runtime": 15946.8861,
+      "train_tokens_per_second": 9355.107
+    },
+    {
+      "epoch": 1.0849436392914653,
+      "grad_norm": 0.6960288428590676,
+      "learning_rate": 4.337965883516805e-06,
+      "loss": 0.3267,
+      "num_input_tokens_seen": 149456000,
+      "step": 2695,
+      "train_runtime": 15979.9953,
+      "train_tokens_per_second": 9352.694
+    },
+    {
+      "epoch": 1.0869565217391304,
+      "grad_norm": 0.7348899856697629,
+      "learning_rate": 4.322299269353095e-06,
+      "loss": 0.3325,
+      "num_input_tokens_seen": 149745208,
+      "step": 2700,
+      "train_runtime": 16012.3101,
+      "train_tokens_per_second": 9351.88
+    },
+    {
+      "epoch": 1.0889694041867954,
+      "grad_norm": 0.7159345301360588,
+      "learning_rate": 4.306639430265392e-06,
+      "loss": 0.3446,
+      "num_input_tokens_seen": 150022800,
+      "step": 2705,
+      "train_runtime": 16044.4857,
+      "train_tokens_per_second": 9350.427
+    },
+    {
+      "epoch": 1.0909822866344605,
+      "grad_norm": 0.6999867262753464,
+      "learning_rate": 4.29098652280746e-06,
+      "loss": 0.3413,
+      "num_input_tokens_seen": 150310680,
+      "step": 2710,
+      "train_runtime": 16080.9578,
+      "train_tokens_per_second": 9347.122
+    },
+    {
+      "epoch": 1.0929951690821256,
+      "grad_norm": 0.8011352760770298,
+      "learning_rate": 4.275340703463767e-06,
+      "loss": 0.3557,
+      "num_input_tokens_seen": 150592344,
+      "step": 2715,
+      "train_runtime": 16105.1877,
+      "train_tokens_per_second": 9350.549
+    },
+    {
+      "epoch": 1.0950080515297906,
+      "grad_norm": 0.7651211371480515,
+      "learning_rate": 4.259702128647918e-06,
+      "loss": 0.3304,
+      "num_input_tokens_seen": 150894872,
+      "step": 2720,
+      "train_runtime": 16139.293,
+      "train_tokens_per_second": 9349.534
+    },
+    {
+      "epoch": 1.0970209339774557,
+      "grad_norm": 0.7488461317524117,
+      "learning_rate": 4.244070954701096e-06,
+      "loss": 0.322,
+      "num_input_tokens_seen": 151178168,
+      "step": 2725,
+      "train_runtime": 16171.7074,
+      "train_tokens_per_second": 9348.312
+    },
+    {
+      "epoch": 1.0990338164251208,
+      "grad_norm": 0.7517975853136228,
+      "learning_rate": 4.2284473378904935e-06,
+      "loss": 0.3399,
+      "num_input_tokens_seen": 151470512,
+      "step": 2730,
+      "train_runtime": 16202.3944,
+      "train_tokens_per_second": 9348.65
+    },
+    {
+      "epoch": 1.1010466988727858,
+      "grad_norm": 0.6903648605822994,
+      "learning_rate": 4.212831434407755e-06,
+      "loss": 0.3418,
+      "num_input_tokens_seen": 151770968,
+      "step": 2735,
+      "train_runtime": 16227.6059,
+      "train_tokens_per_second": 9352.641
+    },
+    {
+      "epoch": 1.103059581320451,
+      "grad_norm": 0.7001218019941555,
+      "learning_rate": 4.197223400367415e-06,
+      "loss": 0.3351,
+      "num_input_tokens_seen": 152063832,
+      "step": 2740,
+      "train_runtime": 16263.5499,
+      "train_tokens_per_second": 9349.978
+    },
+    {
+      "epoch": 1.105072463768116,
+      "grad_norm": 0.7796939128256034,
+      "learning_rate": 4.1816233918053316e-06,
+      "loss": 0.3191,
+      "num_input_tokens_seen": 152305168,
+      "step": 2745,
+      "train_runtime": 16283.3383,
+      "train_tokens_per_second": 9353.436
+    },
+    {
+      "epoch": 1.107085346215781,
+      "grad_norm": 0.7412542135482324,
+      "learning_rate": 4.166031564677138e-06,
+      "loss": 0.3345,
+      "num_input_tokens_seen": 152617232,
+      "step": 2750,
+      "train_runtime": 16306.4992,
+      "train_tokens_per_second": 9359.289
+    },
+    {
+      "epoch": 1.109098228663446,
+      "grad_norm": 0.6644411988813269,
+      "learning_rate": 4.150448074856667e-06,
+      "loss": 0.3327,
+      "num_input_tokens_seen": 152921712,
+      "step": 2755,
+      "train_runtime": 16339.8748,
+      "train_tokens_per_second": 9358.806
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.8709944891016662,
+      "learning_rate": 4.1348730781344105e-06,
+      "loss": 0.3453,
+      "num_input_tokens_seen": 153215800,
+      "step": 2760,
+      "train_runtime": 16371.0277,
+      "train_tokens_per_second": 9358.96
+    },
+    {
+      "epoch": 1.1131239935587762,
+      "grad_norm": 0.6908996264490945,
+      "learning_rate": 4.119306730215951e-06,
+      "loss": 0.3458,
+      "num_input_tokens_seen": 153485136,
+      "step": 2765,
+      "train_runtime": 16402.6706,
+      "train_tokens_per_second": 9357.326
+    },
+    {
+      "epoch": 1.1151368760064413,
+      "grad_norm": 0.8166064151006408,
+      "learning_rate": 4.103749186720405e-06,
+      "loss": 0.3309,
+      "num_input_tokens_seen": 153766608,
+      "step": 2770,
+      "train_runtime": 16438.4293,
+      "train_tokens_per_second": 9354.094
+    },
+    {
+      "epoch": 1.1171497584541064,
+      "grad_norm": 0.7669142007962767,
+      "learning_rate": 4.088200603178872e-06,
+      "loss": 0.3214,
+      "num_input_tokens_seen": 154039072,
+      "step": 2775,
+      "train_runtime": 16468.8793,
+      "train_tokens_per_second": 9353.343
+    },
+    {
+      "epoch": 1.1191626409017714,
+      "grad_norm": 0.8383957043084173,
+      "learning_rate": 4.0726611350328786e-06,
+      "loss": 0.3459,
+      "num_input_tokens_seen": 154302088,
+      "step": 2780,
+      "train_runtime": 16499.8879,
+      "train_tokens_per_second": 9351.705
+    },
+    {
+      "epoch": 1.1211755233494365,
+      "grad_norm": 0.8082993057747515,
+      "learning_rate": 4.057130937632821e-06,
+      "loss": 0.3449,
+      "num_input_tokens_seen": 154586608,
+      "step": 2785,
+      "train_runtime": 16531.3901,
+      "train_tokens_per_second": 9351.096
+    },
+    {
+      "epoch": 1.1231884057971016,
+      "grad_norm": 0.8595591367756767,
+      "learning_rate": 4.041610166236417e-06,
+      "loss": 0.3332,
+      "num_input_tokens_seen": 154849256,
+      "step": 2790,
+      "train_runtime": 16556.4585,
+      "train_tokens_per_second": 9352.801
+    },
+    {
+      "epoch": 1.1252012882447664,
+      "grad_norm": 0.7344879967186686,
+      "learning_rate": 4.026098976007148e-06,
+      "loss": 0.3199,
+      "num_input_tokens_seen": 155156384,
+      "step": 2795,
+      "train_runtime": 16585.4821,
+      "train_tokens_per_second": 9354.952
+    },
+    {
+      "epoch": 1.1272141706924317,
+      "grad_norm": 0.8441235262328253,
+      "learning_rate": 4.010597522012716e-06,
+      "loss": 0.3312,
+      "num_input_tokens_seen": 155461544,
+      "step": 2800,
+      "train_runtime": 16609.5746,
+      "train_tokens_per_second": 9359.755
+    },
+    {
+      "epoch": 1.1292270531400965,
+      "grad_norm": 0.7480706843370859,
+      "learning_rate": 3.9951059592234805e-06,
+      "loss": 0.3394,
+      "num_input_tokens_seen": 155761992,
+      "step": 2805,
+      "train_runtime": 16643.5684,
+      "train_tokens_per_second": 9358.69
+    },
+    {
+      "epoch": 1.1312399355877616,
+      "grad_norm": 0.7469545061052811,
+      "learning_rate": 3.979624442510925e-06,
+      "loss": 0.3567,
+      "num_input_tokens_seen": 156042840,
+      "step": 2810,
+      "train_runtime": 16674.7273,
+      "train_tokens_per_second": 9358.045
+    },
+    {
+      "epoch": 1.1332528180354267,
+      "grad_norm": 0.7075261916927754,
+      "learning_rate": 3.964153126646095e-06,
+      "loss": 0.3225,
+      "num_input_tokens_seen": 156346080,
+      "step": 2815,
+      "train_runtime": 16703.7557,
+      "train_tokens_per_second": 9359.936
+    },
+    {
+      "epoch": 1.1352657004830917,
+      "grad_norm": 0.7276903601759177,
+      "learning_rate": 3.948692166298062e-06,
+      "loss": 0.3238,
+      "num_input_tokens_seen": 156625328,
+      "step": 2820,
+      "train_runtime": 16731.2667,
+      "train_tokens_per_second": 9361.236
+    },
+    {
+      "epoch": 1.1372785829307568,
+      "grad_norm": 0.8128801163294187,
+      "learning_rate": 3.9332417160323685e-06,
+      "loss": 0.3277,
+      "num_input_tokens_seen": 156911664,
+      "step": 2825,
+      "train_runtime": 16758.7837,
+      "train_tokens_per_second": 9362.951
+    },
+    {
+      "epoch": 1.1392914653784219,
+      "grad_norm": 0.8174635590359891,
+      "learning_rate": 3.917801930309486e-06,
+      "loss": 0.3357,
+      "num_input_tokens_seen": 157174264,
+      "step": 2830,
+      "train_runtime": 16785.1431,
+      "train_tokens_per_second": 9363.892
+    },
+    {
+      "epoch": 1.141304347826087,
+      "grad_norm": 0.7374636293364648,
+      "learning_rate": 3.902372963483273e-06,
+      "loss": 0.358,
+      "num_input_tokens_seen": 157451416,
+      "step": 2835,
+      "train_runtime": 16816.6722,
+      "train_tokens_per_second": 9362.817
+    },
+    {
+      "epoch": 1.143317230273752,
+      "grad_norm": 0.7267576407913852,
+      "learning_rate": 3.886954969799427e-06,
+      "loss": 0.3384,
+      "num_input_tokens_seen": 157743512,
+      "step": 2840,
+      "train_runtime": 16852.0007,
+      "train_tokens_per_second": 9360.521
+    },
+    {
+      "epoch": 1.145330112721417,
+      "grad_norm": 0.7485112735217255,
+      "learning_rate": 3.871548103393947e-06,
+      "loss": 0.3269,
+      "num_input_tokens_seen": 158007856,
+      "step": 2845,
+      "train_runtime": 16877.8937,
+      "train_tokens_per_second": 9361.823
+    },
+    {
+      "epoch": 1.1473429951690821,
+      "grad_norm": 0.8528938941319912,
+      "learning_rate": 3.856152518291593e-06,
+      "loss": 0.3431,
+      "num_input_tokens_seen": 158276856,
+      "step": 2850,
+      "train_runtime": 16905.97,
+      "train_tokens_per_second": 9362.187
+    },
+    {
+      "epoch": 1.1493558776167472,
+      "grad_norm": 0.7427854040856345,
+      "learning_rate": 3.84076836840434e-06,
+      "loss": 0.3334,
+      "num_input_tokens_seen": 158558152,
+      "step": 2855,
+      "train_runtime": 16937.2755,
+      "train_tokens_per_second": 9361.491
+    },
+    {
+      "epoch": 1.1513687600644122,
+      "grad_norm": 0.7648667041662149,
+      "learning_rate": 3.825395807529845e-06,
+      "loss": 0.3424,
+      "num_input_tokens_seen": 158838168,
+      "step": 2860,
+      "train_runtime": 16965.676,
+      "train_tokens_per_second": 9362.325
+    },
+    {
+      "epoch": 1.1533816425120773,
+      "grad_norm": 0.7265847958677756,
+      "learning_rate": 3.8100349893499094e-06,
+      "loss": 0.3199,
+      "num_input_tokens_seen": 159088800,
+      "step": 2865,
+      "train_runtime": 16989.3193,
+      "train_tokens_per_second": 9364.048
+    },
+    {
+      "epoch": 1.1553945249597424,
+      "grad_norm": 0.7297815570885989,
+      "learning_rate": 3.79468606742894e-06,
+      "loss": 0.3347,
+      "num_input_tokens_seen": 159374112,
+      "step": 2870,
+      "train_runtime": 17025.0705,
+      "train_tokens_per_second": 9361.143
+    },
+    {
+      "epoch": 1.1574074074074074,
+      "grad_norm": 0.7634669993060619,
+      "learning_rate": 3.779349195212415e-06,
+      "loss": 0.3402,
+      "num_input_tokens_seen": 159639960,
+      "step": 2875,
+      "train_runtime": 17057.2367,
+      "train_tokens_per_second": 9359.075
+    },
+    {
+      "epoch": 1.1594202898550725,
+      "grad_norm": 0.7984823700419286,
+      "learning_rate": 3.7640245260253484e-06,
+      "loss": 0.3638,
+      "num_input_tokens_seen": 159917080,
+      "step": 2880,
+      "train_runtime": 17081.2046,
+      "train_tokens_per_second": 9362.166
+    },
+    {
+      "epoch": 1.1614331723027376,
+      "grad_norm": 0.7418637971072911,
+      "learning_rate": 3.748712213070762e-06,
+      "loss": 0.331,
+      "num_input_tokens_seen": 160178384,
+      "step": 2885,
+      "train_runtime": 17112.7424,
+      "train_tokens_per_second": 9360.182
+    },
+    {
+      "epoch": 1.1634460547504026,
+      "grad_norm": 0.7745596910699212,
+      "learning_rate": 3.733412409428148e-06,
+      "loss": 0.3365,
+      "num_input_tokens_seen": 160459200,
+      "step": 2890,
+      "train_runtime": 17146.7808,
+      "train_tokens_per_second": 9357.978
+    },
+    {
+      "epoch": 1.1654589371980677,
+      "grad_norm": 0.7787015694327291,
+      "learning_rate": 3.7181252680519407e-06,
+      "loss": 0.3259,
+      "num_input_tokens_seen": 160734272,
+      "step": 2895,
+      "train_runtime": 17169.9111,
+      "train_tokens_per_second": 9361.392
+    },
+    {
+      "epoch": 1.1674718196457328,
+      "grad_norm": 0.8393423930767941,
+      "learning_rate": 3.7028509417699915e-06,
+      "loss": 0.3405,
+      "num_input_tokens_seen": 161023640,
+      "step": 2900,
+      "train_runtime": 17198.0475,
+      "train_tokens_per_second": 9362.902
+    },
+    {
+      "epoch": 1.1694847020933978,
+      "grad_norm": 0.8403102908455301,
+      "learning_rate": 3.687589583282031e-06,
+      "loss": 0.3516,
+      "num_input_tokens_seen": 161309400,
+      "step": 2905,
+      "train_runtime": 17224.481,
+      "train_tokens_per_second": 9365.124
+    },
+    {
+      "epoch": 1.171497584541063,
+      "grad_norm": 0.7300955566223247,
+      "learning_rate": 3.6723413451581554e-06,
+      "loss": 0.3309,
+      "num_input_tokens_seen": 161594240,
+      "step": 2910,
+      "train_runtime": 17258.3139,
+      "train_tokens_per_second": 9363.269
+    },
+    {
+      "epoch": 1.173510466988728,
+      "grad_norm": 0.7360500987290836,
+      "learning_rate": 3.6571063798372913e-06,
+      "loss": 0.3388,
+      "num_input_tokens_seen": 161877344,
+      "step": 2915,
+      "train_runtime": 17288.3831,
+      "train_tokens_per_second": 9363.359
+    },
+    {
+      "epoch": 1.1755233494363928,
+      "grad_norm": 0.7782333023344933,
+      "learning_rate": 3.641884839625676e-06,
+      "loss": 0.3528,
+      "num_input_tokens_seen": 162164808,
+      "step": 2920,
+      "train_runtime": 17318.4782,
+      "train_tokens_per_second": 9363.687
+    },
+    {
+      "epoch": 1.177536231884058,
+      "grad_norm": 0.7702460245125943,
+      "learning_rate": 3.626676876695331e-06,
+      "loss": 0.3336,
+      "num_input_tokens_seen": 162457120,
+      "step": 2925,
+      "train_runtime": 17350.4638,
+      "train_tokens_per_second": 9363.272
+    },
+    {
+      "epoch": 1.179549114331723,
+      "grad_norm": 0.6791897147886135,
+      "learning_rate": 3.611482643082548e-06,
+      "loss": 0.328,
+      "num_input_tokens_seen": 162757376,
+      "step": 2930,
+      "train_runtime": 17382.5609,
+      "train_tokens_per_second": 9363.256
+    },
+    {
+      "epoch": 1.181561996779388,
+      "grad_norm": 0.6990507407015166,
+      "learning_rate": 3.5963022906863633e-06,
+      "loss": 0.344,
+      "num_input_tokens_seen": 163057128,
+      "step": 2935,
+      "train_runtime": 17414.4414,
+      "train_tokens_per_second": 9363.328
+    },
+    {
+      "epoch": 1.183574879227053,
+      "grad_norm": 0.7479941255486404,
+      "learning_rate": 3.581135971267039e-06,
+      "loss": 0.3303,
+      "num_input_tokens_seen": 163332224,
+      "step": 2940,
+      "train_runtime": 17446.3665,
+      "train_tokens_per_second": 9361.962
+    },
+    {
+      "epoch": 1.1855877616747181,
+      "grad_norm": 0.7142130787203327,
+      "learning_rate": 3.5659838364445505e-06,
+      "loss": 0.3341,
+      "num_input_tokens_seen": 163630112,
+      "step": 2945,
+      "train_runtime": 17476.7825,
+      "train_tokens_per_second": 9362.714
+    },
+    {
+      "epoch": 1.1876006441223832,
+      "grad_norm": 0.852604387772866,
+      "learning_rate": 3.550846037697063e-06,
+      "loss": 0.334,
+      "num_input_tokens_seen": 163899888,
+      "step": 2950,
+      "train_runtime": 17506.0631,
+      "train_tokens_per_second": 9362.464
+    },
+    {
+      "epoch": 1.1896135265700483,
+      "grad_norm": 0.7559064288714968,
+      "learning_rate": 3.535722726359426e-06,
+      "loss": 0.3316,
+      "num_input_tokens_seen": 164164552,
+      "step": 2955,
+      "train_runtime": 17538.1881,
+      "train_tokens_per_second": 9360.405
+    },
+    {
+      "epoch": 1.1916264090177133,
+      "grad_norm": 0.7857836373605884,
+      "learning_rate": 3.520614053621656e-06,
+      "loss": 0.3326,
+      "num_input_tokens_seen": 164453912,
+      "step": 2960,
+      "train_runtime": 17567.2506,
+      "train_tokens_per_second": 9361.392
+    },
+    {
+      "epoch": 1.1936392914653784,
+      "grad_norm": 0.645775167819383,
+      "learning_rate": 3.5055201705274223e-06,
+      "loss": 0.345,
+      "num_input_tokens_seen": 164734248,
+      "step": 2965,
+      "train_runtime": 17598.6446,
+      "train_tokens_per_second": 9360.621
+    },
+    {
+      "epoch": 1.1956521739130435,
+      "grad_norm": 0.823324829074732,
+      "learning_rate": 3.490441227972542e-06,
+      "loss": 0.3418,
+      "num_input_tokens_seen": 164979064,
+      "step": 2970,
+      "train_runtime": 17621.1536,
+      "train_tokens_per_second": 9362.558
+    },
+    {
+      "epoch": 1.1976650563607085,
+      "grad_norm": 0.7331568857712636,
+      "learning_rate": 3.475377376703469e-06,
+      "loss": 0.3125,
+      "num_input_tokens_seen": 165256632,
+      "step": 2975,
+      "train_runtime": 17653.1146,
+      "train_tokens_per_second": 9361.33
+    },
+    {
+      "epoch": 1.1996779388083736,
+      "grad_norm": 0.7496010920309691,
+      "learning_rate": 3.4603287673157866e-06,
+      "loss": 0.3327,
+      "num_input_tokens_seen": 165548776,
+      "step": 2980,
+      "train_runtime": 17677.7292,
+      "train_tokens_per_second": 9364.821
+    },
+    {
+      "epoch": 1.2016908212560387,
+      "grad_norm": 0.7777467777535843,
+      "learning_rate": 3.445295550252705e-06,
+      "loss": 0.3376,
+      "num_input_tokens_seen": 165831856,
+      "step": 2985,
+      "train_runtime": 17708.3539,
+      "train_tokens_per_second": 9364.612
+    },
+    {
+      "epoch": 1.2037037037037037,
+      "grad_norm": 0.7361665250523245,
+      "learning_rate": 3.430277875803552e-06,
+      "loss": 0.3366,
+      "num_input_tokens_seen": 166105248,
+      "step": 2990,
+      "train_runtime": 17738.5027,
+      "train_tokens_per_second": 9364.108
+    },
+    {
+      "epoch": 1.2057165861513688,
+      "grad_norm": 0.8956938399130385,
+      "learning_rate": 3.4152758941022756e-06,
+      "loss": 0.3613,
+      "num_input_tokens_seen": 166369016,
+      "step": 2995,
+      "train_runtime": 17772.8734,
+      "train_tokens_per_second": 9360.84
+    },
+    {
+      "epoch": 1.2077294685990339,
+      "grad_norm": 0.6907640502988542,
+      "learning_rate": 3.400289755125941e-06,
+      "loss": 0.3354,
+      "num_input_tokens_seen": 166646592,
+      "step": 3000,
+      "train_runtime": 17804.6209,
+      "train_tokens_per_second": 9359.738
+    },
+    {
+      "epoch": 1.209742351046699,
+      "grad_norm": 0.7309740632690966,
+      "learning_rate": 3.3853196086932295e-06,
+      "loss": 0.3338,
+      "num_input_tokens_seen": 166923616,
+      "step": 3005,
+      "train_runtime": 17843.4692,
+      "train_tokens_per_second": 9354.886
+    },
+    {
+      "epoch": 1.211755233494364,
+      "grad_norm": 0.7595352673544153,
+      "learning_rate": 3.370365604462945e-06,
+      "loss": 0.3438,
+      "num_input_tokens_seen": 167215512,
+      "step": 3010,
+      "train_runtime": 17870.6345,
+      "train_tokens_per_second": 9356.999
+    },
+    {
+      "epoch": 1.213768115942029,
+      "grad_norm": 0.7028122076223309,
+      "learning_rate": 3.3554278919325127e-06,
+      "loss": 0.3376,
+      "num_input_tokens_seen": 167509488,
+      "step": 3015,
+      "train_runtime": 17901.1023,
+      "train_tokens_per_second": 9357.496
+    },
+    {
+      "epoch": 1.2157809983896941,
+      "grad_norm": 0.7111819738680615,
+      "learning_rate": 3.34050662043649e-06,
+      "loss": 0.3314,
+      "num_input_tokens_seen": 167789280,
+      "step": 3020,
+      "train_runtime": 17931.195,
+      "train_tokens_per_second": 9357.395
+    },
+    {
+      "epoch": 1.2177938808373592,
+      "grad_norm": 0.7590292894827062,
+      "learning_rate": 3.3256019391450696e-06,
+      "loss": 0.3364,
+      "num_input_tokens_seen": 168056448,
+      "step": 3025,
+      "train_runtime": 17957.8905,
+      "train_tokens_per_second": 9358.362
+    },
+    {
+      "epoch": 1.2198067632850242,
+      "grad_norm": 0.6953403290663392,
+      "learning_rate": 3.310713997062588e-06,
+      "loss": 0.3474,
+      "num_input_tokens_seen": 168336992,
+      "step": 3030,
+      "train_runtime": 17983.59,
+      "train_tokens_per_second": 9360.589
+    },
+    {
+      "epoch": 1.2218196457326893,
+      "grad_norm": 0.751892779351779,
+      "learning_rate": 3.29584294302604e-06,
+      "loss": 0.3311,
+      "num_input_tokens_seen": 168615880,
+      "step": 3035,
+      "train_runtime": 18015.8173,
+      "train_tokens_per_second": 9359.324
+    },
+    {
+      "epoch": 1.2238325281803544,
+      "grad_norm": 0.7621419343747116,
+      "learning_rate": 3.2809889257035853e-06,
+      "loss": 0.3247,
+      "num_input_tokens_seen": 168917480,
+      "step": 3040,
+      "train_runtime": 18042.5222,
+      "train_tokens_per_second": 9362.188
+    },
+    {
+      "epoch": 1.2258454106280192,
+      "grad_norm": 0.7511480008074028,
+      "learning_rate": 3.266152093593069e-06,
+      "loss": 0.3354,
+      "num_input_tokens_seen": 169188184,
+      "step": 3045,
+      "train_runtime": 18074.1421,
+      "train_tokens_per_second": 9360.786
+    },
+    {
+      "epoch": 1.2278582930756845,
+      "grad_norm": 0.8264538043627928,
+      "learning_rate": 3.2513325950205277e-06,
+      "loss": 0.3291,
+      "num_input_tokens_seen": 169435608,
+      "step": 3050,
+      "train_runtime": 18105.7395,
+      "train_tokens_per_second": 9358.116
+    },
+    {
+      "epoch": 1.2298711755233493,
+      "grad_norm": 0.7224058795350149,
+      "learning_rate": 3.2365305781387156e-06,
+      "loss": 0.3244,
+      "num_input_tokens_seen": 169656664,
+      "step": 3055,
+      "train_runtime": 18136.5928,
+      "train_tokens_per_second": 9354.385
+    },
+    {
+      "epoch": 1.2318840579710144,
+      "grad_norm": 0.7231781076781525,
+      "learning_rate": 3.2217461909256186e-06,
+      "loss": 0.3699,
+      "num_input_tokens_seen": 169947560,
+      "step": 3060,
+      "train_runtime": 18175.6461,
+      "train_tokens_per_second": 9350.29
+    },
+    {
+      "epoch": 1.2338969404186795,
+      "grad_norm": 0.7873625781421767,
+      "learning_rate": 3.2069795811829772e-06,
+      "loss": 0.3279,
+      "num_input_tokens_seen": 170231144,
+      "step": 3065,
+      "train_runtime": 18203.8102,
+      "train_tokens_per_second": 9351.402
+    },
+    {
+      "epoch": 1.2359098228663445,
+      "grad_norm": 0.7721173860632746,
+      "learning_rate": 3.192230896534808e-06,
+      "loss": 0.3413,
+      "num_input_tokens_seen": 170502840,
+      "step": 3070,
+      "train_runtime": 18236.1944,
+      "train_tokens_per_second": 9349.694
+    },
+    {
+      "epoch": 1.2379227053140096,
+      "grad_norm": 0.7525581504480072,
+      "learning_rate": 3.1775002844259253e-06,
+      "loss": 0.3264,
+      "num_input_tokens_seen": 170734664,
+      "step": 3075,
+      "train_runtime": 18264.3841,
+      "train_tokens_per_second": 9347.956
+    },
+    {
+      "epoch": 1.2399355877616747,
+      "grad_norm": 0.7816408171271338,
+      "learning_rate": 3.162787892120473e-06,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 170977240,
+      "step": 3080,
+      "train_runtime": 18296.8007,
+      "train_tokens_per_second": 9344.652
+    },
+    {
+      "epoch": 1.2419484702093397,
+      "grad_norm": 0.9986565013254199,
+      "learning_rate": 3.148093866700445e-06,
+      "loss": 0.3323,
+      "num_input_tokens_seen": 171242488,
+      "step": 3085,
+      "train_runtime": 18325.0733,
+      "train_tokens_per_second": 9344.71
+    },
+    {
+      "epoch": 1.2439613526570048,
+      "grad_norm": 0.7799935201338836,
+      "learning_rate": 3.1334183550642217e-06,
+      "loss": 0.3175,
+      "num_input_tokens_seen": 171525896,
+      "step": 3090,
+      "train_runtime": 18357.2773,
+      "train_tokens_per_second": 9343.755
+    },
+    {
+      "epoch": 1.2459742351046699,
+      "grad_norm": 0.6749592540953486,
+      "learning_rate": 3.1187615039250984e-06,
+      "loss": 0.3386,
+      "num_input_tokens_seen": 171811240,
+      "step": 3095,
+      "train_runtime": 18383.1956,
+      "train_tokens_per_second": 9346.103
+    },
+    {
+      "epoch": 1.247987117552335,
+      "grad_norm": 0.7873297191568724,
+      "learning_rate": 3.1041234598098157e-06,
+      "loss": 0.3429,
+      "num_input_tokens_seen": 172090640,
+      "step": 3100,
+      "train_runtime": 18410.6719,
+      "train_tokens_per_second": 9347.331
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7311174087067115,
+      "learning_rate": 3.0895043690571018e-06,
+      "loss": 0.3215,
+      "num_input_tokens_seen": 172368360,
+      "step": 3105,
+      "train_runtime": 18435.6044,
+      "train_tokens_per_second": 9349.754
+    },
+    {
+      "epoch": 1.252012882447665,
+      "grad_norm": 0.7340055274437848,
+      "learning_rate": 3.0749043778162015e-06,
+      "loss": 0.3228,
+      "num_input_tokens_seen": 172676096,
+      "step": 3110,
+      "train_runtime": 18459.303,
+      "train_tokens_per_second": 9354.421
+    },
+    {
+      "epoch": 1.2540257648953301,
+      "grad_norm": 0.8410671774005818,
+      "learning_rate": 3.0603236320454195e-06,
+      "loss": 0.3515,
+      "num_input_tokens_seen": 172936224,
+      "step": 3115,
+      "train_runtime": 18489.5182,
+      "train_tokens_per_second": 9353.203
+    },
+    {
+      "epoch": 1.2560386473429952,
+      "grad_norm": 0.7459570100086268,
+      "learning_rate": 3.0457622775106643e-06,
+      "loss": 0.3425,
+      "num_input_tokens_seen": 173218720,
+      "step": 3120,
+      "train_runtime": 18517.1498,
+      "train_tokens_per_second": 9354.502
+    },
+    {
+      "epoch": 1.2580515297906603,
+      "grad_norm": 0.7684560276734097,
+      "learning_rate": 3.0312204597839824e-06,
+      "loss": 0.337,
+      "num_input_tokens_seen": 173492936,
+      "step": 3125,
+      "train_runtime": 18550.458,
+      "train_tokens_per_second": 9352.488
+    },
+    {
+      "epoch": 1.2600644122383253,
+      "grad_norm": 0.8698419679449874,
+      "learning_rate": 3.01669832424211e-06,
+      "loss": 0.3311,
+      "num_input_tokens_seen": 173767880,
+      "step": 3130,
+      "train_runtime": 18584.1169,
+      "train_tokens_per_second": 9350.344
+    },
+    {
+      "epoch": 1.2620772946859904,
+      "grad_norm": 0.7872327503936718,
+      "learning_rate": 3.0021960160650197e-06,
+      "loss": 0.3308,
+      "num_input_tokens_seen": 174050368,
+      "step": 3135,
+      "train_runtime": 18617.6012,
+      "train_tokens_per_second": 9348.7
+    },
+    {
+      "epoch": 1.2640901771336555,
+      "grad_norm": 0.7433805951484156,
+      "learning_rate": 2.987713680234462e-06,
+      "loss": 0.3531,
+      "num_input_tokens_seen": 174313792,
+      "step": 3140,
+      "train_runtime": 18649.6559,
+      "train_tokens_per_second": 9346.756
+    },
+    {
+      "epoch": 1.2661030595813205,
+      "grad_norm": 0.7919165158640763,
+      "learning_rate": 2.973251461532527e-06,
+      "loss": 0.3143,
+      "num_input_tokens_seen": 174600328,
+      "step": 3145,
+      "train_runtime": 18679.0881,
+      "train_tokens_per_second": 9347.369
+    },
+    {
+      "epoch": 1.2681159420289856,
+      "grad_norm": 0.7468136277033092,
+      "learning_rate": 2.9588095045401844e-06,
+      "loss": 0.3245,
+      "num_input_tokens_seen": 174884848,
+      "step": 3150,
+      "train_runtime": 18715.7118,
+      "train_tokens_per_second": 9344.28
+    },
+    {
+      "epoch": 1.2701288244766507,
+      "grad_norm": 0.8656466132671339,
+      "learning_rate": 2.944387953635849e-06,
+      "loss": 0.3312,
+      "num_input_tokens_seen": 175166224,
+      "step": 3155,
+      "train_runtime": 18741.1804,
+      "train_tokens_per_second": 9346.595
+    },
+    {
+      "epoch": 1.2721417069243155,
+      "grad_norm": 0.7640821703374746,
+      "learning_rate": 2.929986952993933e-06,
+      "loss": 0.3486,
+      "num_input_tokens_seen": 175474184,
+      "step": 3160,
+      "train_runtime": 18771.7891,
+      "train_tokens_per_second": 9347.76
+    },
+    {
+      "epoch": 1.2741545893719808,
+      "grad_norm": 0.8039377052313604,
+      "learning_rate": 2.9156066465833997e-06,
+      "loss": 0.344,
+      "num_input_tokens_seen": 175728456,
+      "step": 3165,
+      "train_runtime": 18805.5129,
+      "train_tokens_per_second": 9344.518
+    },
+    {
+      "epoch": 1.2761674718196456,
+      "grad_norm": 0.8518044177666888,
+      "learning_rate": 2.9012471781663375e-06,
+      "loss": 0.318,
+      "num_input_tokens_seen": 175982264,
+      "step": 3170,
+      "train_runtime": 18830.6646,
+      "train_tokens_per_second": 9345.515
+    },
+    {
+      "epoch": 1.278180354267311,
+      "grad_norm": 0.7749202995717128,
+      "learning_rate": 2.886908691296504e-06,
+      "loss": 0.35,
+      "num_input_tokens_seen": 176253240,
+      "step": 3175,
+      "train_runtime": 18863.0765,
+      "train_tokens_per_second": 9343.823
+    },
+    {
+      "epoch": 1.2801932367149758,
+      "grad_norm": 0.8542732138318762,
+      "learning_rate": 2.872591329317908e-06,
+      "loss": 0.3424,
+      "num_input_tokens_seen": 176536144,
+      "step": 3180,
+      "train_runtime": 18890.6574,
+      "train_tokens_per_second": 9345.156
+    },
+    {
+      "epoch": 1.2822061191626408,
+      "grad_norm": 1.0674926599994714,
+      "learning_rate": 2.8582952353633708e-06,
+      "loss": 0.3211,
+      "num_input_tokens_seen": 176793320,
+      "step": 3185,
+      "train_runtime": 18911.4761,
+      "train_tokens_per_second": 9348.467
+    },
+    {
+      "epoch": 1.2842190016103059,
+      "grad_norm": 0.9387952828326972,
+      "learning_rate": 2.844020552353087e-06,
+      "loss": 0.3323,
+      "num_input_tokens_seen": 177074552,
+      "step": 3190,
+      "train_runtime": 18944.2952,
+      "train_tokens_per_second": 9347.117
+    },
+    {
+      "epoch": 1.286231884057971,
+      "grad_norm": 0.7384989951672372,
+      "learning_rate": 2.8297674229932086e-06,
+      "loss": 0.3416,
+      "num_input_tokens_seen": 177357576,
+      "step": 3195,
+      "train_runtime": 18977.8751,
+      "train_tokens_per_second": 9345.492
+    },
+    {
+      "epoch": 1.288244766505636,
+      "grad_norm": 0.768247162518729,
+      "learning_rate": 2.8155359897744117e-06,
+      "loss": 0.3514,
+      "num_input_tokens_seen": 177630464,
+      "step": 3200,
+      "train_runtime": 19006.2632,
+      "train_tokens_per_second": 9345.891
+    },
+    {
+      "epoch": 1.290257648953301,
+      "grad_norm": 0.772790059841373,
+      "learning_rate": 2.8013263949704706e-06,
+      "loss": 0.3166,
+      "num_input_tokens_seen": 177907472,
+      "step": 3205,
+      "train_runtime": 19038.0646,
+      "train_tokens_per_second": 9344.83
+    },
+    {
+      "epoch": 1.2922705314009661,
+      "grad_norm": 0.7276379642935875,
+      "learning_rate": 2.7871387806368426e-06,
+      "loss": 0.3416,
+      "num_input_tokens_seen": 178215720,
+      "step": 3210,
+      "train_runtime": 19059.93,
+      "train_tokens_per_second": 9350.282
+    },
+    {
+      "epoch": 1.2942834138486312,
+      "grad_norm": 0.806980532147328,
+      "learning_rate": 2.772973288609235e-06,
+      "loss": 0.309,
+      "num_input_tokens_seen": 178482056,
+      "step": 3215,
+      "train_runtime": 19086.7029,
+      "train_tokens_per_second": 9351.12
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.8123491796211115,
+      "learning_rate": 2.7588300605022007e-06,
+      "loss": 0.3634,
+      "num_input_tokens_seen": 178754976,
+      "step": 3220,
+      "train_runtime": 19111.4415,
+      "train_tokens_per_second": 9353.296
+    },
+    {
+      "epoch": 1.2983091787439613,
+      "grad_norm": 0.7108104554957614,
+      "learning_rate": 2.744709237707717e-06,
+      "loss": 0.335,
+      "num_input_tokens_seen": 179058256,
+      "step": 3225,
+      "train_runtime": 19143.45,
+      "train_tokens_per_second": 9353.5
+    },
+    {
+      "epoch": 1.3003220611916264,
+      "grad_norm": 0.7282761881972113,
+      "learning_rate": 2.7306109613937687e-06,
+      "loss": 0.3362,
+      "num_input_tokens_seen": 179339520,
+      "step": 3230,
+      "train_runtime": 19173.9813,
+      "train_tokens_per_second": 9353.275
+    },
+    {
+      "epoch": 1.3023349436392915,
+      "grad_norm": 0.7246329837325194,
+      "learning_rate": 2.7165353725029454e-06,
+      "loss": 0.3185,
+      "num_input_tokens_seen": 179624536,
+      "step": 3235,
+      "train_runtime": 19205.4034,
+      "train_tokens_per_second": 9352.812
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 0.8014353650355803,
+      "learning_rate": 2.7024826117510193e-06,
+      "loss": 0.337,
+      "num_input_tokens_seen": 179905904,
+      "step": 3240,
+      "train_runtime": 19232.9193,
+      "train_tokens_per_second": 9354.061
+    },
+    {
+      "epoch": 1.3063607085346216,
+      "grad_norm": 0.6850413613384719,
+      "learning_rate": 2.688452819625553e-06,
+      "loss": 0.3196,
+      "num_input_tokens_seen": 180211712,
+      "step": 3245,
+      "train_runtime": 19263.0365,
+      "train_tokens_per_second": 9355.312
+    },
+    {
+      "epoch": 1.3083735909822867,
+      "grad_norm": 0.7337323572168991,
+      "learning_rate": 2.674446136384486e-06,
+      "loss": 0.3395,
+      "num_input_tokens_seen": 180477328,
+      "step": 3250,
+      "train_runtime": 19293.7632,
+      "train_tokens_per_second": 9354.18
+    },
+    {
+      "epoch": 1.3103864734299517,
+      "grad_norm": 0.7030498455097879,
+      "learning_rate": 2.6604627020547334e-06,
+      "loss": 0.3552,
+      "num_input_tokens_seen": 180770296,
+      "step": 3255,
+      "train_runtime": 19328.9791,
+      "train_tokens_per_second": 9352.294
+    },
+    {
+      "epoch": 1.3123993558776168,
+      "grad_norm": 0.6953831746788743,
+      "learning_rate": 2.6465026564307924e-06,
+      "loss": 0.3285,
+      "num_input_tokens_seen": 181027688,
+      "step": 3260,
+      "train_runtime": 19358.9249,
+      "train_tokens_per_second": 9351.123
+    },
+    {
+      "epoch": 1.3144122383252819,
+      "grad_norm": 0.7915413597134223,
+      "learning_rate": 2.6325661390733303e-06,
+      "loss": 0.345,
+      "num_input_tokens_seen": 181309080,
+      "step": 3265,
+      "train_runtime": 19391.5932,
+      "train_tokens_per_second": 9349.881
+    },
+    {
+      "epoch": 1.316425120772947,
+      "grad_norm": 0.6774870108488009,
+      "learning_rate": 2.6186532893078087e-06,
+      "loss": 0.3342,
+      "num_input_tokens_seen": 181569064,
+      "step": 3270,
+      "train_runtime": 19422.9557,
+      "train_tokens_per_second": 9348.169
+    },
+    {
+      "epoch": 1.318438003220612,
+      "grad_norm": 0.8367363945695063,
+      "learning_rate": 2.6047642462230756e-06,
+      "loss": 0.3451,
+      "num_input_tokens_seen": 181836376,
+      "step": 3275,
+      "train_runtime": 19457.9527,
+      "train_tokens_per_second": 9345.093
+    },
+    {
+      "epoch": 1.320450885668277,
+      "grad_norm": 0.7317270241532496,
+      "learning_rate": 2.590899148669981e-06,
+      "loss": 0.3188,
+      "num_input_tokens_seen": 182109648,
+      "step": 3280,
+      "train_runtime": 19492.0118,
+      "train_tokens_per_second": 9342.784
+    },
+    {
+      "epoch": 1.322463768115942,
+      "grad_norm": 0.7226234121959383,
+      "learning_rate": 2.5770581352599876e-06,
+      "loss": 0.3311,
+      "num_input_tokens_seen": 182384584,
+      "step": 3285,
+      "train_runtime": 19526.0236,
+      "train_tokens_per_second": 9340.59
+    },
+    {
+      "epoch": 1.3244766505636072,
+      "grad_norm": 0.6899938873543758,
+      "learning_rate": 2.5632413443637887e-06,
+      "loss": 0.3586,
+      "num_input_tokens_seen": 182680016,
+      "step": 3290,
+      "train_runtime": 19556.0873,
+      "train_tokens_per_second": 9341.338
+    },
+    {
+      "epoch": 1.326489533011272,
+      "grad_norm": 0.7820360325459759,
+      "learning_rate": 2.5494489141099155e-06,
+      "loss": 0.3404,
+      "num_input_tokens_seen": 182967400,
+      "step": 3295,
+      "train_runtime": 19583.7937,
+      "train_tokens_per_second": 9342.797
+    },
+    {
+      "epoch": 1.3285024154589373,
+      "grad_norm": 0.7082182159459535,
+      "learning_rate": 2.535680982383364e-06,
+      "loss": 0.3398,
+      "num_input_tokens_seen": 183255320,
+      "step": 3300,
+      "train_runtime": 19612.9858,
+      "train_tokens_per_second": 9343.571
+    },
+    {
+      "epoch": 1.3305152979066022,
+      "grad_norm": 0.7704004053785951,
+      "learning_rate": 2.5219376868242173e-06,
+      "loss": 0.3251,
+      "num_input_tokens_seen": 183522760,
+      "step": 3305,
+      "train_runtime": 19639.0649,
+      "train_tokens_per_second": 9344.781
+    },
+    {
+      "epoch": 1.3325281803542672,
+      "grad_norm": 0.7467779259050853,
+      "learning_rate": 2.5082191648262654e-06,
+      "loss": 0.3322,
+      "num_input_tokens_seen": 183800888,
+      "step": 3310,
+      "train_runtime": 19670.4057,
+      "train_tokens_per_second": 9344.031
+    },
+    {
+      "epoch": 1.3345410628019323,
+      "grad_norm": 0.8075790233799084,
+      "learning_rate": 2.4945255535356343e-06,
+      "loss": 0.3411,
+      "num_input_tokens_seen": 184077008,
+      "step": 3315,
+      "train_runtime": 19702.6205,
+      "train_tokens_per_second": 9342.768
+    },
+    {
+      "epoch": 1.3365539452495974,
+      "grad_norm": 0.8176544087917905,
+      "learning_rate": 2.4808569898494144e-06,
+      "loss": 0.3503,
+      "num_input_tokens_seen": 184356944,
+      "step": 3320,
+      "train_runtime": 19723.853,
+      "train_tokens_per_second": 9346.903
+    },
+    {
+      "epoch": 1.3385668276972624,
+      "grad_norm": 0.6936054607476786,
+      "learning_rate": 2.467213610414286e-06,
+      "loss": 0.3164,
+      "num_input_tokens_seen": 184641776,
+      "step": 3325,
+      "train_runtime": 19756.261,
+      "train_tokens_per_second": 9345.988
+    },
+    {
+      "epoch": 1.3405797101449275,
+      "grad_norm": 0.7705012548742098,
+      "learning_rate": 2.453595551625165e-06,
+      "loss": 0.3492,
+      "num_input_tokens_seen": 184926320,
+      "step": 3330,
+      "train_runtime": 19787.4693,
+      "train_tokens_per_second": 9345.628
+    },
+    {
+      "epoch": 1.3425925925925926,
+      "grad_norm": 0.6663201842465876,
+      "learning_rate": 2.4400029496238312e-06,
+      "loss": 0.328,
+      "num_input_tokens_seen": 185218872,
+      "step": 3335,
+      "train_runtime": 19817.7222,
+      "train_tokens_per_second": 9346.123
+    },
+    {
+      "epoch": 1.3446054750402576,
+      "grad_norm": 0.7247129110820784,
+      "learning_rate": 2.4264359402975683e-06,
+      "loss": 0.3201,
+      "num_input_tokens_seen": 185509824,
+      "step": 3340,
+      "train_runtime": 19847.8942,
+      "train_tokens_per_second": 9346.575
+    },
+    {
+      "epoch": 1.3466183574879227,
+      "grad_norm": 0.6744408261310794,
+      "learning_rate": 2.4128946592778086e-06,
+      "loss": 0.326,
+      "num_input_tokens_seen": 185807552,
+      "step": 3345,
+      "train_runtime": 19877.2199,
+      "train_tokens_per_second": 9347.764
+    },
+    {
+      "epoch": 1.3486312399355878,
+      "grad_norm": 0.678859949650584,
+      "learning_rate": 2.399379241938769e-06,
+      "loss": 0.3266,
+      "num_input_tokens_seen": 186086368,
+      "step": 3350,
+      "train_runtime": 19911.0543,
+      "train_tokens_per_second": 9345.882
+    },
+    {
+      "epoch": 1.3506441223832528,
+      "grad_norm": 0.7520707447256646,
+      "learning_rate": 2.385889823396108e-06,
+      "loss": 0.3347,
+      "num_input_tokens_seen": 186354824,
+      "step": 3355,
+      "train_runtime": 19943.5224,
+      "train_tokens_per_second": 9344.128
+    },
+    {
+      "epoch": 1.3526570048309179,
+      "grad_norm": 0.6762639568655372,
+      "learning_rate": 2.3724265385055697e-06,
+      "loss": 0.318,
+      "num_input_tokens_seen": 186671600,
+      "step": 3360,
+      "train_runtime": 19969.2755,
+      "train_tokens_per_second": 9347.941
+    },
+    {
+      "epoch": 1.354669887278583,
+      "grad_norm": 0.8167801902667485,
+      "learning_rate": 2.3589895218616356e-06,
+      "loss": 0.339,
+      "num_input_tokens_seen": 186944072,
+      "step": 3365,
+      "train_runtime": 20001.9517,
+      "train_tokens_per_second": 9346.292
+    },
+    {
+      "epoch": 1.356682769726248,
+      "grad_norm": 0.7222385240429897,
+      "learning_rate": 2.3455789077961816e-06,
+      "loss": 0.3243,
+      "num_input_tokens_seen": 187240712,
+      "step": 3370,
+      "train_runtime": 20037.0355,
+      "train_tokens_per_second": 9344.731
+    },
+    {
+      "epoch": 1.358695652173913,
+      "grad_norm": 0.8039813920937489,
+      "learning_rate": 2.332194830377127e-06,
+      "loss": 0.3372,
+      "num_input_tokens_seen": 187535336,
+      "step": 3375,
+      "train_runtime": 20067.6403,
+      "train_tokens_per_second": 9345.161
+    },
+    {
+      "epoch": 1.3607085346215781,
+      "grad_norm": 0.8316528793190969,
+      "learning_rate": 2.318837423407106e-06,
+      "loss": 0.3206,
+      "num_input_tokens_seen": 187812248,
+      "step": 3380,
+      "train_runtime": 20096.3947,
+      "train_tokens_per_second": 9345.569
+    },
+    {
+      "epoch": 1.3627214170692432,
+      "grad_norm": 0.7595681685698273,
+      "learning_rate": 2.3055068204221226e-06,
+      "loss": 0.3393,
+      "num_input_tokens_seen": 188073072,
+      "step": 3385,
+      "train_runtime": 20122.2842,
+      "train_tokens_per_second": 9346.507
+    },
+    {
+      "epoch": 1.3647342995169083,
+      "grad_norm": 0.7222074042267652,
+      "learning_rate": 2.292203154690218e-06,
+      "loss": 0.3328,
+      "num_input_tokens_seen": 188376696,
+      "step": 3390,
+      "train_runtime": 20150.7679,
+      "train_tokens_per_second": 9348.363
+    },
+    {
+      "epoch": 1.3667471819645733,
+      "grad_norm": 0.6795902408969277,
+      "learning_rate": 2.278926559210137e-06,
+      "loss": 0.3264,
+      "num_input_tokens_seen": 188653136,
+      "step": 3395,
+      "train_runtime": 20179.3258,
+      "train_tokens_per_second": 9348.832
+    },
+    {
+      "epoch": 1.3687600644122382,
+      "grad_norm": 0.7281120649894878,
+      "learning_rate": 2.2656771667099993e-06,
+      "loss": 0.3247,
+      "num_input_tokens_seen": 188957296,
+      "step": 3400,
+      "train_runtime": 20210.5929,
+      "train_tokens_per_second": 9349.419
+    },
+    {
+      "epoch": 1.3707729468599035,
+      "grad_norm": 0.8169612705560803,
+      "learning_rate": 2.2524551096459703e-06,
+      "loss": 0.3495,
+      "num_input_tokens_seen": 189219824,
+      "step": 3405,
+      "train_runtime": 20234.9838,
+      "train_tokens_per_second": 9351.123
+    },
+    {
+      "epoch": 1.3727858293075683,
+      "grad_norm": 0.7307166976268302,
+      "learning_rate": 2.239260520200941e-06,
+      "loss": 0.3114,
+      "num_input_tokens_seen": 189473144,
+      "step": 3410,
+      "train_runtime": 20270.5784,
+      "train_tokens_per_second": 9347.2
+    },
+    {
+      "epoch": 1.3747987117552336,
+      "grad_norm": 0.7832910961374582,
+      "learning_rate": 2.226093530283205e-06,
+      "loss": 0.3444,
+      "num_input_tokens_seen": 189723464,
+      "step": 3415,
+      "train_runtime": 20300.6805,
+      "train_tokens_per_second": 9345.67
+    },
+    {
+      "epoch": 1.3768115942028984,
+      "grad_norm": 0.7298425031791868,
+      "learning_rate": 2.21295427152514e-06,
+      "loss": 0.3071,
+      "num_input_tokens_seen": 189965672,
+      "step": 3420,
+      "train_runtime": 20328.7908,
+      "train_tokens_per_second": 9344.662
+    },
+    {
+      "epoch": 1.3788244766505637,
+      "grad_norm": 0.8260322411021098,
+      "learning_rate": 2.199842875281886e-06,
+      "loss": 0.3403,
+      "num_input_tokens_seen": 190245224,
+      "step": 3425,
+      "train_runtime": 20356.668,
+      "train_tokens_per_second": 9345.597
+    },
+    {
+      "epoch": 1.3808373590982286,
+      "grad_norm": 0.731742794001471,
+      "learning_rate": 2.186759472630045e-06,
+      "loss": 0.3183,
+      "num_input_tokens_seen": 190536872,
+      "step": 3430,
+      "train_runtime": 20385.4113,
+      "train_tokens_per_second": 9346.727
+    },
+    {
+      "epoch": 1.3828502415458936,
+      "grad_norm": 0.7338901274205261,
+      "learning_rate": 2.173704194366354e-06,
+      "loss": 0.3364,
+      "num_input_tokens_seen": 190816560,
+      "step": 3435,
+      "train_runtime": 20415.6378,
+      "train_tokens_per_second": 9346.588
+    },
+    {
+      "epoch": 1.3848631239935587,
+      "grad_norm": 0.8466352536358828,
+      "learning_rate": 2.1606771710063928e-06,
+      "loss": 0.3579,
+      "num_input_tokens_seen": 191069000,
+      "step": 3440,
+      "train_runtime": 20446.4658,
+      "train_tokens_per_second": 9344.842
+    },
+    {
+      "epoch": 1.3868760064412238,
+      "grad_norm": 0.9601160315693671,
+      "learning_rate": 2.1476785327832715e-06,
+      "loss": 0.3411,
+      "num_input_tokens_seen": 191327320,
+      "step": 3445,
+      "train_runtime": 20474.6232,
+      "train_tokens_per_second": 9344.608
+    },
+    {
+      "epoch": 1.3888888888888888,
+      "grad_norm": 0.7481376156397551,
+      "learning_rate": 2.1347084096463293e-06,
+      "loss": 0.3385,
+      "num_input_tokens_seen": 191614592,
+      "step": 3450,
+      "train_runtime": 20505.0361,
+      "train_tokens_per_second": 9344.758
+    },
+    {
+      "epoch": 1.390901771336554,
+      "grad_norm": 0.7663923690004009,
+      "learning_rate": 2.1217669312598364e-06,
+      "loss": 0.3458,
+      "num_input_tokens_seen": 191894816,
+      "step": 3455,
+      "train_runtime": 20529.7574,
+      "train_tokens_per_second": 9347.155
+    },
+    {
+      "epoch": 1.392914653784219,
+      "grad_norm": 0.7397049845010489,
+      "learning_rate": 2.108854227001693e-06,
+      "loss": 0.3421,
+      "num_input_tokens_seen": 192184888,
+      "step": 3460,
+      "train_runtime": 20559.6789,
+      "train_tokens_per_second": 9347.66
+    },
+    {
+      "epoch": 1.394927536231884,
+      "grad_norm": 0.7927356458888435,
+      "learning_rate": 2.0959704259621457e-06,
+      "loss": 0.3234,
+      "num_input_tokens_seen": 192457000,
+      "step": 3465,
+      "train_runtime": 20585.6324,
+      "train_tokens_per_second": 9349.093
+    },
+    {
+      "epoch": 1.396940418679549,
+      "grad_norm": 0.824558996428323,
+      "learning_rate": 2.0831156569424877e-06,
+      "loss": 0.3269,
+      "num_input_tokens_seen": 192726472,
+      "step": 3470,
+      "train_runtime": 20607.4804,
+      "train_tokens_per_second": 9352.258
+    },
+    {
+      "epoch": 1.3989533011272142,
+      "grad_norm": 0.7749186490987381,
+      "learning_rate": 2.070290048453777e-06,
+      "loss": 0.3272,
+      "num_input_tokens_seen": 193004472,
+      "step": 3475,
+      "train_runtime": 20636.1298,
+      "train_tokens_per_second": 9352.746
+    },
+    {
+      "epoch": 1.4009661835748792,
+      "grad_norm": 0.8341663054298187,
+      "learning_rate": 2.0574937287155496e-06,
+      "loss": 0.3186,
+      "num_input_tokens_seen": 193289112,
+      "step": 3480,
+      "train_runtime": 20665.8966,
+      "train_tokens_per_second": 9353.047
+    },
+    {
+      "epoch": 1.4029790660225443,
+      "grad_norm": 0.8211755789443504,
+      "learning_rate": 2.0447268256545327e-06,
+      "loss": 0.3457,
+      "num_input_tokens_seen": 193544272,
+      "step": 3485,
+      "train_runtime": 20689.4626,
+      "train_tokens_per_second": 9354.727
+    },
+    {
+      "epoch": 1.4049919484702094,
+      "grad_norm": 0.7073558545556239,
+      "learning_rate": 2.0319894669033758e-06,
+      "loss": 0.3205,
+      "num_input_tokens_seen": 193824776,
+      "step": 3490,
+      "train_runtime": 20720.5934,
+      "train_tokens_per_second": 9354.21
+    },
+    {
+      "epoch": 1.4070048309178744,
+      "grad_norm": 0.711399657032163,
+      "learning_rate": 2.0192817797993674e-06,
+      "loss": 0.3224,
+      "num_input_tokens_seen": 194114008,
+      "step": 3495,
+      "train_runtime": 20754.5625,
+      "train_tokens_per_second": 9352.835
+    },
+    {
+      "epoch": 1.4090177133655395,
+      "grad_norm": 0.7647394388497878,
+      "learning_rate": 2.0066038913831653e-06,
+      "loss": 0.3419,
+      "num_input_tokens_seen": 194341864,
+      "step": 3500,
+      "train_runtime": 20780.9041,
+      "train_tokens_per_second": 9351.945
+    },
+    {
+      "epoch": 1.4110305958132046,
+      "grad_norm": 0.7670540090543413,
+      "learning_rate": 1.9939559283975237e-06,
+      "loss": 0.323,
+      "num_input_tokens_seen": 194613464,
+      "step": 3505,
+      "train_runtime": 20811.8588,
+      "train_tokens_per_second": 9351.085
+    },
+    {
+      "epoch": 1.4130434782608696,
+      "grad_norm": 0.7364376095932686,
+      "learning_rate": 1.9813380172860305e-06,
+      "loss": 0.3141,
+      "num_input_tokens_seen": 194922320,
+      "step": 3510,
+      "train_runtime": 20841.3974,
+      "train_tokens_per_second": 9352.651
+    },
+    {
+      "epoch": 1.4150563607085347,
+      "grad_norm": 0.8203672018695578,
+      "learning_rate": 1.9687502841918345e-06,
+      "loss": 0.3237,
+      "num_input_tokens_seen": 195209048,
+      "step": 3515,
+      "train_runtime": 20868.3945,
+      "train_tokens_per_second": 9354.292
+    },
+    {
+      "epoch": 1.4170692431561998,
+      "grad_norm": 0.720277277547583,
+      "learning_rate": 1.956192854956397e-06,
+      "loss": 0.3181,
+      "num_input_tokens_seen": 195495712,
+      "step": 3520,
+      "train_runtime": 20895.04,
+      "train_tokens_per_second": 9356.082
+    },
+    {
+      "epoch": 1.4190821256038646,
+      "grad_norm": 0.7815451198268726,
+      "learning_rate": 1.9436658551182234e-06,
+      "loss": 0.3469,
+      "num_input_tokens_seen": 195782624,
+      "step": 3525,
+      "train_runtime": 20928.2545,
+      "train_tokens_per_second": 9354.943
+    },
+    {
+      "epoch": 1.4210950080515299,
+      "grad_norm": 0.7711741089338209,
+      "learning_rate": 1.9311694099116143e-06,
+      "loss": 0.3269,
+      "num_input_tokens_seen": 196062136,
+      "step": 3530,
+      "train_runtime": 20956.2551,
+      "train_tokens_per_second": 9355.781
+    },
+    {
+      "epoch": 1.4231078904991947,
+      "grad_norm": 0.8390379841615295,
+      "learning_rate": 1.9187036442654093e-06,
+      "loss": 0.3526,
+      "num_input_tokens_seen": 196327816,
+      "step": 3535,
+      "train_runtime": 20978.8526,
+      "train_tokens_per_second": 9358.368
+    },
+    {
+      "epoch": 1.42512077294686,
+      "grad_norm": 0.7667132719622095,
+      "learning_rate": 1.906268682801739e-06,
+      "loss": 0.3434,
+      "num_input_tokens_seen": 196587800,
+      "step": 3540,
+      "train_runtime": 21006.0881,
+      "train_tokens_per_second": 9358.611
+    },
+    {
+      "epoch": 1.4271336553945249,
+      "grad_norm": 0.7860988349811444,
+      "learning_rate": 1.893864649834783e-06,
+      "loss": 0.3354,
+      "num_input_tokens_seen": 196871416,
+      "step": 3545,
+      "train_runtime": 21041.6887,
+      "train_tokens_per_second": 9356.256
+    },
+    {
+      "epoch": 1.4291465378421901,
+      "grad_norm": 0.6849830883231917,
+      "learning_rate": 1.881491669369519e-06,
+      "loss": 0.319,
+      "num_input_tokens_seen": 197135528,
+      "step": 3550,
+      "train_runtime": 21072.1759,
+      "train_tokens_per_second": 9355.253
+    },
+    {
+      "epoch": 1.431159420289855,
+      "grad_norm": 0.7140942317202141,
+      "learning_rate": 1.8691498651004936e-06,
+      "loss": 0.3186,
+      "num_input_tokens_seen": 197398224,
+      "step": 3555,
+      "train_runtime": 21105.1794,
+      "train_tokens_per_second": 9353.07
+    },
+    {
+      "epoch": 1.43317230273752,
+      "grad_norm": 0.7574411567964284,
+      "learning_rate": 1.8568393604105778e-06,
+      "loss": 0.3354,
+      "num_input_tokens_seen": 197661288,
+      "step": 3560,
+      "train_runtime": 21130.2104,
+      "train_tokens_per_second": 9354.44
+    },
+    {
+      "epoch": 1.4351851851851851,
+      "grad_norm": 0.7080334559936231,
+      "learning_rate": 1.8445602783697375e-06,
+      "loss": 0.3398,
+      "num_input_tokens_seen": 197926592,
+      "step": 3565,
+      "train_runtime": 21163.2783,
+      "train_tokens_per_second": 9352.36
+    },
+    {
+      "epoch": 1.4371980676328502,
+      "grad_norm": 0.7957755143734638,
+      "learning_rate": 1.8323127417337977e-06,
+      "loss": 0.3395,
+      "num_input_tokens_seen": 198206904,
+      "step": 3570,
+      "train_runtime": 21194.1838,
+      "train_tokens_per_second": 9351.948
+    },
+    {
+      "epoch": 1.4392109500805152,
+      "grad_norm": 0.756589190658684,
+      "learning_rate": 1.8200968729432233e-06,
+      "loss": 0.342,
+      "num_input_tokens_seen": 198469936,
+      "step": 3575,
+      "train_runtime": 21229.5588,
+      "train_tokens_per_second": 9348.755
+    },
+    {
+      "epoch": 1.4412238325281803,
+      "grad_norm": 0.7455924352605094,
+      "learning_rate": 1.8079127941218894e-06,
+      "loss": 0.3365,
+      "num_input_tokens_seen": 198741632,
+      "step": 3580,
+      "train_runtime": 21252.4639,
+      "train_tokens_per_second": 9351.463
+    },
+    {
+      "epoch": 1.4432367149758454,
+      "grad_norm": 0.7563901585223551,
+      "learning_rate": 1.7957606270758626e-06,
+      "loss": 0.3311,
+      "num_input_tokens_seen": 199015992,
+      "step": 3585,
+      "train_runtime": 21278.2066,
+      "train_tokens_per_second": 9353.044
+    },
+    {
+      "epoch": 1.4452495974235104,
+      "grad_norm": 0.8050684971800192,
+      "learning_rate": 1.7836404932921842e-06,
+      "loss": 0.3381,
+      "num_input_tokens_seen": 199274512,
+      "step": 3590,
+      "train_runtime": 21301.4041,
+      "train_tokens_per_second": 9354.994
+    },
+    {
+      "epoch": 1.4472624798711755,
+      "grad_norm": 0.8128310140434445,
+      "learning_rate": 1.7715525139376482e-06,
+      "loss": 0.3316,
+      "num_input_tokens_seen": 199553176,
+      "step": 3595,
+      "train_runtime": 21327.7516,
+      "train_tokens_per_second": 9356.503
+    },
+    {
+      "epoch": 1.4492753623188406,
+      "grad_norm": 0.80186236632579,
+      "learning_rate": 1.7594968098576027e-06,
+      "loss": 0.3249,
+      "num_input_tokens_seen": 199817680,
+      "step": 3600,
+      "train_runtime": 21355.1788,
+      "train_tokens_per_second": 9356.872
+    },
+    {
+      "epoch": 1.4512882447665056,
+      "grad_norm": 0.7500745799891086,
+      "learning_rate": 1.7474735015747312e-06,
+      "loss": 0.323,
+      "num_input_tokens_seen": 200090712,
+      "step": 3605,
+      "train_runtime": 21380.3989,
+      "train_tokens_per_second": 9358.605
+    },
+    {
+      "epoch": 1.4533011272141707,
+      "grad_norm": 0.762084141671551,
+      "learning_rate": 1.735482709287854e-06,
+      "loss": 0.3295,
+      "num_input_tokens_seen": 200356408,
+      "step": 3610,
+      "train_runtime": 21408.4274,
+      "train_tokens_per_second": 9358.763
+    },
+    {
+      "epoch": 1.4553140096618358,
+      "grad_norm": 0.7327176618051108,
+      "learning_rate": 1.723524552870724e-06,
+      "loss": 0.3279,
+      "num_input_tokens_seen": 200644032,
+      "step": 3615,
+      "train_runtime": 21439.8746,
+      "train_tokens_per_second": 9358.452
+    },
+    {
+      "epoch": 1.4573268921095008,
+      "grad_norm": 0.7604501772538239,
+      "learning_rate": 1.711599151870823e-06,
+      "loss": 0.3332,
+      "num_input_tokens_seen": 200918008,
+      "step": 3620,
+      "train_runtime": 21460.7022,
+      "train_tokens_per_second": 9362.136
+    },
+    {
+      "epoch": 1.459339774557166,
+      "grad_norm": 0.7700182673423906,
+      "learning_rate": 1.6997066255081795e-06,
+      "loss": 0.3413,
+      "num_input_tokens_seen": 201217456,
+      "step": 3625,
+      "train_runtime": 21492.6058,
+      "train_tokens_per_second": 9362.171
+    },
+    {
+      "epoch": 1.461352657004831,
+      "grad_norm": 0.9077159732449974,
+      "learning_rate": 1.6878470926741646e-06,
+      "loss": 0.3363,
+      "num_input_tokens_seen": 201481680,
+      "step": 3630,
+      "train_runtime": 21530.8795,
+      "train_tokens_per_second": 9357.801
+    },
+    {
+      "epoch": 1.463365539452496,
+      "grad_norm": 0.7111378403358045,
+      "learning_rate": 1.6760206719303107e-06,
+      "loss": 0.3297,
+      "num_input_tokens_seen": 201762592,
+      "step": 3635,
+      "train_runtime": 21562.8521,
+      "train_tokens_per_second": 9356.953
+    },
+    {
+      "epoch": 1.465378421900161,
+      "grad_norm": 0.7664635794716855,
+      "learning_rate": 1.6642274815071213e-06,
+      "loss": 0.3243,
+      "num_input_tokens_seen": 202049912,
+      "step": 3640,
+      "train_runtime": 21596.7035,
+      "train_tokens_per_second": 9355.59
+    },
+    {
+      "epoch": 1.4673913043478262,
+      "grad_norm": 0.7340027598660989,
+      "learning_rate": 1.6524676393028955e-06,
+      "loss": 0.3416,
+      "num_input_tokens_seen": 202333320,
+      "step": 3645,
+      "train_runtime": 21628.0847,
+      "train_tokens_per_second": 9355.12
+    },
+    {
+      "epoch": 1.469404186795491,
+      "grad_norm": 0.8241694980062337,
+      "learning_rate": 1.640741262882538e-06,
+      "loss": 0.3344,
+      "num_input_tokens_seen": 202592424,
+      "step": 3650,
+      "train_runtime": 21652.8571,
+      "train_tokens_per_second": 9356.383
+    },
+    {
+      "epoch": 1.4714170692431563,
+      "grad_norm": 0.7255725103479399,
+      "learning_rate": 1.629048469476398e-06,
+      "loss": 0.318,
+      "num_input_tokens_seen": 202863200,
+      "step": 3655,
+      "train_runtime": 21678.0008,
+      "train_tokens_per_second": 9358.022
+    },
+    {
+      "epoch": 1.4734299516908211,
+      "grad_norm": 0.7058338249231842,
+      "learning_rate": 1.6173893759790871e-06,
+      "loss": 0.3243,
+      "num_input_tokens_seen": 203166800,
+      "step": 3660,
+      "train_runtime": 21711.1822,
+      "train_tokens_per_second": 9357.703
+    },
+    {
+      "epoch": 1.4754428341384864,
+      "grad_norm": 0.7789722213993872,
+      "learning_rate": 1.6057640989483158e-06,
+      "loss": 0.3158,
+      "num_input_tokens_seen": 203436976,
+      "step": 3665,
+      "train_runtime": 21743.9953,
+      "train_tokens_per_second": 9356.007
+    },
+    {
+      "epoch": 1.4774557165861513,
+      "grad_norm": 0.7524179352776947,
+      "learning_rate": 1.5941727546037223e-06,
+      "loss": 0.3177,
+      "num_input_tokens_seen": 203699416,
+      "step": 3670,
+      "train_runtime": 21773.4951,
+      "train_tokens_per_second": 9355.384
+    },
+    {
+      "epoch": 1.4794685990338166,
+      "grad_norm": 0.8260746313639024,
+      "learning_rate": 1.5826154588257204e-06,
+      "loss": 0.3176,
+      "num_input_tokens_seen": 203972208,
+      "step": 3675,
+      "train_runtime": 21796.8548,
+      "train_tokens_per_second": 9357.873
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.7696205164544199,
+      "learning_rate": 1.5710923271543282e-06,
+      "loss": 0.32,
+      "num_input_tokens_seen": 204233264,
+      "step": 3680,
+      "train_runtime": 21825.7251,
+      "train_tokens_per_second": 9357.456
+    },
+    {
+      "epoch": 1.4834943639291465,
+      "grad_norm": 0.6996295075938915,
+      "learning_rate": 1.5596034747880263e-06,
+      "loss": 0.3355,
+      "num_input_tokens_seen": 204494824,
+      "step": 3685,
+      "train_runtime": 21861.2893,
+      "train_tokens_per_second": 9354.198
+    },
+    {
+      "epoch": 1.4855072463768115,
+      "grad_norm": 0.7091627233641642,
+      "learning_rate": 1.5481490165825969e-06,
+      "loss": 0.3288,
+      "num_input_tokens_seen": 204799728,
+      "step": 3690,
+      "train_runtime": 21894.3832,
+      "train_tokens_per_second": 9353.985
+    },
+    {
+      "epoch": 1.4875201288244766,
+      "grad_norm": 0.834338576517347,
+      "learning_rate": 1.5367290670499802e-06,
+      "loss": 0.3284,
+      "num_input_tokens_seen": 205070904,
+      "step": 3695,
+      "train_runtime": 21921.2625,
+      "train_tokens_per_second": 9354.886
+    },
+    {
+      "epoch": 1.4895330112721417,
+      "grad_norm": 0.7019097217407337,
+      "learning_rate": 1.525343740357128e-06,
+      "loss": 0.3319,
+      "num_input_tokens_seen": 205343752,
+      "step": 3700,
+      "train_runtime": 21955.3724,
+      "train_tokens_per_second": 9352.779
+    },
+    {
+      "epoch": 1.4915458937198067,
+      "grad_norm": 0.7545572873350569,
+      "learning_rate": 1.5139931503248595e-06,
+      "loss": 0.3266,
+      "num_input_tokens_seen": 205628256,
+      "step": 3705,
+      "train_runtime": 21984.1537,
+      "train_tokens_per_second": 9353.476
+    },
+    {
+      "epoch": 1.4935587761674718,
+      "grad_norm": 0.7852431362121378,
+      "learning_rate": 1.5026774104267289e-06,
+      "loss": 0.3226,
+      "num_input_tokens_seen": 205912760,
+      "step": 3710,
+      "train_runtime": 22015.2182,
+      "train_tokens_per_second": 9353.201
+    },
+    {
+      "epoch": 1.4955716586151369,
+      "grad_norm": 0.7698037548572059,
+      "learning_rate": 1.4913966337878876e-06,
+      "loss": 0.3201,
+      "num_input_tokens_seen": 206174696,
+      "step": 3715,
+      "train_runtime": 22041.5864,
+      "train_tokens_per_second": 9353.896
+    },
+    {
+      "epoch": 1.497584541062802,
+      "grad_norm": 0.7328145417287896,
+      "learning_rate": 1.4801509331839553e-06,
+      "loss": 0.3276,
+      "num_input_tokens_seen": 206447872,
+      "step": 3720,
+      "train_runtime": 22073.2414,
+      "train_tokens_per_second": 9352.857
+    },
+    {
+      "epoch": 1.499597423510467,
+      "grad_norm": 0.7503172679945208,
+      "learning_rate": 1.4689404210398906e-06,
+      "loss": 0.3371,
+      "num_input_tokens_seen": 206738232,
+      "step": 3725,
+      "train_runtime": 22095.9797,
+      "train_tokens_per_second": 9356.373
+    },
+    {
+      "epoch": 1.501610305958132,
+      "grad_norm": 0.8280843366980284,
+      "learning_rate": 1.4577652094288652e-06,
+      "loss": 0.3509,
+      "num_input_tokens_seen": 207007936,
+      "step": 3730,
+      "train_runtime": 22128.0882,
+      "train_tokens_per_second": 9354.985
+    },
+    {
+      "epoch": 1.5036231884057971,
+      "grad_norm": 0.8111843946824732,
+      "learning_rate": 1.4466254100711496e-06,
+      "loss": 0.329,
+      "num_input_tokens_seen": 207275768,
+      "step": 3735,
+      "train_runtime": 22162.9449,
+      "train_tokens_per_second": 9352.357
+    },
+    {
+      "epoch": 1.5056360708534622,
+      "grad_norm": 0.850331935572953,
+      "learning_rate": 1.4355211343329912e-06,
+      "loss": 0.3273,
+      "num_input_tokens_seen": 207542112,
+      "step": 3740,
+      "train_runtime": 22182.175,
+      "train_tokens_per_second": 9356.256
+    },
+    {
+      "epoch": 1.5076489533011272,
+      "grad_norm": 0.7573499208213532,
+      "learning_rate": 1.4244524932255026e-06,
+      "loss": 0.2991,
+      "num_input_tokens_seen": 207818368,
+      "step": 3745,
+      "train_runtime": 22209.3056,
+      "train_tokens_per_second": 9357.265
+    },
+    {
+      "epoch": 1.5096618357487923,
+      "grad_norm": 0.7396393574795177,
+      "learning_rate": 1.4134195974035525e-06,
+      "loss": 0.3412,
+      "num_input_tokens_seen": 208118848,
+      "step": 3750,
+      "train_runtime": 22243.4626,
+      "train_tokens_per_second": 9356.405
+    },
+    {
+      "epoch": 1.5116747181964574,
+      "grad_norm": 0.7224774993646524,
+      "learning_rate": 1.4024225571646582e-06,
+      "loss": 0.3414,
+      "num_input_tokens_seen": 208398136,
+      "step": 3755,
+      "train_runtime": 22270.1861,
+      "train_tokens_per_second": 9357.719
+    },
+    {
+      "epoch": 1.5136876006441224,
+      "grad_norm": 0.7953918942859293,
+      "learning_rate": 1.39146148244788e-06,
+      "loss": 0.3437,
+      "num_input_tokens_seen": 208662024,
+      "step": 3760,
+      "train_runtime": 22298.0039,
+      "train_tokens_per_second": 9357.879
+    },
+    {
+      "epoch": 1.5157004830917873,
+      "grad_norm": 0.7777212876172017,
+      "learning_rate": 1.3805364828327305e-06,
+      "loss": 0.3299,
+      "num_input_tokens_seen": 208943320,
+      "step": 3765,
+      "train_runtime": 22324.0807,
+      "train_tokens_per_second": 9359.549
+    },
+    {
+      "epoch": 1.5177133655394526,
+      "grad_norm": 0.7623150666705597,
+      "learning_rate": 1.3696476675380722e-06,
+      "loss": 0.3327,
+      "num_input_tokens_seen": 209217648,
+      "step": 3770,
+      "train_runtime": 22357.4149,
+      "train_tokens_per_second": 9357.864
+    },
+    {
+      "epoch": 1.5197262479871174,
+      "grad_norm": 0.7568601746782276,
+      "learning_rate": 1.3587951454210256e-06,
+      "loss": 0.3271,
+      "num_input_tokens_seen": 209517184,
+      "step": 3775,
+      "train_runtime": 22386.6847,
+      "train_tokens_per_second": 9359.009
+    },
+    {
+      "epoch": 1.5217391304347827,
+      "grad_norm": 0.7370857438605858,
+      "learning_rate": 1.3479790249758885e-06,
+      "loss": 0.3213,
+      "num_input_tokens_seen": 209801536,
+      "step": 3780,
+      "train_runtime": 22415.6806,
+      "train_tokens_per_second": 9359.588
+    },
+    {
+      "epoch": 1.5237520128824475,
+      "grad_norm": 0.7058856821143202,
+      "learning_rate": 1.337199414333037e-06,
+      "loss": 0.3166,
+      "num_input_tokens_seen": 210091416,
+      "step": 3785,
+      "train_runtime": 22447.4296,
+      "train_tokens_per_second": 9359.264
+    },
+    {
+      "epoch": 1.5257648953301128,
+      "grad_norm": 0.7767901668042033,
+      "learning_rate": 1.3264564212578613e-06,
+      "loss": 0.3243,
+      "num_input_tokens_seen": 210368136,
+      "step": 3790,
+      "train_runtime": 22472.7891,
+      "train_tokens_per_second": 9361.016
+    },
+    {
+      "epoch": 1.5277777777777777,
+      "grad_norm": 0.7599039346238826,
+      "learning_rate": 1.3157501531496746e-06,
+      "loss": 0.3485,
+      "num_input_tokens_seen": 210669744,
+      "step": 3795,
+      "train_runtime": 22503.9995,
+      "train_tokens_per_second": 9361.436
+    },
+    {
+      "epoch": 1.529790660225443,
+      "grad_norm": 0.7132737750226479,
+      "learning_rate": 1.3050807170406483e-06,
+      "loss": 0.336,
+      "num_input_tokens_seen": 210944912,
+      "step": 3800,
+      "train_runtime": 22530.668,
+      "train_tokens_per_second": 9362.568
+    },
+    {
+      "epoch": 1.5318035426731078,
+      "grad_norm": 0.7583636525631505,
+      "learning_rate": 1.2944482195947384e-06,
+      "loss": 0.3453,
+      "num_input_tokens_seen": 211230096,
+      "step": 3805,
+      "train_runtime": 22561.3939,
+      "train_tokens_per_second": 9362.458
+    },
+    {
+      "epoch": 1.533816425120773,
+      "grad_norm": 0.7841629541575967,
+      "learning_rate": 1.2838527671066198e-06,
+      "loss": 0.3462,
+      "num_input_tokens_seen": 211519208,
+      "step": 3810,
+      "train_runtime": 22594.3602,
+      "train_tokens_per_second": 9361.593
+    },
+    {
+      "epoch": 1.535829307568438,
+      "grad_norm": 0.7890567809065828,
+      "learning_rate": 1.2732944655006203e-06,
+      "loss": 0.335,
+      "num_input_tokens_seen": 211775088,
+      "step": 3815,
+      "train_runtime": 22619.7005,
+      "train_tokens_per_second": 9362.418
+    },
+    {
+      "epoch": 1.537842190016103,
+      "grad_norm": 0.7548051299367953,
+      "learning_rate": 1.2627734203296688e-06,
+      "loss": 0.3482,
+      "num_input_tokens_seen": 212032792,
+      "step": 3820,
+      "train_runtime": 22648.4694,
+      "train_tokens_per_second": 9361.904
+    },
+    {
+      "epoch": 1.539855072463768,
+      "grad_norm": 0.7652241851121122,
+      "learning_rate": 1.2522897367742337e-06,
+      "loss": 0.3465,
+      "num_input_tokens_seen": 212334584,
+      "step": 3825,
+      "train_runtime": 22678.4469,
+      "train_tokens_per_second": 9362.836
+    },
+    {
+      "epoch": 1.5418679549114331,
+      "grad_norm": 0.719029800964307,
+      "learning_rate": 1.2418435196412754e-06,
+      "loss": 0.3228,
+      "num_input_tokens_seen": 212589368,
+      "step": 3830,
+      "train_runtime": 22707.2221,
+      "train_tokens_per_second": 9362.192
+    },
+    {
+      "epoch": 1.5438808373590982,
+      "grad_norm": 0.8211021829253593,
+      "learning_rate": 1.2314348733631958e-06,
+      "loss": 0.3262,
+      "num_input_tokens_seen": 212859664,
+      "step": 3835,
+      "train_runtime": 22733.0631,
+      "train_tokens_per_second": 9363.44
+    },
+    {
+      "epoch": 1.5458937198067633,
+      "grad_norm": 1.5217341999992973,
+      "learning_rate": 1.2210639019967929e-06,
+      "loss": 0.3386,
+      "num_input_tokens_seen": 213157880,
+      "step": 3840,
+      "train_runtime": 22766.5682,
+      "train_tokens_per_second": 9362.759
+    },
+    {
+      "epoch": 1.5479066022544283,
+      "grad_norm": 0.7735934109063405,
+      "learning_rate": 1.2107307092222276e-06,
+      "loss": 0.3254,
+      "num_input_tokens_seen": 213432864,
+      "step": 3845,
+      "train_runtime": 22795.2149,
+      "train_tokens_per_second": 9363.056
+    },
+    {
+      "epoch": 1.5499194847020934,
+      "grad_norm": 0.761887458555926,
+      "learning_rate": 1.2004353983419797e-06,
+      "loss": 0.33,
+      "num_input_tokens_seen": 213735872,
+      "step": 3850,
+      "train_runtime": 22833.5735,
+      "train_tokens_per_second": 9360.597
+    },
+    {
+      "epoch": 1.5519323671497585,
+      "grad_norm": 0.735252078614553,
+      "learning_rate": 1.1901780722798183e-06,
+      "loss": 0.3211,
+      "num_input_tokens_seen": 214051832,
+      "step": 3855,
+      "train_runtime": 22856.0943,
+      "train_tokens_per_second": 9365.197
+    },
+    {
+      "epoch": 1.5539452495974235,
+      "grad_norm": 0.7288543290026211,
+      "learning_rate": 1.1799588335797735e-06,
+      "loss": 0.3384,
+      "num_input_tokens_seen": 214322768,
+      "step": 3860,
+      "train_runtime": 22881.8014,
+      "train_tokens_per_second": 9366.516
+    },
+    {
+      "epoch": 1.5559581320450886,
+      "grad_norm": 0.7429868883033396,
+      "learning_rate": 1.1697777844051105e-06,
+      "loss": 0.3353,
+      "num_input_tokens_seen": 214597800,
+      "step": 3865,
+      "train_runtime": 22908.3321,
+      "train_tokens_per_second": 9367.675
+    },
+    {
+      "epoch": 1.5579710144927537,
+      "grad_norm": 0.7635901853850191,
+      "learning_rate": 1.1596350265373053e-06,
+      "loss": 0.3411,
+      "num_input_tokens_seen": 214891392,
+      "step": 3870,
+      "train_runtime": 22940.9091,
+      "train_tokens_per_second": 9367.17
+    },
+    {
+      "epoch": 1.5599838969404187,
+      "grad_norm": 0.689533131635848,
+      "learning_rate": 1.1495306613750324e-06,
+      "loss": 0.3228,
+      "num_input_tokens_seen": 215175696,
+      "step": 3875,
+      "train_runtime": 22972.3823,
+      "train_tokens_per_second": 9366.712
+    },
+    {
+      "epoch": 1.5619967793880838,
+      "grad_norm": 0.8542015506714442,
+      "learning_rate": 1.1394647899331484e-06,
+      "loss": 0.3163,
+      "num_input_tokens_seen": 215451904,
+      "step": 3880,
+      "train_runtime": 22995.273,
+      "train_tokens_per_second": 9369.4
+    },
+    {
+      "epoch": 1.5640096618357489,
+      "grad_norm": 0.7368396531482924,
+      "learning_rate": 1.129437512841683e-06,
+      "loss": 0.3331,
+      "num_input_tokens_seen": 215733512,
+      "step": 3885,
+      "train_runtime": 23025.7572,
+      "train_tokens_per_second": 9369.226
+    },
+    {
+      "epoch": 1.5660225442834137,
+      "grad_norm": 1.0547431934412872,
+      "learning_rate": 1.1194489303448318e-06,
+      "loss": 0.3652,
+      "num_input_tokens_seen": 215993512,
+      "step": 3890,
+      "train_runtime": 23049.335,
+      "train_tokens_per_second": 9370.922
+    },
+    {
+      "epoch": 1.568035426731079,
+      "grad_norm": 0.7889470165863994,
+      "learning_rate": 1.109499142299953e-06,
+      "loss": 0.3255,
+      "num_input_tokens_seen": 216268800,
+      "step": 3895,
+      "train_runtime": 23077.4613,
+      "train_tokens_per_second": 9371.429
+    },
+    {
+      "epoch": 1.5700483091787438,
+      "grad_norm": 0.8257936511738263,
+      "learning_rate": 1.0995882481765724e-06,
+      "loss": 0.3361,
+      "num_input_tokens_seen": 216575512,
+      "step": 3900,
+      "train_runtime": 23110.0727,
+      "train_tokens_per_second": 9371.477
+    },
+    {
+      "epoch": 1.5720611916264091,
+      "grad_norm": 0.8619751100027543,
+      "learning_rate": 1.089716347055388e-06,
+      "loss": 0.3162,
+      "num_input_tokens_seen": 216837920,
+      "step": 3905,
+      "train_runtime": 23137.165,
+      "train_tokens_per_second": 9371.845
+    },
+    {
+      "epoch": 1.574074074074074,
+      "grad_norm": 0.7176090477456033,
+      "learning_rate": 1.0798835376272782e-06,
+      "loss": 0.33,
+      "num_input_tokens_seen": 217123624,
+      "step": 3910,
+      "train_runtime": 23173.599,
+      "train_tokens_per_second": 9369.439
+    },
+    {
+      "epoch": 1.5760869565217392,
+      "grad_norm": 0.7006677474391395,
+      "learning_rate": 1.070089918192317e-06,
+      "loss": 0.3347,
+      "num_input_tokens_seen": 217391624,
+      "step": 3915,
+      "train_runtime": 23195.6833,
+      "train_tokens_per_second": 9372.072
+    },
+    {
+      "epoch": 1.578099838969404,
+      "grad_norm": 0.7831116819382761,
+      "learning_rate": 1.0603355866587883e-06,
+      "loss": 0.3253,
+      "num_input_tokens_seen": 217662616,
+      "step": 3920,
+      "train_runtime": 23228.258,
+      "train_tokens_per_second": 9370.596
+    },
+    {
+      "epoch": 1.5801127214170694,
+      "grad_norm": 0.7623065283085293,
+      "learning_rate": 1.050620640542208e-06,
+      "loss": 0.3248,
+      "num_input_tokens_seen": 217949272,
+      "step": 3925,
+      "train_runtime": 23265.6696,
+      "train_tokens_per_second": 9367.849
+    },
+    {
+      "epoch": 1.5821256038647342,
+      "grad_norm": 0.8605799970777358,
+      "learning_rate": 1.040945176964353e-06,
+      "loss": 0.3256,
+      "num_input_tokens_seen": 218237752,
+      "step": 3930,
+      "train_runtime": 23296.3997,
+      "train_tokens_per_second": 9367.875
+    },
+    {
+      "epoch": 1.5841384863123995,
+      "grad_norm": 0.741040121519462,
+      "learning_rate": 1.0313092926522865e-06,
+      "loss": 0.3177,
+      "num_input_tokens_seen": 218526008,
+      "step": 3935,
+      "train_runtime": 23325.0281,
+      "train_tokens_per_second": 9368.735
+    },
+    {
+      "epoch": 1.5861513687600644,
+      "grad_norm": 0.7991985835562613,
+      "learning_rate": 1.0217130839373913e-06,
+      "loss": 0.3283,
+      "num_input_tokens_seen": 218801648,
+      "step": 3940,
+      "train_runtime": 23354.206,
+      "train_tokens_per_second": 9368.833
+    },
+    {
+      "epoch": 1.5881642512077294,
+      "grad_norm": 0.6816928221089462,
+      "learning_rate": 1.0121566467544085e-06,
+      "loss": 0.3361,
+      "num_input_tokens_seen": 219080992,
+      "step": 3945,
+      "train_runtime": 23387.3049,
+      "train_tokens_per_second": 9367.518
+    },
+    {
+      "epoch": 1.5901771336553945,
+      "grad_norm": 0.7179563889694311,
+      "learning_rate": 1.0026400766404736e-06,
+      "loss": 0.3334,
+      "num_input_tokens_seen": 219374736,
+      "step": 3950,
+      "train_runtime": 23417.5712,
+      "train_tokens_per_second": 9367.954
+    },
+    {
+      "epoch": 1.5921900161030595,
+      "grad_norm": 0.8030065233999752,
+      "learning_rate": 9.931634687341695e-07,
+      "loss": 0.3289,
+      "num_input_tokens_seen": 219645792,
+      "step": 3955,
+      "train_runtime": 23446.2343,
+      "train_tokens_per_second": 9368.063
+    },
+    {
+      "epoch": 1.5942028985507246,
+      "grad_norm": 0.752344455406048,
+      "learning_rate": 9.837269177745683e-07,
+      "loss": 0.3211,
+      "num_input_tokens_seen": 219942200,
+      "step": 3960,
+      "train_runtime": 23471.9302,
+      "train_tokens_per_second": 9370.435
+    },
+    {
+      "epoch": 1.5962157809983897,
+      "grad_norm": 0.9225543222415034,
+      "learning_rate": 9.743305181002877e-07,
+      "loss": 0.3075,
+      "num_input_tokens_seen": 220214488,
+      "step": 3965,
+      "train_runtime": 23499.7716,
+      "train_tokens_per_second": 9370.92
+    },
+    {
+      "epoch": 1.5982286634460547,
+      "grad_norm": 0.7382625708287228,
+      "learning_rate": 9.64974363648548e-07,
+      "loss": 0.3209,
+      "num_input_tokens_seen": 220469832,
+      "step": 3970,
+      "train_runtime": 23528.556,
+      "train_tokens_per_second": 9370.309
+    },
+    {
+      "epoch": 1.6002415458937198,
+      "grad_norm": 0.9101105571618404,
+      "learning_rate": 9.556585479542285e-07,
+      "loss": 0.3333,
+      "num_input_tokens_seen": 220769464,
+      "step": 3975,
+      "train_runtime": 23556.1692,
+      "train_tokens_per_second": 9372.044
+    },
+    {
+      "epoch": 1.6022544283413849,
+      "grad_norm": 0.746913025499908,
+      "learning_rate": 9.463831641489391e-07,
+      "loss": 0.3282,
+      "num_input_tokens_seen": 221056816,
+      "step": 3980,
+      "train_runtime": 23590.4489,
+      "train_tokens_per_second": 9370.607
+    },
+    {
+      "epoch": 1.60426731078905,
+      "grad_norm": 0.7913120028541547,
+      "learning_rate": 9.371483049600849e-07,
+      "loss": 0.3326,
+      "num_input_tokens_seen": 221336336,
+      "step": 3985,
+      "train_runtime": 23623.142,
+      "train_tokens_per_second": 9369.471
+    },
+    {
+      "epoch": 1.606280193236715,
+      "grad_norm": 0.7885748325954427,
+      "learning_rate": 9.279540627099409e-07,
+      "loss": 0.3372,
+      "num_input_tokens_seen": 221624032,
+      "step": 3990,
+      "train_runtime": 23652.8384,
+      "train_tokens_per_second": 9369.87
+    },
+    {
+      "epoch": 1.60829307568438,
+      "grad_norm": 0.7570465081161281,
+      "learning_rate": 9.188005293147273e-07,
+      "loss": 0.3426,
+      "num_input_tokens_seen": 221924728,
+      "step": 3995,
+      "train_runtime": 23676.2684,
+      "train_tokens_per_second": 9373.298
+    },
+    {
+      "epoch": 1.6103059581320451,
+      "grad_norm": 0.7342840072573682,
+      "learning_rate": 9.096877962836941e-07,
+      "loss": 0.3322,
+      "num_input_tokens_seen": 222209696,
+      "step": 4000,
+      "train_runtime": 23707.3594,
+      "train_tokens_per_second": 9373.026
+    },
+    {
+      "epoch": 1.6123188405797102,
+      "grad_norm": 0.7076717686937853,
+      "learning_rate": 9.006159547182003e-07,
+      "loss": 0.3283,
+      "num_input_tokens_seen": 222501176,
+      "step": 4005,
+      "train_runtime": 23731.7899,
+      "train_tokens_per_second": 9375.659
+    },
+    {
+      "epoch": 1.6143317230273753,
+      "grad_norm": 0.8270895934966705,
+      "learning_rate": 8.915850953108096e-07,
+      "loss": 0.3212,
+      "num_input_tokens_seen": 222774880,
+      "step": 4010,
+      "train_runtime": 23754.7975,
+      "train_tokens_per_second": 9378.101
+    },
+    {
+      "epoch": 1.61634460547504,
+      "grad_norm": 0.7144503451693651,
+      "learning_rate": 8.8259530834438e-07,
+      "loss": 0.3271,
+      "num_input_tokens_seen": 223056112,
+      "step": 4015,
+      "train_runtime": 23787.3871,
+      "train_tokens_per_second": 9377.075
+    },
+    {
+      "epoch": 1.6183574879227054,
+      "grad_norm": 0.7893572554634869,
+      "learning_rate": 8.736466836911622e-07,
+      "loss": 0.3276,
+      "num_input_tokens_seen": 223315064,
+      "step": 4020,
+      "train_runtime": 23811.2754,
+      "train_tokens_per_second": 9378.543
+    },
+    {
+      "epoch": 1.6203703703703702,
+      "grad_norm": 0.711839495636752,
+      "learning_rate": 8.647393108119023e-07,
+      "loss": 0.3239,
+      "num_input_tokens_seen": 223605536,
+      "step": 4025,
+      "train_runtime": 23843.3925,
+      "train_tokens_per_second": 9378.092
+    },
+    {
+      "epoch": 1.6223832528180355,
+      "grad_norm": 0.7633440203784806,
+      "learning_rate": 8.558732787549429e-07,
+      "loss": 0.3307,
+      "num_input_tokens_seen": 223864176,
+      "step": 4030,
+      "train_runtime": 23869.0811,
+      "train_tokens_per_second": 9378.835
+    },
+    {
+      "epoch": 1.6243961352657004,
+      "grad_norm": 0.811672430999867,
+      "learning_rate": 8.470486761553398e-07,
+      "loss": 0.3482,
+      "num_input_tokens_seen": 224156968,
+      "step": 4035,
+      "train_runtime": 23897.5997,
+      "train_tokens_per_second": 9379.895
+    },
+    {
+      "epoch": 1.6264090177133657,
+      "grad_norm": 0.8023225724148835,
+      "learning_rate": 8.38265591233971e-07,
+      "loss": 0.3196,
+      "num_input_tokens_seen": 224410576,
+      "step": 4040,
+      "train_runtime": 23922.3154,
+      "train_tokens_per_second": 9380.805
+    },
+    {
+      "epoch": 1.6284219001610305,
+      "grad_norm": 0.6915544413761353,
+      "learning_rate": 8.29524111796654e-07,
+      "loss": 0.3344,
+      "num_input_tokens_seen": 224700560,
+      "step": 4045,
+      "train_runtime": 23958.959,
+      "train_tokens_per_second": 9378.561
+    },
+    {
+      "epoch": 1.6304347826086958,
+      "grad_norm": 0.710435158096271,
+      "learning_rate": 8.208243252332737e-07,
+      "loss": 0.3346,
+      "num_input_tokens_seen": 224988920,
+      "step": 4050,
+      "train_runtime": 23992.9876,
+      "train_tokens_per_second": 9377.278
+    },
+    {
+      "epoch": 1.6324476650563606,
+      "grad_norm": 0.7805604446381733,
+      "learning_rate": 8.121663185169032e-07,
+      "loss": 0.3399,
+      "num_input_tokens_seen": 225239008,
+      "step": 4055,
+      "train_runtime": 24024.7611,
+      "train_tokens_per_second": 9375.286
+    },
+    {
+      "epoch": 1.634460547504026,
+      "grad_norm": 0.7779017964507423,
+      "learning_rate": 8.035501782029348e-07,
+      "loss": 0.3128,
+      "num_input_tokens_seen": 225516448,
+      "step": 4060,
+      "train_runtime": 24054.9921,
+      "train_tokens_per_second": 9375.037
+    },
+    {
+      "epoch": 1.6364734299516908,
+      "grad_norm": 0.8267859710567748,
+      "learning_rate": 7.949759904282184e-07,
+      "loss": 0.3354,
+      "num_input_tokens_seen": 225807880,
+      "step": 4065,
+      "train_runtime": 24084.9719,
+      "train_tokens_per_second": 9375.468
+    },
+    {
+      "epoch": 1.6384863123993558,
+      "grad_norm": 0.7184112265151104,
+      "learning_rate": 7.864438409101971e-07,
+      "loss": 0.3235,
+      "num_input_tokens_seen": 226078456,
+      "step": 4070,
+      "train_runtime": 24111.754,
+      "train_tokens_per_second": 9376.276
+    },
+    {
+      "epoch": 1.640499194847021,
+      "grad_norm": 0.8079086332350327,
+      "learning_rate": 7.779538149460514e-07,
+      "loss": 0.3356,
+      "num_input_tokens_seen": 226347576,
+      "step": 4075,
+      "train_runtime": 24140.9796,
+      "train_tokens_per_second": 9376.073
+    },
+    {
+      "epoch": 1.642512077294686,
+      "grad_norm": 0.9725163014475132,
+      "learning_rate": 7.695059974118468e-07,
+      "loss": 0.3289,
+      "num_input_tokens_seen": 226625696,
+      "step": 4080,
+      "train_runtime": 24169.1123,
+      "train_tokens_per_second": 9376.666
+    },
+    {
+      "epoch": 1.644524959742351,
+      "grad_norm": 0.7997223733781428,
+      "learning_rate": 7.611004727616833e-07,
+      "loss": 0.3237,
+      "num_input_tokens_seen": 226918856,
+      "step": 4085,
+      "train_runtime": 24205.3541,
+      "train_tokens_per_second": 9374.738
+    },
+    {
+      "epoch": 1.646537842190016,
+      "grad_norm": 0.7437592300271824,
+      "learning_rate": 7.527373250268533e-07,
+      "loss": 0.3288,
+      "num_input_tokens_seen": 227197600,
+      "step": 4090,
+      "train_runtime": 24238.6357,
+      "train_tokens_per_second": 9373.366
+    },
+    {
+      "epoch": 1.6485507246376812,
+      "grad_norm": 0.8926501959554938,
+      "learning_rate": 7.444166378150014e-07,
+      "loss": 0.3396,
+      "num_input_tokens_seen": 227453160,
+      "step": 4095,
+      "train_runtime": 24269.7627,
+      "train_tokens_per_second": 9371.874
+    },
+    {
+      "epoch": 1.6505636070853462,
+      "grad_norm": 0.6945130835700737,
+      "learning_rate": 7.361384943092881e-07,
+      "loss": 0.3122,
+      "num_input_tokens_seen": 227735952,
+      "step": 4100,
+      "train_runtime": 24303.3859,
+      "train_tokens_per_second": 9370.544
+    },
+    {
+      "epoch": 1.6525764895330113,
+      "grad_norm": 0.7874575323702854,
+      "learning_rate": 7.279029772675572e-07,
+      "loss": 0.332,
+      "num_input_tokens_seen": 228024336,
+      "step": 4105,
+      "train_runtime": 24335.0174,
+      "train_tokens_per_second": 9370.215
+    },
+    {
+      "epoch": 1.6545893719806763,
+      "grad_norm": 0.7864716590931821,
+      "learning_rate": 7.197101690215103e-07,
+      "loss": 0.3354,
+      "num_input_tokens_seen": 228295336,
+      "step": 4110,
+      "train_runtime": 24370.1254,
+      "train_tokens_per_second": 9367.836
+    },
+    {
+      "epoch": 1.6566022544283414,
+      "grad_norm": 0.7295846759852129,
+      "learning_rate": 7.115601514758808e-07,
+      "loss": 0.3193,
+      "num_input_tokens_seen": 228565976,
+      "step": 4115,
+      "train_runtime": 24400.2267,
+      "train_tokens_per_second": 9367.371
+    },
+    {
+      "epoch": 1.6586151368760065,
+      "grad_norm": 0.8283187457493462,
+      "learning_rate": 7.034530061076183e-07,
+      "loss": 0.3357,
+      "num_input_tokens_seen": 228823480,
+      "step": 4120,
+      "train_runtime": 24430.4272,
+      "train_tokens_per_second": 9366.332
+    },
+    {
+      "epoch": 1.6606280193236715,
+      "grad_norm": 0.7557175907200036,
+      "learning_rate": 6.953888139650744e-07,
+      "loss": 0.3249,
+      "num_input_tokens_seen": 229111408,
+      "step": 4125,
+      "train_runtime": 24461.0257,
+      "train_tokens_per_second": 9366.386
+    },
+    {
+      "epoch": 1.6626409017713364,
+      "grad_norm": 0.7962931945604196,
+      "learning_rate": 6.873676556671882e-07,
+      "loss": 0.3366,
+      "num_input_tokens_seen": 229354968,
+      "step": 4130,
+      "train_runtime": 24491.8735,
+      "train_tokens_per_second": 9364.533
+    },
+    {
+      "epoch": 1.6646537842190017,
+      "grad_norm": 0.7502386473995064,
+      "learning_rate": 6.79389611402686e-07,
+      "loss": 0.2955,
+      "num_input_tokens_seen": 229622848,
+      "step": 4135,
+      "train_runtime": 24523.2159,
+      "train_tokens_per_second": 9363.488
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.8854707969637451,
+      "learning_rate": 6.714547609292715e-07,
+      "loss": 0.3385,
+      "num_input_tokens_seen": 229881872,
+      "step": 4140,
+      "train_runtime": 24553.1854,
+      "train_tokens_per_second": 9362.609
+    },
+    {
+      "epoch": 1.6686795491143318,
+      "grad_norm": 0.7978393640599302,
+      "learning_rate": 6.63563183572839e-07,
+      "loss": 0.3363,
+      "num_input_tokens_seen": 230150944,
+      "step": 4145,
+      "train_runtime": 24582.8002,
+      "train_tokens_per_second": 9362.275
+    },
+    {
+      "epoch": 1.6706924315619966,
+      "grad_norm": 0.7846940236044779,
+      "learning_rate": 6.557149582266726e-07,
+      "loss": 0.3226,
+      "num_input_tokens_seen": 230404984,
+      "step": 4150,
+      "train_runtime": 24609.5429,
+      "train_tokens_per_second": 9362.424
+    },
+    {
+      "epoch": 1.672705314009662,
+      "grad_norm": 0.8111523832897525,
+      "learning_rate": 6.479101633506596e-07,
+      "loss": 0.3501,
+      "num_input_tokens_seen": 230671544,
+      "step": 4155,
+      "train_runtime": 24638.5253,
+      "train_tokens_per_second": 9362.23
+    },
+    {
+      "epoch": 1.6747181964573268,
+      "grad_norm": 0.7056322889636739,
+      "learning_rate": 6.401488769705083e-07,
+      "loss": 0.3308,
+      "num_input_tokens_seen": 230970448,
+      "step": 4160,
+      "train_runtime": 24671.0992,
+      "train_tokens_per_second": 9361.985
+    },
+    {
+      "epoch": 1.676731078904992,
+      "grad_norm": 0.737347905742684,
+      "learning_rate": 6.324311766769631e-07,
+      "loss": 0.3248,
+      "num_input_tokens_seen": 231258312,
+      "step": 4165,
+      "train_runtime": 24697.8594,
+      "train_tokens_per_second": 9363.496
+    },
+    {
+      "epoch": 1.678743961352657,
+      "grad_norm": 0.788657901267551,
+      "learning_rate": 6.247571396250318e-07,
+      "loss": 0.3268,
+      "num_input_tokens_seen": 231531768,
+      "step": 4170,
+      "train_runtime": 24723.4168,
+      "train_tokens_per_second": 9364.877
+    },
+    {
+      "epoch": 1.6807568438003222,
+      "grad_norm": 0.8480722380499536,
+      "learning_rate": 6.171268425332178e-07,
+      "loss": 0.3205,
+      "num_input_tokens_seen": 231824216,
+      "step": 4175,
+      "train_runtime": 24752.6175,
+      "train_tokens_per_second": 9365.644
+    },
+    {
+      "epoch": 1.682769726247987,
+      "grad_norm": 0.8533528991884664,
+      "learning_rate": 6.095403616827467e-07,
+      "loss": 0.3325,
+      "num_input_tokens_seen": 232086976,
+      "step": 4180,
+      "train_runtime": 24783.6916,
+      "train_tokens_per_second": 9364.504
+    },
+    {
+      "epoch": 1.6847826086956523,
+      "grad_norm": 0.7607851188345848,
+      "learning_rate": 6.01997772916807e-07,
+      "loss": 0.3303,
+      "num_input_tokens_seen": 232356456,
+      "step": 4185,
+      "train_runtime": 24815.3282,
+      "train_tokens_per_second": 9363.425
+    },
+    {
+      "epoch": 1.6867954911433172,
+      "grad_norm": 0.7673057889224089,
+      "learning_rate": 5.944991516397924e-07,
+      "loss": 0.3449,
+      "num_input_tokens_seen": 232613080,
+      "step": 4190,
+      "train_runtime": 24851.1567,
+      "train_tokens_per_second": 9360.252
+    },
+    {
+      "epoch": 1.6888083735909822,
+      "grad_norm": 0.762852969022318,
+      "learning_rate": 5.870445728165447e-07,
+      "loss": 0.3337,
+      "num_input_tokens_seen": 232913968,
+      "step": 4195,
+      "train_runtime": 24876.6978,
+      "train_tokens_per_second": 9362.737
+    },
+    {
+      "epoch": 1.6908212560386473,
+      "grad_norm": 0.7199577692805973,
+      "learning_rate": 5.796341109716097e-07,
+      "loss": 0.3187,
+      "num_input_tokens_seen": 233215016,
+      "step": 4200,
+      "train_runtime": 24904.8894,
+      "train_tokens_per_second": 9364.226
+    },
+    {
+      "epoch": 1.6928341384863124,
+      "grad_norm": 0.803062519650011,
+      "learning_rate": 5.722678401884868e-07,
+      "loss": 0.3306,
+      "num_input_tokens_seen": 233517040,
+      "step": 4205,
+      "train_runtime": 24934.5071,
+      "train_tokens_per_second": 9365.216
+    },
+    {
+      "epoch": 1.6948470209339774,
+      "grad_norm": 0.7772987135840412,
+      "learning_rate": 5.649458341088915e-07,
+      "loss": 0.3318,
+      "num_input_tokens_seen": 233795160,
+      "step": 4210,
+      "train_runtime": 24959.9535,
+      "train_tokens_per_second": 9366.811
+    },
+    {
+      "epoch": 1.6968599033816425,
+      "grad_norm": 0.6771339695865659,
+      "learning_rate": 5.576681659320199e-07,
+      "loss": 0.299,
+      "num_input_tokens_seen": 234065480,
+      "step": 4215,
+      "train_runtime": 24990.24,
+      "train_tokens_per_second": 9366.276
+    },
+    {
+      "epoch": 1.6988727858293076,
+      "grad_norm": 0.8029014126940806,
+      "learning_rate": 5.504349084138122e-07,
+      "loss": 0.3062,
+      "num_input_tokens_seen": 234341016,
+      "step": 4220,
+      "train_runtime": 25025.2579,
+      "train_tokens_per_second": 9364.18
+    },
+    {
+      "epoch": 1.7008856682769726,
+      "grad_norm": 0.6798104689313403,
+      "learning_rate": 5.43246133866231e-07,
+      "loss": 0.3353,
+      "num_input_tokens_seen": 234623672,
+      "step": 4225,
+      "train_runtime": 25054.3876,
+      "train_tokens_per_second": 9364.574
+    },
+    {
+      "epoch": 1.7028985507246377,
+      "grad_norm": 0.7122713964574555,
+      "learning_rate": 5.361019141565355e-07,
+      "loss": 0.348,
+      "num_input_tokens_seen": 234924992,
+      "step": 4230,
+      "train_runtime": 25082.1009,
+      "train_tokens_per_second": 9366.241
+    },
+    {
+      "epoch": 1.7049114331723028,
+      "grad_norm": 0.7422047457584408,
+      "learning_rate": 5.290023207065631e-07,
+      "loss": 0.319,
+      "num_input_tokens_seen": 235200872,
+      "step": 4235,
+      "train_runtime": 25113.5089,
+      "train_tokens_per_second": 9365.512
+    },
+    {
+      "epoch": 1.7069243156199678,
+      "grad_norm": 0.7167271366349289,
+      "learning_rate": 5.219474244920164e-07,
+      "loss": 0.3158,
+      "num_input_tokens_seen": 235501128,
+      "step": 4240,
+      "train_runtime": 25139.08,
+      "train_tokens_per_second": 9367.929
+    },
+    {
+      "epoch": 1.708937198067633,
+      "grad_norm": 0.7811185266673254,
+      "learning_rate": 5.149372960417537e-07,
+      "loss": 0.3312,
+      "num_input_tokens_seen": 235765184,
+      "step": 4245,
+      "train_runtime": 25175.6394,
+      "train_tokens_per_second": 9364.814
+    },
+    {
+      "epoch": 1.710950080515298,
+      "grad_norm": 0.8134374020183804,
+      "learning_rate": 5.07972005437079e-07,
+      "loss": 0.3168,
+      "num_input_tokens_seen": 236038880,
+      "step": 4250,
+      "train_runtime": 25205.7441,
+      "train_tokens_per_second": 9364.488
+    },
+    {
+      "epoch": 1.7129629629629628,
+      "grad_norm": 0.825576722770212,
+      "learning_rate": 5.010516223110512e-07,
+      "loss": 0.3157,
+      "num_input_tokens_seen": 236312816,
+      "step": 4255,
+      "train_runtime": 25235.9213,
+      "train_tokens_per_second": 9364.145
+    },
+    {
+      "epoch": 1.714975845410628,
+      "grad_norm": 0.8483761347452623,
+      "learning_rate": 4.941762158477798e-07,
+      "loss": 0.3085,
+      "num_input_tokens_seen": 236568240,
+      "step": 4260,
+      "train_runtime": 25270.095,
+      "train_tokens_per_second": 9361.589
+    },
+    {
+      "epoch": 1.716988727858293,
+      "grad_norm": 0.7112690478984135,
+      "learning_rate": 4.87345854781735e-07,
+      "loss": 0.3295,
+      "num_input_tokens_seen": 236857840,
+      "step": 4265,
+      "train_runtime": 25298.6455,
+      "train_tokens_per_second": 9362.471
+    },
+    {
+      "epoch": 1.7190016103059582,
+      "grad_norm": 0.7985569938095428,
+      "learning_rate": 4.805606073970642e-07,
+      "loss": 0.3152,
+      "num_input_tokens_seen": 237112384,
+      "step": 4270,
+      "train_runtime": 25326.2616,
+      "train_tokens_per_second": 9362.313
+    },
+    {
+      "epoch": 1.721014492753623,
+      "grad_norm": 0.8956824502085036,
+      "learning_rate": 4.7382054152690314e-07,
+      "loss": 0.3575,
+      "num_input_tokens_seen": 237393616,
+      "step": 4275,
+      "train_runtime": 25352.6837,
+      "train_tokens_per_second": 9363.648
+    },
+    {
+      "epoch": 1.7230273752012883,
+      "grad_norm": 0.7210994760039046,
+      "learning_rate": 4.671257245527028e-07,
+      "loss": 0.3518,
+      "num_input_tokens_seen": 237692784,
+      "step": 4280,
+      "train_runtime": 25382.1151,
+      "train_tokens_per_second": 9364.577
+    },
+    {
+      "epoch": 1.7250402576489532,
+      "grad_norm": 0.7294071065949437,
+      "learning_rate": 4.604762234035548e-07,
+      "loss": 0.326,
+      "num_input_tokens_seen": 237983312,
+      "step": 4285,
+      "train_runtime": 25415.2627,
+      "train_tokens_per_second": 9363.795
+    },
+    {
+      "epoch": 1.7270531400966185,
+      "grad_norm": 0.7270119920970302,
+      "learning_rate": 4.538721045555222e-07,
+      "loss": 0.3273,
+      "num_input_tokens_seen": 238255928,
+      "step": 4290,
+      "train_runtime": 25446.5974,
+      "train_tokens_per_second": 9362.978
+    },
+    {
+      "epoch": 1.7290660225442833,
+      "grad_norm": 0.7671198512818569,
+      "learning_rate": 4.473134340309704e-07,
+      "loss": 0.3142,
+      "num_input_tokens_seen": 238523560,
+      "step": 4295,
+      "train_runtime": 25469.2074,
+      "train_tokens_per_second": 9365.174
+    },
+    {
+      "epoch": 1.7310789049919486,
+      "grad_norm": 0.8277465101806594,
+      "learning_rate": 4.408002773979164e-07,
+      "loss": 0.3126,
+      "num_input_tokens_seen": 238761784,
+      "step": 4300,
+      "train_runtime": 25498.9919,
+      "train_tokens_per_second": 9363.577
+    },
+    {
+      "epoch": 1.7330917874396135,
+      "grad_norm": 0.7988799633027643,
+      "learning_rate": 4.343326997693637e-07,
+      "loss": 0.3103,
+      "num_input_tokens_seen": 239045784,
+      "step": 4305,
+      "train_runtime": 25528.3495,
+      "train_tokens_per_second": 9363.934
+    },
+    {
+      "epoch": 1.7351046698872787,
+      "grad_norm": 0.7264979355934109,
+      "learning_rate": 4.279107658026582e-07,
+      "loss": 0.3187,
+      "num_input_tokens_seen": 239322384,
+      "step": 4310,
+      "train_runtime": 25564.717,
+      "train_tokens_per_second": 9361.433
+    },
+    {
+      "epoch": 1.7371175523349436,
+      "grad_norm": 0.8742452116086884,
+      "learning_rate": 4.2153453969883796e-07,
+      "loss": 0.3283,
+      "num_input_tokens_seen": 239575784,
+      "step": 4315,
+      "train_runtime": 25596.0519,
+      "train_tokens_per_second": 9359.873
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 0.7976387729789014,
+      "learning_rate": 4.1520408520199397e-07,
+      "loss": 0.3374,
+      "num_input_tokens_seen": 239857624,
+      "step": 4320,
+      "train_runtime": 25625.444,
+      "train_tokens_per_second": 9360.135
+    },
+    {
+      "epoch": 1.7411433172302737,
+      "grad_norm": 0.7589841183500853,
+      "learning_rate": 4.089194655986306e-07,
+      "loss": 0.3162,
+      "num_input_tokens_seen": 240148600,
+      "step": 4325,
+      "train_runtime": 25650.8597,
+      "train_tokens_per_second": 9362.205
+    },
+    {
+      "epoch": 1.7431561996779388,
+      "grad_norm": 0.7743152167346375,
+      "learning_rate": 4.0268074371703324e-07,
+      "loss": 0.3197,
+      "num_input_tokens_seen": 240414752,
+      "step": 4330,
+      "train_runtime": 25678.7774,
+      "train_tokens_per_second": 9362.391
+    },
+    {
+      "epoch": 1.7451690821256038,
+      "grad_norm": 0.9041200400518119,
+      "learning_rate": 3.964879819266426e-07,
+      "loss": 0.3329,
+      "num_input_tokens_seen": 240660528,
+      "step": 4335,
+      "train_runtime": 25709.5318,
+      "train_tokens_per_second": 9360.751
+    },
+    {
+      "epoch": 1.747181964573269,
+      "grad_norm": 0.775352752538963,
+      "learning_rate": 3.9034124213742807e-07,
+      "loss": 0.3115,
+      "num_input_tokens_seen": 240951984,
+      "step": 4340,
+      "train_runtime": 25735.9577,
+      "train_tokens_per_second": 9362.464
+    },
+    {
+      "epoch": 1.749194847020934,
+      "grad_norm": 0.7252894822269488,
+      "learning_rate": 3.8424058579927147e-07,
+      "loss": 0.3121,
+      "num_input_tokens_seen": 241224008,
+      "step": 4345,
+      "train_runtime": 25766.639,
+      "train_tokens_per_second": 9361.873
+    },
+    {
+      "epoch": 1.751207729468599,
+      "grad_norm": 0.7781072199995613,
+      "learning_rate": 3.7818607390134933e-07,
+      "loss": 0.3097,
+      "num_input_tokens_seen": 241507016,
+      "step": 4350,
+      "train_runtime": 25801.0511,
+      "train_tokens_per_second": 9360.356
+    },
+    {
+      "epoch": 1.753220611916264,
+      "grad_norm": 0.7287518109243901,
+      "learning_rate": 3.721777669715287e-07,
+      "loss": 0.3327,
+      "num_input_tokens_seen": 241789352,
+      "step": 4355,
+      "train_runtime": 25826.2713,
+      "train_tokens_per_second": 9362.147
+    },
+    {
+      "epoch": 1.7552334943639292,
+      "grad_norm": 0.6893910188650302,
+      "learning_rate": 3.662157250757542e-07,
+      "loss": 0.3182,
+      "num_input_tokens_seen": 242073264,
+      "step": 4360,
+      "train_runtime": 25855.3411,
+      "train_tokens_per_second": 9362.602
+    },
+    {
+      "epoch": 1.7572463768115942,
+      "grad_norm": 0.8180772525625999,
+      "learning_rate": 3.6030000781745513e-07,
+      "loss": 0.3143,
+      "num_input_tokens_seen": 242347328,
+      "step": 4365,
+      "train_runtime": 25884.5006,
+      "train_tokens_per_second": 9362.643
+    },
+    {
+      "epoch": 1.7592592592592593,
+      "grad_norm": 0.756560782407874,
+      "learning_rate": 3.5443067433694525e-07,
+      "loss": 0.3052,
+      "num_input_tokens_seen": 242613888,
+      "step": 4370,
+      "train_runtime": 25918.3472,
+      "train_tokens_per_second": 9360.701
+    },
+    {
+      "epoch": 1.7612721417069244,
+      "grad_norm": 0.9345320976260738,
+      "learning_rate": 3.486077833108342e-07,
+      "loss": 0.3258,
+      "num_input_tokens_seen": 242873240,
+      "step": 4375,
+      "train_runtime": 25951.0981,
+      "train_tokens_per_second": 9358.881
+    },
+    {
+      "epoch": 1.7632850241545892,
+      "grad_norm": 0.8335245931242411,
+      "learning_rate": 3.428313929514376e-07,
+      "loss": 0.3296,
+      "num_input_tokens_seen": 243138080,
+      "step": 4380,
+      "train_runtime": 25982.8517,
+      "train_tokens_per_second": 9357.636
+    },
+    {
+      "epoch": 1.7652979066022545,
+      "grad_norm": 0.8637489681174857,
+      "learning_rate": 3.371015610061973e-07,
+      "loss": 0.3167,
+      "num_input_tokens_seen": 243439728,
+      "step": 4385,
+      "train_runtime": 26014.8498,
+      "train_tokens_per_second": 9357.722
+    },
+    {
+      "epoch": 1.7673107890499193,
+      "grad_norm": 0.7563371461839502,
+      "learning_rate": 3.31418344757104e-07,
+      "loss": 0.3228,
+      "num_input_tokens_seen": 243711576,
+      "step": 4390,
+      "train_runtime": 26043.9886,
+      "train_tokens_per_second": 9357.69
+    },
+    {
+      "epoch": 1.7693236714975846,
+      "grad_norm": 0.8472172338384234,
+      "learning_rate": 3.2578180102012414e-07,
+      "loss": 0.3347,
+      "num_input_tokens_seen": 243969600,
+      "step": 4395,
+      "train_runtime": 26071.6761,
+      "train_tokens_per_second": 9357.649
+    },
+    {
+      "epoch": 1.7713365539452495,
+      "grad_norm": 0.6857594083337077,
+      "learning_rate": 3.2019198614463285e-07,
+      "loss": 0.3305,
+      "num_input_tokens_seen": 244257824,
+      "step": 4400,
+      "train_runtime": 26105.413,
+      "train_tokens_per_second": 9356.597
+    },
+    {
+      "epoch": 1.7733494363929148,
+      "grad_norm": 0.7729840343779738,
+      "learning_rate": 3.146489560128496e-07,
+      "loss": 0.3511,
+      "num_input_tokens_seen": 244538784,
+      "step": 4405,
+      "train_runtime": 26131.8865,
+      "train_tokens_per_second": 9357.87
+    },
+    {
+      "epoch": 1.7753623188405796,
+      "grad_norm": 0.792021018194511,
+      "learning_rate": 3.091527660392779e-07,
+      "loss": 0.3185,
+      "num_input_tokens_seen": 244812552,
+      "step": 4410,
+      "train_runtime": 26157.6284,
+      "train_tokens_per_second": 9359.126
+    },
+    {
+      "epoch": 1.7773752012882449,
+      "grad_norm": 0.7090501983680757,
+      "learning_rate": 3.037034711701553e-07,
+      "loss": 0.3341,
+      "num_input_tokens_seen": 245105832,
+      "step": 4415,
+      "train_runtime": 26186.2217,
+      "train_tokens_per_second": 9360.107
+    },
+    {
+      "epoch": 1.7793880837359097,
+      "grad_norm": 0.7710194420665034,
+      "learning_rate": 2.983011258829005e-07,
+      "loss": 0.3258,
+      "num_input_tokens_seen": 245393776,
+      "step": 4420,
+      "train_runtime": 26223.2928,
+      "train_tokens_per_second": 9357.855
+    },
+    {
+      "epoch": 1.781400966183575,
+      "grad_norm": 0.7042050812538667,
+      "learning_rate": 2.929457841855715e-07,
+      "loss": 0.3164,
+      "num_input_tokens_seen": 245668688,
+      "step": 4425,
+      "train_runtime": 26256.6761,
+      "train_tokens_per_second": 9356.428
+    },
+    {
+      "epoch": 1.7834138486312399,
+      "grad_norm": 0.8907120980114343,
+      "learning_rate": 2.876374996163239e-07,
+      "loss": 0.3247,
+      "num_input_tokens_seen": 245916384,
+      "step": 4430,
+      "train_runtime": 26284.6078,
+      "train_tokens_per_second": 9355.908
+    },
+    {
+      "epoch": 1.7854267310789051,
+      "grad_norm": 0.7629486959102406,
+      "learning_rate": 2.823763252428757e-07,
+      "loss": 0.3312,
+      "num_input_tokens_seen": 246190864,
+      "step": 4435,
+      "train_runtime": 26319.5729,
+      "train_tokens_per_second": 9353.908
+    },
+    {
+      "epoch": 1.78743961352657,
+      "grad_norm": 0.783065145594715,
+      "learning_rate": 2.7716231366197663e-07,
+      "loss": 0.3166,
+      "num_input_tokens_seen": 246467664,
+      "step": 4440,
+      "train_runtime": 26344.8992,
+      "train_tokens_per_second": 9355.423
+    },
+    {
+      "epoch": 1.789452495974235,
+      "grad_norm": 0.7919972725500392,
+      "learning_rate": 2.719955169988847e-07,
+      "loss": 0.3282,
+      "num_input_tokens_seen": 246710208,
+      "step": 4445,
+      "train_runtime": 26375.7013,
+      "train_tokens_per_second": 9353.693
+    },
+    {
+      "epoch": 1.7914653784219001,
+      "grad_norm": 0.7837608209075699,
+      "learning_rate": 2.668759869068421e-07,
+      "loss": 0.332,
+      "num_input_tokens_seen": 246977928,
+      "step": 4450,
+      "train_runtime": 26406.0738,
+      "train_tokens_per_second": 9353.073
+    },
+    {
+      "epoch": 1.7934782608695652,
+      "grad_norm": 0.7669721045273009,
+      "learning_rate": 2.6180377456656023e-07,
+      "loss": 0.3244,
+      "num_input_tokens_seen": 247269376,
+      "step": 4455,
+      "train_runtime": 26437.0955,
+      "train_tokens_per_second": 9353.122
+    },
+    {
+      "epoch": 1.7954911433172303,
+      "grad_norm": 0.7653740355371631,
+      "learning_rate": 2.5677893068570767e-07,
+      "loss": 0.3083,
+      "num_input_tokens_seen": 247539224,
+      "step": 4460,
+      "train_runtime": 26467.6314,
+      "train_tokens_per_second": 9352.527
+    },
+    {
+      "epoch": 1.7975040257648953,
+      "grad_norm": 0.7999194458492196,
+      "learning_rate": 2.518015054984041e-07,
+      "loss": 0.3287,
+      "num_input_tokens_seen": 247826880,
+      "step": 4465,
+      "train_runtime": 26493.2771,
+      "train_tokens_per_second": 9354.331
+    },
+    {
+      "epoch": 1.7995169082125604,
+      "grad_norm": 0.7072665174428415,
+      "learning_rate": 2.468715487647155e-07,
+      "loss": 0.3203,
+      "num_input_tokens_seen": 248099328,
+      "step": 4470,
+      "train_runtime": 26518.3156,
+      "train_tokens_per_second": 9355.773
+    },
+    {
+      "epoch": 1.8015297906602254,
+      "grad_norm": 0.6917605648044121,
+      "learning_rate": 2.41989109770161e-07,
+      "loss": 0.3274,
+      "num_input_tokens_seen": 248403448,
+      "step": 4475,
+      "train_runtime": 26544.782,
+      "train_tokens_per_second": 9357.901
+    },
+    {
+      "epoch": 1.8035426731078905,
+      "grad_norm": 0.7644034936421807,
+      "learning_rate": 2.371542373252156e-07,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 248654000,
+      "step": 4480,
+      "train_runtime": 26568.4749,
+      "train_tokens_per_second": 9358.987
+    },
+    {
+      "epoch": 1.8055555555555556,
+      "grad_norm": 0.8433107297681439,
+      "learning_rate": 2.3236697976482703e-07,
+      "loss": 0.3243,
+      "num_input_tokens_seen": 248948704,
+      "step": 4485,
+      "train_runtime": 26601.5745,
+      "train_tokens_per_second": 9358.42
+    },
+    {
+      "epoch": 1.8075684380032206,
+      "grad_norm": 0.8133515975979622,
+      "learning_rate": 2.276273849479277e-07,
+      "loss": 0.3363,
+      "num_input_tokens_seen": 249209968,
+      "step": 4490,
+      "train_runtime": 26629.0549,
+      "train_tokens_per_second": 9358.574
+    },
+    {
+      "epoch": 1.8095813204508857,
+      "grad_norm": 0.6765546275506685,
+      "learning_rate": 2.2293550025695776e-07,
+      "loss": 0.3168,
+      "num_input_tokens_seen": 249486336,
+      "step": 4495,
+      "train_runtime": 26659.546,
+      "train_tokens_per_second": 9358.236
+    },
+    {
+      "epoch": 1.8115942028985508,
+      "grad_norm": 0.6870170498142121,
+      "learning_rate": 2.1829137259739453e-07,
+      "loss": 0.354,
+      "num_input_tokens_seen": 249789232,
+      "step": 4500,
+      "train_runtime": 26691.6787,
+      "train_tokens_per_second": 9358.319
+    },
+    {
+      "epoch": 1.8136070853462156,
+      "grad_norm": 0.8413378192941108,
+      "learning_rate": 2.136950483972794e-07,
+      "loss": 0.3329,
+      "num_input_tokens_seen": 250062680,
+      "step": 4505,
+      "train_runtime": 26726.0645,
+      "train_tokens_per_second": 9356.51
+    },
+    {
+      "epoch": 1.815619967793881,
+      "grad_norm": 0.7134073687499528,
+      "learning_rate": 2.0914657360675606e-07,
+      "loss": 0.3381,
+      "num_input_tokens_seen": 250323856,
+      "step": 4510,
+      "train_runtime": 26758.7151,
+      "train_tokens_per_second": 9354.853
+    },
+    {
+      "epoch": 1.8176328502415457,
+      "grad_norm": 0.6941928260181837,
+      "learning_rate": 2.046459936976103e-07,
+      "loss": 0.3317,
+      "num_input_tokens_seen": 250632152,
+      "step": 4515,
+      "train_runtime": 26783.1073,
+      "train_tokens_per_second": 9357.844
+    },
+    {
+      "epoch": 1.819645732689211,
+      "grad_norm": 0.8437674526866175,
+      "learning_rate": 2.0019335366281533e-07,
+      "loss": 0.3124,
+      "num_input_tokens_seen": 250923760,
+      "step": 4520,
+      "train_runtime": 26813.9213,
+      "train_tokens_per_second": 9357.966
+    },
+    {
+      "epoch": 1.8216586151368759,
+      "grad_norm": 0.7780396491976735,
+      "learning_rate": 1.9578869801608168e-07,
+      "loss": 0.3024,
+      "num_input_tokens_seen": 251214576,
+      "step": 4525,
+      "train_runtime": 26845.0328,
+      "train_tokens_per_second": 9357.954
+    },
+    {
+      "epoch": 1.8236714975845412,
+      "grad_norm": 0.7144185909467665,
+      "learning_rate": 1.9143207079141523e-07,
+      "loss": 0.3285,
+      "num_input_tokens_seen": 251498880,
+      "step": 4530,
+      "train_runtime": 26881.5086,
+      "train_tokens_per_second": 9355.832
+    },
+    {
+      "epoch": 1.825684380032206,
+      "grad_norm": 1.027627743075786,
+      "learning_rate": 1.8712351554267206e-07,
+      "loss": 0.3196,
+      "num_input_tokens_seen": 251758600,
+      "step": 4535,
+      "train_runtime": 26907.9316,
+      "train_tokens_per_second": 9356.297
+    },
+    {
+      "epoch": 1.8276972624798713,
+      "grad_norm": 1.3189239086881603,
+      "learning_rate": 1.8286307534312542e-07,
+      "loss": 0.3165,
+      "num_input_tokens_seen": 252052760,
+      "step": 4540,
+      "train_runtime": 26938.4763,
+      "train_tokens_per_second": 9356.608
+    },
+    {
+      "epoch": 1.8297101449275361,
+      "grad_norm": 0.7898025504989091,
+      "learning_rate": 1.7865079278503727e-07,
+      "loss": 0.3352,
+      "num_input_tokens_seen": 252319768,
+      "step": 4545,
+      "train_runtime": 26972.4027,
+      "train_tokens_per_second": 9354.738
+    },
+    {
+      "epoch": 1.8317230273752014,
+      "grad_norm": 0.8199840128420028,
+      "learning_rate": 1.74486709979228e-07,
+      "loss": 0.3314,
+      "num_input_tokens_seen": 252613568,
+      "step": 4550,
+      "train_runtime": 27005.4364,
+      "train_tokens_per_second": 9354.175
+    },
+    {
+      "epoch": 1.8337359098228663,
+      "grad_norm": 0.7905130187587479,
+      "learning_rate": 1.7037086855465902e-07,
+      "loss": 0.3308,
+      "num_input_tokens_seen": 252891568,
+      "step": 4555,
+      "train_runtime": 27036.0901,
+      "train_tokens_per_second": 9353.851
+    },
+    {
+      "epoch": 1.8357487922705316,
+      "grad_norm": 0.7492204565397131,
+      "learning_rate": 1.6630330965801578e-07,
+      "loss": 0.3132,
+      "num_input_tokens_seen": 253180104,
+      "step": 4560,
+      "train_runtime": 27061.3951,
+      "train_tokens_per_second": 9355.767
+    },
+    {
+      "epoch": 1.8377616747181964,
+      "grad_norm": 0.8606625928655156,
+      "learning_rate": 1.6228407395329548e-07,
+      "loss": 0.3253,
+      "num_input_tokens_seen": 253471976,
+      "step": 4565,
+      "train_runtime": 27091.5671,
+      "train_tokens_per_second": 9356.121
+    },
+    {
+      "epoch": 1.8397745571658615,
+      "grad_norm": 0.750001287077284,
+      "learning_rate": 1.5831320162140172e-07,
+      "loss": 0.3179,
+      "num_input_tokens_seen": 253767792,
+      "step": 4570,
+      "train_runtime": 27113.6784,
+      "train_tokens_per_second": 9359.401
+    },
+    {
+      "epoch": 1.8417874396135265,
+      "grad_norm": 0.7172785462775401,
+      "learning_rate": 1.54390732359741e-07,
+      "loss": 0.3261,
+      "num_input_tokens_seen": 254059848,
+      "step": 4575,
+      "train_runtime": 27146.7045,
+      "train_tokens_per_second": 9358.773
+    },
+    {
+      "epoch": 1.8438003220611916,
+      "grad_norm": 0.7171552626345205,
+      "learning_rate": 1.5051670538182793e-07,
+      "loss": 0.3079,
+      "num_input_tokens_seen": 254334200,
+      "step": 4580,
+      "train_runtime": 27164.6151,
+      "train_tokens_per_second": 9362.702
+    },
+    {
+      "epoch": 1.8458132045088567,
+      "grad_norm": 0.7172156724331626,
+      "learning_rate": 1.4669115941689182e-07,
+      "loss": 0.3138,
+      "num_input_tokens_seen": 254598752,
+      "step": 4585,
+      "train_runtime": 27193.4655,
+      "train_tokens_per_second": 9362.497
+    },
+    {
+      "epoch": 1.8478260869565217,
+      "grad_norm": 0.6924037486981617,
+      "learning_rate": 1.429141327094913e-07,
+      "loss": 0.2999,
+      "num_input_tokens_seen": 254888928,
+      "step": 4590,
+      "train_runtime": 27219.6154,
+      "train_tokens_per_second": 9364.163
+    },
+    {
+      "epoch": 1.8498389694041868,
+      "grad_norm": 0.7750181355651895,
+      "learning_rate": 1.391856630191285e-07,
+      "loss": 0.336,
+      "num_input_tokens_seen": 255168480,
+      "step": 4595,
+      "train_runtime": 27251.1389,
+      "train_tokens_per_second": 9363.59
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 0.6986119708306424,
+      "learning_rate": 1.3550578761987675e-07,
+      "loss": 0.3083,
+      "num_input_tokens_seen": 255438072,
+      "step": 4600,
+      "train_runtime": 27276.9578,
+      "train_tokens_per_second": 9364.61
+    },
+    {
+      "epoch": 1.853864734299517,
+      "grad_norm": 0.8379643594581179,
+      "learning_rate": 1.3187454330000127e-07,
+      "loss": 0.3302,
+      "num_input_tokens_seen": 255733000,
+      "step": 4605,
+      "train_runtime": 27307.633,
+      "train_tokens_per_second": 9364.891
+    },
+    {
+      "epoch": 1.855877616747182,
+      "grad_norm": 0.6793729940600891,
+      "learning_rate": 1.282919663615978e-07,
+      "loss": 0.3355,
+      "num_input_tokens_seen": 255990800,
+      "step": 4610,
+      "train_runtime": 27342.7866,
+      "train_tokens_per_second": 9362.279
+    },
+    {
+      "epoch": 1.857890499194847,
+      "grad_norm": 0.7023924345706357,
+      "learning_rate": 1.2475809262022632e-07,
+      "loss": 0.3115,
+      "num_input_tokens_seen": 256269840,
+      "step": 4615,
+      "train_runtime": 27369.0605,
+      "train_tokens_per_second": 9363.487
+    },
+    {
+      "epoch": 1.8599033816425121,
+      "grad_norm": 0.7714521066519529,
+      "learning_rate": 1.2127295740455348e-07,
+      "loss": 0.3056,
+      "num_input_tokens_seen": 256551224,
+      "step": 4620,
+      "train_runtime": 27400.4216,
+      "train_tokens_per_second": 9363.039
+    },
+    {
+      "epoch": 1.8619162640901772,
+      "grad_norm": 0.7805388958476632,
+      "learning_rate": 1.1783659555600013e-07,
+      "loss": 0.3297,
+      "num_input_tokens_seen": 256821232,
+      "step": 4625,
+      "train_runtime": 27423.9655,
+      "train_tokens_per_second": 9364.847
+    },
+    {
+      "epoch": 1.863929146537842,
+      "grad_norm": 0.7532854692726214,
+      "learning_rate": 1.1444904142838997e-07,
+      "loss": 0.3261,
+      "num_input_tokens_seen": 257102072,
+      "step": 4630,
+      "train_runtime": 27449.8963,
+      "train_tokens_per_second": 9366.231
+    },
+    {
+      "epoch": 1.8659420289855073,
+      "grad_norm": 0.7922669189746588,
+      "learning_rate": 1.1111032888761253e-07,
+      "loss": 0.3321,
+      "num_input_tokens_seen": 257376384,
+      "step": 4635,
+      "train_runtime": 27477.9274,
+      "train_tokens_per_second": 9366.659
+    },
+    {
+      "epoch": 1.8679549114331722,
+      "grad_norm": 0.789017121099722,
+      "learning_rate": 1.0782049131127736e-07,
+      "loss": 0.3354,
+      "num_input_tokens_seen": 257658656,
+      "step": 4640,
+      "train_runtime": 27509.7063,
+      "train_tokens_per_second": 9366.1
+    },
+    {
+      "epoch": 1.8699677938808374,
+      "grad_norm": 0.7657621503780188,
+      "learning_rate": 1.0457956158838545e-07,
+      "loss": 0.3342,
+      "num_input_tokens_seen": 257927704,
+      "step": 4645,
+      "train_runtime": 27539.0691,
+      "train_tokens_per_second": 9365.883
+    },
+    {
+      "epoch": 1.8719806763285023,
+      "grad_norm": 0.8293094341623964,
+      "learning_rate": 1.013875721189983e-07,
+      "loss": 0.3161,
+      "num_input_tokens_seen": 258190216,
+      "step": 4650,
+      "train_runtime": 27563.3578,
+      "train_tokens_per_second": 9367.154
+    },
+    {
+      "epoch": 1.8739935587761676,
+      "grad_norm": 0.7525078900929507,
+      "learning_rate": 9.824455481391381e-08,
+      "loss": 0.3312,
+      "num_input_tokens_seen": 258452600,
+      "step": 4655,
+      "train_runtime": 27585.1669,
+      "train_tokens_per_second": 9369.26
+    },
+    {
+      "epoch": 1.8760064412238324,
+      "grad_norm": 0.7593919977246997,
+      "learning_rate": 9.515054109434819e-08,
+      "loss": 0.31,
+      "num_input_tokens_seen": 258721912,
+      "step": 4660,
+      "train_runtime": 27606.2703,
+      "train_tokens_per_second": 9371.853
+    },
+    {
+      "epoch": 1.8780193236714977,
+      "grad_norm": 0.8045308753706758,
+      "learning_rate": 9.210556189162056e-08,
+      "loss": 0.3236,
+      "num_input_tokens_seen": 258969488,
+      "step": 4665,
+      "train_runtime": 27632.2752,
+      "train_tokens_per_second": 9371.993
+    },
+    {
+      "epoch": 1.8800322061191626,
+      "grad_norm": 0.6752546309278759,
+      "learning_rate": 8.910964764684671e-08,
+      "loss": 0.3096,
+      "num_input_tokens_seen": 259258480,
+      "step": 4670,
+      "train_runtime": 27668.2561,
+      "train_tokens_per_second": 9370.25
+    },
+    {
+      "epoch": 1.8820450885668278,
+      "grad_norm": 0.6975824885561981,
+      "learning_rate": 8.616282831063083e-08,
+      "loss": 0.3162,
+      "num_input_tokens_seen": 259548968,
+      "step": 4675,
+      "train_runtime": 27695.7658,
+      "train_tokens_per_second": 9371.431
+    },
+    {
+      "epoch": 1.8840579710144927,
+      "grad_norm": 0.7960122008845532,
+      "learning_rate": 8.326513334276864e-08,
+      "loss": 0.2966,
+      "num_input_tokens_seen": 259821040,
+      "step": 4680,
+      "train_runtime": 27728.2066,
+      "train_tokens_per_second": 9370.279
+    },
+    {
+      "epoch": 1.886070853462158,
+      "grad_norm": 0.7670827810838102,
+      "learning_rate": 8.041659171195149e-08,
+      "loss": 0.3256,
+      "num_input_tokens_seen": 260064200,
+      "step": 4685,
+      "train_runtime": 27751.4212,
+      "train_tokens_per_second": 9371.203
+    },
+    {
+      "epoch": 1.8880837359098228,
+      "grad_norm": 0.9089265268812889,
+      "learning_rate": 7.76172318954782e-08,
+      "loss": 0.3395,
+      "num_input_tokens_seen": 260343640,
+      "step": 4690,
+      "train_runtime": 27779.2474,
+      "train_tokens_per_second": 9371.875
+    },
+    {
+      "epoch": 1.8900966183574879,
+      "grad_norm": 0.8002638768101971,
+      "learning_rate": 7.486708187896986e-08,
+      "loss": 0.3217,
+      "num_input_tokens_seen": 260608136,
+      "step": 4695,
+      "train_runtime": 27807.7469,
+      "train_tokens_per_second": 9371.782
+    },
+    {
+      "epoch": 1.892109500805153,
+      "grad_norm": 0.712867722095006,
+      "learning_rate": 7.216616915608876e-08,
+      "loss": 0.3216,
+      "num_input_tokens_seen": 260913592,
+      "step": 4700,
+      "train_runtime": 27842.3744,
+      "train_tokens_per_second": 9371.097
+    },
+    {
+      "epoch": 1.894122383252818,
+      "grad_norm": 0.8452604847702806,
+      "learning_rate": 6.951452072826547e-08,
+      "loss": 0.3356,
+      "num_input_tokens_seen": 261219328,
+      "step": 4705,
+      "train_runtime": 27862.6276,
+      "train_tokens_per_second": 9375.258
+    },
+    {
+      "epoch": 1.896135265700483,
+      "grad_norm": 0.7302148900435511,
+      "learning_rate": 6.691216310442783e-08,
+      "loss": 0.3169,
+      "num_input_tokens_seen": 261484544,
+      "step": 4710,
+      "train_runtime": 27893.2041,
+      "train_tokens_per_second": 9374.489
+    },
+    {
+      "epoch": 1.8981481481481481,
+      "grad_norm": 0.7186849455926788,
+      "learning_rate": 6.435912230073504e-08,
+      "loss": 0.336,
+      "num_input_tokens_seen": 261745496,
+      "step": 4715,
+      "train_runtime": 27928.3183,
+      "train_tokens_per_second": 9372.046
+    },
+    {
+      "epoch": 1.9001610305958132,
+      "grad_norm": 0.7482558219077388,
+      "learning_rate": 6.185542384032018e-08,
+      "loss": 0.315,
+      "num_input_tokens_seen": 262023384,
+      "step": 4720,
+      "train_runtime": 27958.644,
+      "train_tokens_per_second": 9371.82
+    },
+    {
+      "epoch": 1.9021739130434783,
+      "grad_norm": 0.7091459846241603,
+      "learning_rate": 5.9401092753033075e-08,
+      "loss": 0.3182,
+      "num_input_tokens_seen": 262274224,
+      "step": 4725,
+      "train_runtime": 27983.8638,
+      "train_tokens_per_second": 9372.338
+    },
+    {
+      "epoch": 1.9041867954911433,
+      "grad_norm": 0.8149612613401902,
+      "learning_rate": 5.6996153575189504e-08,
+      "loss": 0.3039,
+      "num_input_tokens_seen": 262520656,
+      "step": 4730,
+      "train_runtime": 28012.4822,
+      "train_tokens_per_second": 9371.56
+    },
+    {
+      "epoch": 1.9061996779388084,
+      "grad_norm": 0.8391247525729111,
+      "learning_rate": 5.4640630349329094e-08,
+      "loss": 0.3147,
+      "num_input_tokens_seen": 262787184,
+      "step": 4735,
+      "train_runtime": 28037.4445,
+      "train_tokens_per_second": 9372.722
+    },
+    {
+      "epoch": 1.9082125603864735,
+      "grad_norm": 0.7418230334249278,
+      "learning_rate": 5.233454662396942e-08,
+      "loss": 0.3426,
+      "num_input_tokens_seen": 263085176,
+      "step": 4740,
+      "train_runtime": 28063.1074,
+      "train_tokens_per_second": 9374.77
+    },
+    {
+      "epoch": 1.9102254428341385,
+      "grad_norm": 0.8492850554774655,
+      "learning_rate": 5.0077925453377865e-08,
+      "loss": 0.3203,
+      "num_input_tokens_seen": 263326152,
+      "step": 4745,
+      "train_runtime": 28085.6375,
+      "train_tokens_per_second": 9375.83
+    },
+    {
+      "epoch": 1.9122383252818036,
+      "grad_norm": 0.7391961977926107,
+      "learning_rate": 4.7870789397333475e-08,
+      "loss": 0.3336,
+      "num_input_tokens_seen": 263611048,
+      "step": 4750,
+      "train_runtime": 28114.7313,
+      "train_tokens_per_second": 9376.261
+    },
+    {
+      "epoch": 1.9142512077294684,
+      "grad_norm": 0.6960233088869753,
+      "learning_rate": 4.571316052090824e-08,
+      "loss": 0.3316,
+      "num_input_tokens_seen": 263884712,
+      "step": 4755,
+      "train_runtime": 28143.1565,
+      "train_tokens_per_second": 9376.514
+    },
+    {
+      "epoch": 1.9162640901771337,
+      "grad_norm": 0.8064978583923966,
+      "learning_rate": 4.360506039424173e-08,
+      "loss": 0.3297,
+      "num_input_tokens_seen": 264180136,
+      "step": 4760,
+      "train_runtime": 28174.0679,
+      "train_tokens_per_second": 9376.713
+    },
+    {
+      "epoch": 1.9182769726247986,
+      "grad_norm": 0.7485423616445349,
+      "learning_rate": 4.1546510092327906e-08,
+      "loss": 0.3101,
+      "num_input_tokens_seen": 264471000,
+      "step": 4765,
+      "train_runtime": 28206.4533,
+      "train_tokens_per_second": 9376.259
+    },
+    {
+      "epoch": 1.9202898550724639,
+      "grad_norm": 0.7705284948487796,
+      "learning_rate": 3.953753019480255e-08,
+      "loss": 0.3268,
+      "num_input_tokens_seen": 264745232,
+      "step": 4770,
+      "train_runtime": 28228.7688,
+      "train_tokens_per_second": 9378.561
+    },
+    {
+      "epoch": 1.9223027375201287,
+      "grad_norm": 0.7471363161350103,
+      "learning_rate": 3.757814078574007e-08,
+      "loss": 0.3308,
+      "num_input_tokens_seen": 264997920,
+      "step": 4775,
+      "train_runtime": 28262.0075,
+      "train_tokens_per_second": 9376.472
+    },
+    {
+      "epoch": 1.924315619967794,
+      "grad_norm": 0.7075930106471708,
+      "learning_rate": 3.566836145345032e-08,
+      "loss": 0.3295,
+      "num_input_tokens_seen": 265267704,
+      "step": 4780,
+      "train_runtime": 28294.2476,
+      "train_tokens_per_second": 9375.323
+    },
+    {
+      "epoch": 1.9263285024154588,
+      "grad_norm": 0.8712935865283123,
+      "learning_rate": 3.3808211290284886e-08,
+      "loss": 0.3288,
+      "num_input_tokens_seen": 265527112,
+      "step": 4785,
+      "train_runtime": 28320.9721,
+      "train_tokens_per_second": 9375.636
+    },
+    {
+      "epoch": 1.9283413848631241,
+      "grad_norm": 0.7850508984809887,
+      "learning_rate": 3.199770889244447e-08,
+      "loss": 0.3264,
+      "num_input_tokens_seen": 265786448,
+      "step": 4790,
+      "train_runtime": 28351.0497,
+      "train_tokens_per_second": 9374.836
+    },
+    {
+      "epoch": 1.930354267310789,
+      "grad_norm": 0.8341057086918056,
+      "learning_rate": 3.023687235979233e-08,
+      "loss": 0.3347,
+      "num_input_tokens_seen": 266082088,
+      "step": 4795,
+      "train_runtime": 28387.6001,
+      "train_tokens_per_second": 9373.18
+    },
+    {
+      "epoch": 1.9323671497584543,
+      "grad_norm": 0.727081503990615,
+      "learning_rate": 2.8525719295676712e-08,
+      "loss": 0.3256,
+      "num_input_tokens_seen": 266349272,
+      "step": 4800,
+      "train_runtime": 28411.2924,
+      "train_tokens_per_second": 9374.768
+    },
+    {
+      "epoch": 1.934380032206119,
+      "grad_norm": 0.7445659854163786,
+      "learning_rate": 2.6864266806751472e-08,
+      "loss": 0.3252,
+      "num_input_tokens_seen": 266649048,
+      "step": 4805,
+      "train_runtime": 28434.8264,
+      "train_tokens_per_second": 9377.551
+    },
+    {
+      "epoch": 1.9363929146537844,
+      "grad_norm": 0.6878338494390128,
+      "learning_rate": 2.525253150280682e-08,
+      "loss": 0.3257,
+      "num_input_tokens_seen": 266924600,
+      "step": 4810,
+      "train_runtime": 28464.0401,
+      "train_tokens_per_second": 9377.608
+    },
+    {
+      "epoch": 1.9384057971014492,
+      "grad_norm": 0.7110103013685544,
+      "learning_rate": 2.3690529496602777e-08,
+      "loss": 0.3108,
+      "num_input_tokens_seen": 267238432,
+      "step": 4815,
+      "train_runtime": 28495.09,
+      "train_tokens_per_second": 9378.403
+    },
+    {
+      "epoch": 1.9404186795491143,
+      "grad_norm": 0.777145374037139,
+      "learning_rate": 2.2178276403708175e-08,
+      "loss": 0.3388,
+      "num_input_tokens_seen": 267542320,
+      "step": 4820,
+      "train_runtime": 28526.393,
+      "train_tokens_per_second": 9378.764
+    },
+    {
+      "epoch": 1.9424315619967794,
+      "grad_norm": 0.7856916907848265,
+      "learning_rate": 2.0715787342343586e-08,
+      "loss": 0.3038,
+      "num_input_tokens_seen": 267830064,
+      "step": 4825,
+      "train_runtime": 28557.5312,
+      "train_tokens_per_second": 9378.614
+    },
+    {
+      "epoch": 1.9444444444444444,
+      "grad_norm": 0.7703028441247792,
+      "learning_rate": 1.9303076933233077e-08,
+      "loss": 0.3134,
+      "num_input_tokens_seen": 268093680,
+      "step": 4830,
+      "train_runtime": 28587.3387,
+      "train_tokens_per_second": 9378.057
+    },
+    {
+      "epoch": 1.9464573268921095,
+      "grad_norm": 0.7204253683028499,
+      "learning_rate": 1.7940159299453806e-08,
+      "loss": 0.3522,
+      "num_input_tokens_seen": 268380248,
+      "step": 4835,
+      "train_runtime": 28623.5846,
+      "train_tokens_per_second": 9376.193
+    },
+    {
+      "epoch": 1.9484702093397746,
+      "grad_norm": 0.7344953739759582,
+      "learning_rate": 1.662704806629889e-08,
+      "loss": 0.3252,
+      "num_input_tokens_seen": 268685488,
+      "step": 4840,
+      "train_runtime": 28654.6334,
+      "train_tokens_per_second": 9376.686
+    },
+    {
+      "epoch": 1.9504830917874396,
+      "grad_norm": 0.7287185157430882,
+      "learning_rate": 1.5363756361138072e-08,
+      "loss": 0.3122,
+      "num_input_tokens_seen": 268971544,
+      "step": 4845,
+      "train_runtime": 28681.3518,
+      "train_tokens_per_second": 9377.924
+    },
+    {
+      "epoch": 1.9524959742351047,
+      "grad_norm": 0.8254639204771651,
+      "learning_rate": 1.4150296813288388e-08,
+      "loss": 0.2881,
+      "num_input_tokens_seen": 269253440,
+      "step": 4850,
+      "train_runtime": 28709.6249,
+      "train_tokens_per_second": 9378.508
+    },
+    {
+      "epoch": 1.9545088566827697,
+      "grad_norm": 0.8095115861492299,
+      "learning_rate": 1.2986681553887048e-08,
+      "loss": 0.3183,
+      "num_input_tokens_seen": 269536808,
+      "step": 4855,
+      "train_runtime": 28734.3739,
+      "train_tokens_per_second": 9380.292
+    },
+    {
+      "epoch": 1.9565217391304348,
+      "grad_norm": 0.7135156397524652,
+      "learning_rate": 1.1872922215770966e-08,
+      "loss": 0.3394,
+      "num_input_tokens_seen": 269814056,
+      "step": 4860,
+      "train_runtime": 28758.6697,
+      "train_tokens_per_second": 9382.008
+    },
+    {
+      "epoch": 1.9585346215780999,
+      "grad_norm": 0.802633675015495,
+      "learning_rate": 1.0809029933359083e-08,
+      "loss": 0.3203,
+      "num_input_tokens_seen": 270093208,
+      "step": 4865,
+      "train_runtime": 28794.5384,
+      "train_tokens_per_second": 9380.015
+    },
+    {
+      "epoch": 1.960547504025765,
+      "grad_norm": 0.8114569469668379,
+      "learning_rate": 9.795015342543013e-09,
+      "loss": 0.3311,
+      "num_input_tokens_seen": 270367160,
+      "step": 4870,
+      "train_runtime": 28825.2732,
+      "train_tokens_per_second": 9379.518
+    },
+    {
+      "epoch": 1.96256038647343,
+      "grad_norm": 0.7628875850949784,
+      "learning_rate": 8.830888580578233e-09,
+      "loss": 0.3031,
+      "num_input_tokens_seen": 270625544,
+      "step": 4875,
+      "train_runtime": 28855.0566,
+      "train_tokens_per_second": 9378.791
+    },
+    {
+      "epoch": 1.9645732689210949,
+      "grad_norm": 0.7764391453673559,
+      "learning_rate": 7.916659285985284e-09,
+      "loss": 0.3299,
+      "num_input_tokens_seen": 270901192,
+      "step": 4880,
+      "train_runtime": 28885.1884,
+      "train_tokens_per_second": 9378.55
+    },
+    {
+      "epoch": 1.9665861513687601,
+      "grad_norm": 0.9139938687366106,
+      "learning_rate": 7.052336598451504e-09,
+      "loss": 0.322,
+      "num_input_tokens_seen": 271164784,
+      "step": 4885,
+      "train_runtime": 28915.592,
+      "train_tokens_per_second": 9377.805
+    },
+    {
+      "epoch": 1.968599033816425,
+      "grad_norm": 0.7227095780593157,
+      "learning_rate": 6.237929158741107e-09,
+      "loss": 0.3202,
+      "num_input_tokens_seen": 271434272,
+      "step": 4890,
+      "train_runtime": 28947.587,
+      "train_tokens_per_second": 9376.749
+    },
+    {
+      "epoch": 1.9706119162640903,
+      "grad_norm": 0.83949267784495,
+      "learning_rate": 5.473445108606923e-09,
+      "loss": 0.3268,
+      "num_input_tokens_seen": 271697952,
+      "step": 4895,
+      "train_runtime": 28979.2225,
+      "train_tokens_per_second": 9375.612
+    },
+    {
+      "epoch": 1.9726247987117551,
+      "grad_norm": 0.7931098692968355,
+      "learning_rate": 4.758892090711009e-09,
+      "loss": 0.3295,
+      "num_input_tokens_seen": 271967552,
+      "step": 4900,
+      "train_runtime": 29001.3034,
+      "train_tokens_per_second": 9377.77
+    },
+    {
+      "epoch": 1.9746376811594204,
+      "grad_norm": 0.7602519201676003,
+      "learning_rate": 4.094277248548051e-09,
+      "loss": 0.3407,
+      "num_input_tokens_seen": 272234544,
+      "step": 4905,
+      "train_runtime": 29024.584,
+      "train_tokens_per_second": 9379.447
+    },
+    {
+      "epoch": 1.9766505636070852,
+      "grad_norm": 0.8150284644035048,
+      "learning_rate": 3.4796072263715284e-09,
+      "loss": 0.3148,
+      "num_input_tokens_seen": 272495800,
+      "step": 4910,
+      "train_runtime": 29050.2239,
+      "train_tokens_per_second": 9380.162
+    },
+    {
+      "epoch": 1.9786634460547505,
+      "grad_norm": 0.7731598316187499,
+      "learning_rate": 2.9148881691298812e-09,
+      "loss": 0.3233,
+      "num_input_tokens_seen": 272773144,
+      "step": 4915,
+      "train_runtime": 29081.5731,
+      "train_tokens_per_second": 9379.587
+    },
+    {
+      "epoch": 1.9806763285024154,
+      "grad_norm": 0.6977054372397983,
+      "learning_rate": 2.400125722404889e-09,
+      "loss": 0.3301,
+      "num_input_tokens_seen": 273044976,
+      "step": 4920,
+      "train_runtime": 29104.0034,
+      "train_tokens_per_second": 9381.698
+    },
+    {
+      "epoch": 1.9826892109500807,
+      "grad_norm": 0.7449776659422748,
+      "learning_rate": 1.9353250323539407e-09,
+      "loss": 0.3379,
+      "num_input_tokens_seen": 273314184,
+      "step": 4925,
+      "train_runtime": 29139.3197,
+      "train_tokens_per_second": 9379.566
+    },
+    {
+      "epoch": 1.9847020933977455,
+      "grad_norm": 0.7503151520732206,
+      "learning_rate": 1.5204907456595197e-09,
+      "loss": 0.3203,
+      "num_input_tokens_seen": 273634520,
+      "step": 4930,
+      "train_runtime": 29167.9044,
+      "train_tokens_per_second": 9381.357
+    },
+    {
+      "epoch": 1.9867149758454108,
+      "grad_norm": 0.9510330425123027,
+      "learning_rate": 1.1556270094820189e-09,
+      "loss": 0.3264,
+      "num_input_tokens_seen": 273902336,
+      "step": 4935,
+      "train_runtime": 29201.4993,
+      "train_tokens_per_second": 9379.735
+    },
+    {
+      "epoch": 1.9887278582930756,
+      "grad_norm": 0.7796532707408381,
+      "learning_rate": 8.407374714192174e-10,
+      "loss": 0.3399,
+      "num_input_tokens_seen": 274183168,
+      "step": 4940,
+      "train_runtime": 29239.061,
+      "train_tokens_per_second": 9377.29
+    },
+    {
+      "epoch": 1.9907407407407407,
+      "grad_norm": 0.8264341965323302,
+      "learning_rate": 5.758252794690888e-10,
+      "loss": 0.3314,
+      "num_input_tokens_seen": 274462280,
+      "step": 4945,
+      "train_runtime": 29265.8976,
+      "train_tokens_per_second": 9378.229
+    },
+    {
+      "epoch": 1.9927536231884058,
+      "grad_norm": 0.7920653896898673,
+      "learning_rate": 3.608930819981593e-10,
+      "loss": 0.3105,
+      "num_input_tokens_seen": 274708976,
+      "step": 4950,
+      "train_runtime": 29294.5876,
+      "train_tokens_per_second": 9377.465
+    },
+    {
+      "epoch": 1.9947665056360708,
+      "grad_norm": 0.6609197175202728,
+      "learning_rate": 1.9594302771652752e-10,
+      "loss": 0.3266,
+      "num_input_tokens_seen": 274999376,
+      "step": 4955,
+      "train_runtime": 29323.0009,
+      "train_tokens_per_second": 9378.282
+    },
+    {
+      "epoch": 1.996779388083736,
+      "grad_norm": 0.7042685126356053,
+      "learning_rate": 8.097676565399504e-11,
+      "loss": 0.301,
+      "num_input_tokens_seen": 275242304,
+      "step": 4960,
+      "train_runtime": 29354.9514,
+      "train_tokens_per_second": 9376.35
+    },
+    {
+      "epoch": 1.998792270531401,
+      "grad_norm": 0.7323046256494772,
+      "learning_rate": 1.599544514674367e-11,
+      "loss": 0.3388,
+      "num_input_tokens_seen": 275524496,
+      "step": 4965,
+      "train_runtime": 29388.7866,
+      "train_tokens_per_second": 9375.157
+    },
+    {
+      "epoch": 2.0,
+      "num_input_tokens_seen": 275688256,
+      "step": 4968,
+      "total_flos": 509842421743616.0,
+      "train_loss": 0.38044202104570785,
+      "train_runtime": 29517.3177,
+      "train_samples_per_second": 0.673,
+      "train_steps_per_second": 0.168
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4968,
+  "num_input_tokens_seen": 275688256,
+  "num_train_epochs": 2,
+  "save_steps": 0,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 509842421743616.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}