diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9574 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 4768,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0020973154362416107,
+      "grad_norm": 1.926086246297868,
+      "learning_rate": 9.999982634512015e-06,
+      "loss": 0.7613,
+      "num_input_tokens_seen": 265584,
+      "step": 5,
+      "train_runtime": 39.3222,
+      "train_tokens_per_second": 6754.055
+    },
+    {
+      "epoch": 0.0041946308724832215,
+      "grad_norm": 1.261439278052809,
+      "learning_rate": 9.99991208742381e-06,
+      "loss": 0.6113,
+      "num_input_tokens_seen": 547280,
+      "step": 10,
+      "train_runtime": 73.6556,
+      "train_tokens_per_second": 7430.258
+    },
+    {
+      "epoch": 0.006291946308724832,
+      "grad_norm": 1.1859621194132017,
+      "learning_rate": 9.999787274157475e-06,
+      "loss": 0.5919,
+      "num_input_tokens_seen": 816920,
+      "step": 15,
+      "train_runtime": 104.6879,
+      "train_tokens_per_second": 7803.388
+    },
+    {
+      "epoch": 0.008389261744966443,
+      "grad_norm": 0.999921552861195,
+      "learning_rate": 9.99960819606766e-06,
+      "loss": 0.5286,
+      "num_input_tokens_seen": 1109840,
+      "step": 20,
+      "train_runtime": 138.3979,
+      "train_tokens_per_second": 8019.196
+    },
+    {
+      "epoch": 0.010486577181208054,
+      "grad_norm": 0.941416766648573,
+      "learning_rate": 9.999374855097978e-06,
+      "loss": 0.5701,
+      "num_input_tokens_seen": 1381240,
+      "step": 25,
+      "train_runtime": 164.4132,
+      "train_tokens_per_second": 8401.027
+    },
+    {
+      "epoch": 0.012583892617449664,
+      "grad_norm": 0.9587203784754852,
+      "learning_rate": 9.999087253780976e-06,
+      "loss": 0.5163,
+      "num_input_tokens_seen": 1669392,
+      "step": 30,
+      "train_runtime": 192.4425,
+      "train_tokens_per_second": 8674.759
+    },
+    {
+      "epoch": 0.014681208053691275,
+      "grad_norm": 0.8636585005416914,
+      "learning_rate": 9.998745395238115e-06,
+      "loss": 0.5224,
+      "num_input_tokens_seen": 1957496,
+      "step": 35,
+      "train_runtime": 221.2608,
+      "train_tokens_per_second": 8847.01
+    },
+    {
+      "epoch": 0.016778523489932886,
+      "grad_norm": 0.9194126763356688,
+      "learning_rate": 9.998349283179731e-06,
+      "loss": 0.473,
+      "num_input_tokens_seen": 2220816,
+      "step": 40,
+      "train_runtime": 248.4373,
+      "train_tokens_per_second": 8939.142
+    },
+    {
+      "epoch": 0.018875838926174497,
+      "grad_norm": 1.1483873576882146,
+      "learning_rate": 9.997898921904999e-06,
+      "loss": 0.4748,
+      "num_input_tokens_seen": 2505816,
+      "step": 45,
+      "train_runtime": 276.9879,
+      "train_tokens_per_second": 9046.663
+    },
+    {
+      "epoch": 0.02097315436241611,
+      "grad_norm": 0.8472408793809074,
+      "learning_rate": 9.997394316301878e-06,
+      "loss": 0.5127,
+      "num_input_tokens_seen": 2802512,
+      "step": 50,
+      "train_runtime": 298.1538,
+      "train_tokens_per_second": 9399.55
+    },
+    {
+      "epoch": 0.02307046979865772,
+      "grad_norm": 0.8593870670961665,
+      "learning_rate": 9.99683547184707e-06,
+      "loss": 0.5038,
+      "num_input_tokens_seen": 3071272,
+      "step": 55,
+      "train_runtime": 324.6159,
+      "train_tokens_per_second": 9461.25
+    },
+    {
+      "epoch": 0.025167785234899327,
+      "grad_norm": 0.9842907992226929,
+      "learning_rate": 9.996222394605954e-06,
+      "loss": 0.4908,
+      "num_input_tokens_seen": 3340120,
+      "step": 60,
+      "train_runtime": 348.724,
+      "train_tokens_per_second": 9578.118
+    },
+    {
+      "epoch": 0.02726510067114094,
+      "grad_norm": 0.9813885651159995,
+      "learning_rate": 9.995555091232516e-06,
+      "loss": 0.4603,
+      "num_input_tokens_seen": 3594408,
+      "step": 65,
+      "train_runtime": 372.3455,
+      "train_tokens_per_second": 9653.421
+    },
+    {
+      "epoch": 0.02936241610738255,
+      "grad_norm": 0.8531148421966384,
+      "learning_rate": 9.994833568969286e-06,
+      "loss": 0.4847,
+      "num_input_tokens_seen": 3873856,
+      "step": 70,
+      "train_runtime": 400.2223,
+      "train_tokens_per_second": 9679.26
+    },
+    {
+      "epoch": 0.031459731543624164,
+      "grad_norm": 0.931067544886876,
+      "learning_rate": 9.994057835647251e-06,
+      "loss": 0.5083,
+      "num_input_tokens_seen": 4152112,
+      "step": 75,
+      "train_runtime": 431.6934,
+      "train_tokens_per_second": 9618.197
+    },
+    {
+      "epoch": 0.03355704697986577,
+      "grad_norm": 0.9934685022213706,
+      "learning_rate": 9.993227899685777e-06,
+      "loss": 0.4829,
+      "num_input_tokens_seen": 4409536,
+      "step": 80,
+      "train_runtime": 454.1543,
+      "train_tokens_per_second": 9709.334
+    },
+    {
+      "epoch": 0.03565436241610738,
+      "grad_norm": 0.8989586490246144,
+      "learning_rate": 9.992343770092513e-06,
+      "loss": 0.4681,
+      "num_input_tokens_seen": 4685192,
+      "step": 85,
+      "train_runtime": 483.1737,
+      "train_tokens_per_second": 9696.704
+    },
+    {
+      "epoch": 0.037751677852348994,
+      "grad_norm": 1.249228624767336,
+      "learning_rate": 9.991405456463293e-06,
+      "loss": 0.4921,
+      "num_input_tokens_seen": 4950288,
+      "step": 90,
+      "train_runtime": 505.4868,
+      "train_tokens_per_second": 9793.111
+    },
+    {
+      "epoch": 0.0398489932885906,
+      "grad_norm": 0.8764946003568892,
+      "learning_rate": 9.990412968982037e-06,
+      "loss": 0.4711,
+      "num_input_tokens_seen": 5233216,
+      "step": 95,
+      "train_runtime": 538.3762,
+      "train_tokens_per_second": 9720.37
+    },
+    {
+      "epoch": 0.04194630872483222,
+      "grad_norm": 0.9320182309832723,
+      "learning_rate": 9.989366318420632e-06,
+      "loss": 0.4702,
+      "num_input_tokens_seen": 5497216,
+      "step": 100,
+      "train_runtime": 568.5595,
+      "train_tokens_per_second": 9668.673
+    },
+    {
+      "epoch": 0.044043624161073824,
+      "grad_norm": 0.9176640661039285,
+      "learning_rate": 9.988265516138827e-06,
+      "loss": 0.4489,
+      "num_input_tokens_seen": 5800408,
+      "step": 105,
+      "train_runtime": 598.1794,
+      "train_tokens_per_second": 9696.771
+    },
+    {
+      "epoch": 0.04614093959731544,
+      "grad_norm": 0.8479137702050689,
+      "learning_rate": 9.987110574084095e-06,
+      "loss": 0.4598,
+      "num_input_tokens_seen": 6070288,
+      "step": 110,
+      "train_runtime": 625.1984,
+      "train_tokens_per_second": 9709.379
+    },
+    {
+      "epoch": 0.04823825503355705,
+      "grad_norm": 0.9365799665779686,
+      "learning_rate": 9.985901504791516e-06,
+      "loss": 0.4527,
+      "num_input_tokens_seen": 6316624,
+      "step": 115,
+      "train_runtime": 655.6199,
+      "train_tokens_per_second": 9634.583
+    },
+    {
+      "epoch": 0.050335570469798654,
+      "grad_norm": 0.9545535838318295,
+      "learning_rate": 9.984638321383634e-06,
+      "loss": 0.4506,
+      "num_input_tokens_seen": 6550120,
+      "step": 120,
+      "train_runtime": 686.195,
+      "train_tokens_per_second": 9545.566
+    },
+    {
+      "epoch": 0.05243288590604027,
+      "grad_norm": 1.2794303060835923,
+      "learning_rate": 9.98332103757032e-06,
+      "loss": 0.4531,
+      "num_input_tokens_seen": 6810960,
+      "step": 125,
+      "train_runtime": 714.678,
+      "train_tokens_per_second": 9530.111
+    },
+    {
+      "epoch": 0.05453020134228188,
+      "grad_norm": 0.8452520052776062,
+      "learning_rate": 9.981949667648614e-06,
+      "loss": 0.4407,
+      "num_input_tokens_seen": 7085912,
+      "step": 130,
+      "train_runtime": 741.8582,
+      "train_tokens_per_second": 9551.572
+    },
+    {
+      "epoch": 0.05662751677852349,
+      "grad_norm": 0.881153491503371,
+      "learning_rate": 9.98052422650258e-06,
+      "loss": 0.4534,
+      "num_input_tokens_seen": 7359320,
+      "step": 135,
+      "train_runtime": 772.9461,
+      "train_tokens_per_second": 9521.13
+    },
+    {
+      "epoch": 0.0587248322147651,
+      "grad_norm": 0.9216626635324037,
+      "learning_rate": 9.979044729603136e-06,
+      "loss": 0.4311,
+      "num_input_tokens_seen": 7617792,
+      "step": 140,
+      "train_runtime": 798.1193,
+      "train_tokens_per_second": 9544.678
+    },
+    {
+      "epoch": 0.060822147651006714,
+      "grad_norm": 0.9458441171957542,
+      "learning_rate": 9.977511193007896e-06,
+      "loss": 0.4679,
+      "num_input_tokens_seen": 7884768,
+      "step": 145,
+      "train_runtime": 830.5768,
+      "train_tokens_per_second": 9493.123
+    },
+    {
+      "epoch": 0.06291946308724833,
+      "grad_norm": 0.9204873422710974,
+      "learning_rate": 9.975923633360985e-06,
+      "loss": 0.3999,
+      "num_input_tokens_seen": 8134416,
+      "step": 150,
+      "train_runtime": 857.5877,
+      "train_tokens_per_second": 9485.23
+    },
+    {
+      "epoch": 0.06501677852348993,
+      "grad_norm": 0.8366181085149906,
+      "learning_rate": 9.974282067892867e-06,
+      "loss": 0.4406,
+      "num_input_tokens_seen": 8410704,
+      "step": 155,
+      "train_runtime": 883.0057,
+      "train_tokens_per_second": 9525.085
+    },
+    {
+      "epoch": 0.06711409395973154,
+      "grad_norm": 0.8964724549065827,
+      "learning_rate": 9.972586514420148e-06,
+      "loss": 0.48,
+      "num_input_tokens_seen": 8669152,
+      "step": 160,
+      "train_runtime": 916.9802,
+      "train_tokens_per_second": 9454.023
+    },
+    {
+      "epoch": 0.06921140939597316,
+      "grad_norm": 0.8113109733917397,
+      "learning_rate": 9.970836991345398e-06,
+      "loss": 0.4468,
+      "num_input_tokens_seen": 8947264,
+      "step": 165,
+      "train_runtime": 949.2018,
+      "train_tokens_per_second": 9426.092
+    },
+    {
+      "epoch": 0.07130872483221476,
+      "grad_norm": 0.8522870915267229,
+      "learning_rate": 9.969033517656933e-06,
+      "loss": 0.4367,
+      "num_input_tokens_seen": 9241264,
+      "step": 170,
+      "train_runtime": 974.312,
+      "train_tokens_per_second": 9484.913
+    },
+    {
+      "epoch": 0.07340604026845637,
+      "grad_norm": 0.823162742316672,
+      "learning_rate": 9.967176112928622e-06,
+      "loss": 0.4574,
+      "num_input_tokens_seen": 9525624,
+      "step": 175,
+      "train_runtime": 1005.163,
+      "train_tokens_per_second": 9476.695
+    },
+    {
+      "epoch": 0.07550335570469799,
+      "grad_norm": 1.0240908363524686,
+      "learning_rate": 9.965264797319675e-06,
+      "loss": 0.4566,
+      "num_input_tokens_seen": 9819112,
+      "step": 180,
+      "train_runtime": 1030.8663,
+      "train_tokens_per_second": 9525.107
+    },
+    {
+      "epoch": 0.0776006711409396,
+      "grad_norm": 0.8199624395664985,
+      "learning_rate": 9.963299591574411e-06,
+      "loss": 0.4323,
+      "num_input_tokens_seen": 10093360,
+      "step": 185,
+      "train_runtime": 1064.9912,
+      "train_tokens_per_second": 9477.411
+    },
+    {
+      "epoch": 0.0796979865771812,
+      "grad_norm": 1.0461945003503732,
+      "learning_rate": 9.961280517022048e-06,
+      "loss": 0.4426,
+      "num_input_tokens_seen": 10346968,
+      "step": 190,
+      "train_runtime": 1095.3152,
+      "train_tokens_per_second": 9446.567
+    },
+    {
+      "epoch": 0.08179530201342282,
+      "grad_norm": 0.8167699545795873,
+      "learning_rate": 9.959207595576464e-06,
+      "loss": 0.4296,
+      "num_input_tokens_seen": 10597208,
+      "step": 195,
+      "train_runtime": 1122.0762,
+      "train_tokens_per_second": 9444.286
+    },
+    {
+      "epoch": 0.08389261744966443,
+      "grad_norm": 0.9261355996776974,
+      "learning_rate": 9.957080849735955e-06,
+      "loss": 0.4317,
+      "num_input_tokens_seen": 10860136,
+      "step": 200,
+      "train_runtime": 1151.7632,
+      "train_tokens_per_second": 9429.14
+    },
+    {
+      "epoch": 0.08598993288590603,
+      "grad_norm": 0.8768807089664538,
+      "learning_rate": 9.954900302583007e-06,
+      "loss": 0.4608,
+      "num_input_tokens_seen": 11133816,
+      "step": 205,
+      "train_runtime": 1177.259,
+      "train_tokens_per_second": 9457.406
+    },
+    {
+      "epoch": 0.08808724832214765,
+      "grad_norm": 0.8455004712755052,
+      "learning_rate": 9.952665977784022e-06,
+      "loss": 0.3996,
+      "num_input_tokens_seen": 11413952,
+      "step": 210,
+      "train_runtime": 1207.3546,
+      "train_tokens_per_second": 9453.687
+    },
+    {
+      "epoch": 0.09018456375838926,
+      "grad_norm": 1.6535121419979995,
+      "learning_rate": 9.950377899589083e-06,
+      "loss": 0.4334,
+      "num_input_tokens_seen": 11704272,
+      "step": 215,
+      "train_runtime": 1227.1611,
+      "train_tokens_per_second": 9537.682
+    },
+    {
+      "epoch": 0.09228187919463088,
+      "grad_norm": 0.9996119639988559,
+      "learning_rate": 9.948036092831679e-06,
+      "loss": 0.4211,
+      "num_input_tokens_seen": 11974192,
+      "step": 220,
+      "train_runtime": 1261.5351,
+      "train_tokens_per_second": 9491.763
+    },
+    {
+      "epoch": 0.09437919463087248,
+      "grad_norm": 0.893662938029802,
+      "learning_rate": 9.945640582928438e-06,
+      "loss": 0.4205,
+      "num_input_tokens_seen": 12254464,
+      "step": 225,
+      "train_runtime": 1289.5649,
+      "train_tokens_per_second": 9502.789
+    },
+    {
+      "epoch": 0.0964765100671141,
+      "grad_norm": 0.9362007968042841,
+      "learning_rate": 9.943191395878848e-06,
+      "loss": 0.4442,
+      "num_input_tokens_seen": 12542824,
+      "step": 230,
+      "train_runtime": 1317.9099,
+      "train_tokens_per_second": 9517.209
+    },
+    {
+      "epoch": 0.09857382550335571,
+      "grad_norm": 0.9282776330755926,
+      "learning_rate": 9.940688558264982e-06,
+      "loss": 0.4534,
+      "num_input_tokens_seen": 12829488,
+      "step": 235,
+      "train_runtime": 1348.7876,
+      "train_tokens_per_second": 9511.867
+    },
+    {
+      "epoch": 0.10067114093959731,
+      "grad_norm": 0.732215808654604,
+      "learning_rate": 9.938132097251202e-06,
+      "loss": 0.4379,
+      "num_input_tokens_seen": 13104904,
+      "step": 240,
+      "train_runtime": 1379.6651,
+      "train_tokens_per_second": 9498.612
+    },
+    {
+      "epoch": 0.10276845637583892,
+      "grad_norm": 0.8564740174215485,
+      "learning_rate": 9.935522040583871e-06,
+      "loss": 0.442,
+      "num_input_tokens_seen": 13385128,
+      "step": 245,
+      "train_runtime": 1413.4412,
+      "train_tokens_per_second": 9469.887
+    },
+    {
+      "epoch": 0.10486577181208054,
+      "grad_norm": 0.845679911455843,
+      "learning_rate": 9.93285841659105e-06,
+      "loss": 0.4123,
+      "num_input_tokens_seen": 13628968,
+      "step": 250,
+      "train_runtime": 1439.075,
+      "train_tokens_per_second": 9470.645
+    },
+    {
+      "epoch": 0.10696308724832215,
+      "grad_norm": 1.0257801986295452,
+      "learning_rate": 9.930141254182182e-06,
+      "loss": 0.4254,
+      "num_input_tokens_seen": 13917968,
+      "step": 255,
+      "train_runtime": 1466.325,
+      "train_tokens_per_second": 9491.735
+    },
+    {
+      "epoch": 0.10906040268456375,
+      "grad_norm": 0.8484468656955171,
+      "learning_rate": 9.927370582847791e-06,
+      "loss": 0.4171,
+      "num_input_tokens_seen": 14188760,
+      "step": 260,
+      "train_runtime": 1489.9239,
+      "train_tokens_per_second": 9523.144
+    },
+    {
+      "epoch": 0.11115771812080537,
+      "grad_norm": 0.8358975784763669,
+      "learning_rate": 9.924546432659155e-06,
+      "loss": 0.4383,
+      "num_input_tokens_seen": 14458952,
+      "step": 265,
+      "train_runtime": 1517.716,
+      "train_tokens_per_second": 9526.784
+    },
+    {
+      "epoch": 0.11325503355704698,
+      "grad_norm": 0.7732161939636181,
+      "learning_rate": 9.921668834267979e-06,
+      "loss": 0.4185,
+      "num_input_tokens_seen": 14721776,
+      "step": 270,
+      "train_runtime": 1549.1494,
+      "train_tokens_per_second": 9503.135
+    },
+    {
+      "epoch": 0.11535234899328858,
+      "grad_norm": 0.7793159548804247,
+      "learning_rate": 9.918737818906066e-06,
+      "loss": 0.4147,
+      "num_input_tokens_seen": 15010920,
+      "step": 275,
+      "train_runtime": 1575.8051,
+      "train_tokens_per_second": 9525.873
+    },
+    {
+      "epoch": 0.1174496644295302,
+      "grad_norm": 0.873983234720518,
+      "learning_rate": 9.915753418384978e-06,
+      "loss": 0.4209,
+      "num_input_tokens_seen": 15296360,
+      "step": 280,
+      "train_runtime": 1603.1797,
+      "train_tokens_per_second": 9541.263
+    },
+    {
+      "epoch": 0.11954697986577181,
+      "grad_norm": 0.7120314266092039,
+      "learning_rate": 9.912715665095683e-06,
+      "loss": 0.428,
+      "num_input_tokens_seen": 15586088,
+      "step": 285,
+      "train_runtime": 1630.8567,
+      "train_tokens_per_second": 9556.994
+    },
+    {
+      "epoch": 0.12164429530201343,
+      "grad_norm": 0.8080744270444242,
+      "learning_rate": 9.909624592008213e-06,
+      "loss": 0.4118,
+      "num_input_tokens_seen": 15901872,
+      "step": 290,
+      "train_runtime": 1660.1884,
+      "train_tokens_per_second": 9578.354
+    },
+    {
+      "epoch": 0.12374161073825503,
+      "grad_norm": 0.7799741436491128,
+      "learning_rate": 9.906480232671305e-06,
+      "loss": 0.4247,
+      "num_input_tokens_seen": 16178168,
+      "step": 295,
+      "train_runtime": 1686.6812,
+      "train_tokens_per_second": 9591.716
+    },
+    {
+      "epoch": 0.12583892617449666,
+      "grad_norm": 0.9160389553050224,
+      "learning_rate": 9.903282621212032e-06,
+      "loss": 0.4485,
+      "num_input_tokens_seen": 16439584,
+      "step": 300,
+      "train_runtime": 1717.122,
+      "train_tokens_per_second": 9573.917
+    },
+    {
+      "epoch": 0.12793624161073824,
+      "grad_norm": 0.768718296583002,
+      "learning_rate": 9.900031792335432e-06,
+      "loss": 0.4185,
+      "num_input_tokens_seen": 16725632,
+      "step": 305,
+      "train_runtime": 1738.8017,
+      "train_tokens_per_second": 9619.057
+    },
+    {
+      "epoch": 0.13003355704697986,
+      "grad_norm": 0.7913531195796439,
+      "learning_rate": 9.896727781324138e-06,
+      "loss": 0.3945,
+      "num_input_tokens_seen": 17001792,
+      "step": 310,
+      "train_runtime": 1772.2213,
+      "train_tokens_per_second": 9593.492
+    },
+    {
+      "epoch": 0.13213087248322147,
+      "grad_norm": 0.8964681244634585,
+      "learning_rate": 9.893370624037992e-06,
+      "loss": 0.4561,
+      "num_input_tokens_seen": 17289624,
+      "step": 315,
+      "train_runtime": 1800.4858,
+      "train_tokens_per_second": 9602.755
+    },
+    {
+      "epoch": 0.1342281879194631,
+      "grad_norm": 0.9300410992117513,
+      "learning_rate": 9.889960356913654e-06,
+      "loss": 0.416,
+      "num_input_tokens_seen": 17537824,
+      "step": 320,
+      "train_runtime": 1825.1801,
+      "train_tokens_per_second": 9608.818
+    },
+    {
+      "epoch": 0.1363255033557047,
+      "grad_norm": 0.8458438892868644,
+      "learning_rate": 9.886497016964204e-06,
+      "loss": 0.452,
+      "num_input_tokens_seen": 17820448,
+      "step": 325,
+      "train_runtime": 1852.2044,
+      "train_tokens_per_second": 9621.21
+    },
+    {
+      "epoch": 0.13842281879194632,
+      "grad_norm": 0.8881389528906882,
+      "learning_rate": 9.882980641778752e-06,
+      "loss": 0.454,
+      "num_input_tokens_seen": 18097568,
+      "step": 330,
+      "train_runtime": 1881.9774,
+      "train_tokens_per_second": 9616.252
+    },
+    {
+      "epoch": 0.14052013422818793,
+      "grad_norm": 0.9062766684747616,
+      "learning_rate": 9.879411269522015e-06,
+      "loss": 0.3719,
+      "num_input_tokens_seen": 18373568,
+      "step": 335,
+      "train_runtime": 1916.0332,
+      "train_tokens_per_second": 9589.379
+    },
+    {
+      "epoch": 0.14261744966442952,
+      "grad_norm": 0.7513940069803771,
+      "learning_rate": 9.875788938933911e-06,
+      "loss": 0.3836,
+      "num_input_tokens_seen": 18664920,
+      "step": 340,
+      "train_runtime": 1940.0733,
+      "train_tokens_per_second": 9620.729
+    },
+    {
+      "epoch": 0.14471476510067113,
+      "grad_norm": 0.8584062957561568,
+      "learning_rate": 9.872113689329142e-06,
+      "loss": 0.4396,
+      "num_input_tokens_seen": 18938760,
+      "step": 345,
+      "train_runtime": 1968.7965,
+      "train_tokens_per_second": 9619.46
+    },
+    {
+      "epoch": 0.14681208053691275,
+      "grad_norm": 0.8402747842330597,
+      "learning_rate": 9.868385560596756e-06,
+      "loss": 0.4202,
+      "num_input_tokens_seen": 19183256,
+      "step": 350,
+      "train_runtime": 1995.7898,
+      "train_tokens_per_second": 9611.862
+    },
+    {
+      "epoch": 0.14890939597315436,
+      "grad_norm": 0.8945298343553769,
+      "learning_rate": 9.864604593199724e-06,
+      "loss": 0.4278,
+      "num_input_tokens_seen": 19459744,
+      "step": 355,
+      "train_runtime": 2019.4842,
+      "train_tokens_per_second": 9635.997
+    },
+    {
+      "epoch": 0.15100671140939598,
+      "grad_norm": 0.7901612196128083,
+      "learning_rate": 9.860770828174502e-06,
+      "loss": 0.4049,
+      "num_input_tokens_seen": 19731560,
+      "step": 360,
+      "train_runtime": 2052.3892,
+      "train_tokens_per_second": 9613.946
+    },
+    {
+      "epoch": 0.1531040268456376,
+      "grad_norm": 0.8413058218736352,
+      "learning_rate": 9.856884307130572e-06,
+      "loss": 0.4084,
+      "num_input_tokens_seen": 20027032,
+      "step": 365,
+      "train_runtime": 2084.1027,
+      "train_tokens_per_second": 9609.427
+    },
+    {
+      "epoch": 0.1552013422818792,
+      "grad_norm": 1.0179443824773717,
+      "learning_rate": 9.852945072250006e-06,
+      "loss": 0.401,
+      "num_input_tokens_seen": 20289072,
+      "step": 370,
+      "train_runtime": 2112.6217,
+      "train_tokens_per_second": 9603.741
+    },
+    {
+      "epoch": 0.1572986577181208,
+      "grad_norm": 0.8714737863085417,
+      "learning_rate": 9.848953166287003e-06,
+      "loss": 0.4247,
+      "num_input_tokens_seen": 20572928,
+      "step": 375,
+      "train_runtime": 2140.6015,
+      "train_tokens_per_second": 9610.816
+    },
+    {
+      "epoch": 0.1593959731543624,
+      "grad_norm": 0.797239488634744,
+      "learning_rate": 9.844908632567415e-06,
+      "loss": 0.4186,
+      "num_input_tokens_seen": 20855840,
+      "step": 380,
+      "train_runtime": 2177.2736,
+      "train_tokens_per_second": 9578.879
+    },
+    {
+      "epoch": 0.16149328859060402,
+      "grad_norm": 0.8107766703866764,
+      "learning_rate": 9.840811514988294e-06,
+      "loss": 0.4049,
+      "num_input_tokens_seen": 21145488,
+      "step": 385,
+      "train_runtime": 2200.0579,
+      "train_tokens_per_second": 9611.333
+    },
+    {
+      "epoch": 0.16359060402684564,
+      "grad_norm": 0.9578054041829148,
+      "learning_rate": 9.836661858017406e-06,
+      "loss": 0.4099,
+      "num_input_tokens_seen": 21406184,
+      "step": 390,
+      "train_runtime": 2232.3054,
+      "train_tokens_per_second": 9589.272
+    },
+    {
+      "epoch": 0.16568791946308725,
+      "grad_norm": 0.8633596258086575,
+      "learning_rate": 9.832459706692745e-06,
+      "loss": 0.3998,
+      "num_input_tokens_seen": 21690352,
+      "step": 395,
+      "train_runtime": 2260.318,
+      "train_tokens_per_second": 9596.151
+    },
+    {
+      "epoch": 0.16778523489932887,
+      "grad_norm": 0.8204382028799451,
+      "learning_rate": 9.828205106622055e-06,
+      "loss": 0.4086,
+      "num_input_tokens_seen": 21968760,
+      "step": 400,
+      "train_runtime": 2287.346,
+      "train_tokens_per_second": 9604.476
+    },
+    {
+      "epoch": 0.16988255033557048,
+      "grad_norm": 0.8397997700565066,
+      "learning_rate": 9.82389810398232e-06,
+      "loss": 0.4277,
+      "num_input_tokens_seen": 22244792,
+      "step": 405,
+      "train_runtime": 2317.2517,
+      "train_tokens_per_second": 9599.644
+    },
+    {
+      "epoch": 0.17197986577181207,
+      "grad_norm": 0.8194361178298493,
+      "learning_rate": 9.819538745519281e-06,
+      "loss": 0.3891,
+      "num_input_tokens_seen": 22525216,
+      "step": 410,
+      "train_runtime": 2352.8023,
+      "train_tokens_per_second": 9573.782
+    },
+    {
+      "epoch": 0.17407718120805368,
+      "grad_norm": 0.8019528780122389,
+      "learning_rate": 9.815127078546911e-06,
+      "loss": 0.4081,
+      "num_input_tokens_seen": 22788904,
+      "step": 415,
+      "train_runtime": 2381.0707,
+      "train_tokens_per_second": 9570.864
+    },
+    {
+      "epoch": 0.1761744966442953,
+      "grad_norm": 0.8169532758399493,
+      "learning_rate": 9.810663150946915e-06,
+      "loss": 0.4335,
+      "num_input_tokens_seen": 23054832,
+      "step": 420,
+      "train_runtime": 2402.3558,
+      "train_tokens_per_second": 9596.76
+    },
+    {
+      "epoch": 0.1782718120805369,
+      "grad_norm": 0.7719781050126183,
+      "learning_rate": 9.806147011168203e-06,
+      "loss": 0.3992,
+      "num_input_tokens_seen": 23349128,
+      "step": 425,
+      "train_runtime": 2423.8854,
+      "train_tokens_per_second": 9632.934
+    },
+    {
+      "epoch": 0.18036912751677853,
+      "grad_norm": 0.7836007506856649,
+      "learning_rate": 9.801578708226366e-06,
+      "loss": 0.4017,
+      "num_input_tokens_seen": 23636024,
+      "step": 430,
+      "train_runtime": 2457.0477,
+      "train_tokens_per_second": 9619.685
+    },
+    {
+      "epoch": 0.18246644295302014,
+      "grad_norm": 0.8187382513184207,
+      "learning_rate": 9.796958291703142e-06,
+      "loss": 0.4045,
+      "num_input_tokens_seen": 23894616,
+      "step": 435,
+      "train_runtime": 2490.7185,
+      "train_tokens_per_second": 9593.463
+    },
+    {
+      "epoch": 0.18456375838926176,
+      "grad_norm": 0.7147337887183921,
+      "learning_rate": 9.792285811745883e-06,
+      "loss": 0.3978,
+      "num_input_tokens_seen": 24189024,
+      "step": 440,
+      "train_runtime": 2523.7237,
+      "train_tokens_per_second": 9584.656
+    },
+    {
+      "epoch": 0.18666107382550334,
+      "grad_norm": 0.7813123002835518,
+      "learning_rate": 9.787561319067009e-06,
+      "loss": 0.4146,
+      "num_input_tokens_seen": 24484136,
+      "step": 445,
+      "train_runtime": 2548.9248,
+      "train_tokens_per_second": 9605.672
+    },
+    {
+      "epoch": 0.18875838926174496,
+      "grad_norm": 0.7578792927673074,
+      "learning_rate": 9.78278486494345e-06,
+      "loss": 0.4394,
+      "num_input_tokens_seen": 24780424,
+      "step": 450,
+      "train_runtime": 2576.2036,
+      "train_tokens_per_second": 9618.97
+    },
+    {
+      "epoch": 0.19085570469798657,
+      "grad_norm": 0.8231053970457799,
+      "learning_rate": 9.777956501216103e-06,
+      "loss": 0.416,
+      "num_input_tokens_seen": 25040968,
+      "step": 455,
+      "train_runtime": 2605.8409,
+      "train_tokens_per_second": 9609.554
+    },
+    {
+      "epoch": 0.1929530201342282,
+      "grad_norm": 0.7706198655389167,
+      "learning_rate": 9.773076280289256e-06,
+      "loss": 0.4154,
+      "num_input_tokens_seen": 25313472,
+      "step": 460,
+      "train_runtime": 2640.5442,
+      "train_tokens_per_second": 9586.46
+    },
+    {
+      "epoch": 0.1950503355704698,
+      "grad_norm": 0.8575786534102728,
+      "learning_rate": 9.76814425513003e-06,
+      "loss": 0.4132,
+      "num_input_tokens_seen": 25604248,
+      "step": 465,
+      "train_runtime": 2664.3469,
+      "train_tokens_per_second": 9609.953
+    },
+    {
+      "epoch": 0.19714765100671142,
+      "grad_norm": 0.8706275951329838,
+      "learning_rate": 9.763160479267795e-06,
+      "loss": 0.4001,
+      "num_input_tokens_seen": 25875224,
+      "step": 470,
+      "train_runtime": 2688.2418,
+      "train_tokens_per_second": 9625.334
+    },
+    {
+      "epoch": 0.19924496644295303,
+      "grad_norm": 0.7819240875987011,
+      "learning_rate": 9.758125006793596e-06,
+      "loss": 0.4267,
+      "num_input_tokens_seen": 26149984,
+      "step": 475,
+      "train_runtime": 2719.4157,
+      "train_tokens_per_second": 9616.031
+    },
+    {
+      "epoch": 0.20134228187919462,
+      "grad_norm": 0.8133393121489669,
+      "learning_rate": 9.753037892359565e-06,
+      "loss": 0.4159,
+      "num_input_tokens_seen": 26445456,
+      "step": 480,
+      "train_runtime": 2749.5829,
+      "train_tokens_per_second": 9617.988
+    },
+    {
+      "epoch": 0.20343959731543623,
+      "grad_norm": 0.7217209389383089,
+      "learning_rate": 9.747899191178325e-06,
+      "loss": 0.3864,
+      "num_input_tokens_seen": 26703632,
+      "step": 485,
+      "train_runtime": 2775.7335,
+      "train_tokens_per_second": 9620.388
+    },
+    {
+      "epoch": 0.20553691275167785,
+      "grad_norm": 0.8368590471522286,
+      "learning_rate": 9.74270895902239e-06,
+      "loss": 0.3972,
+      "num_input_tokens_seen": 26966632,
+      "step": 490,
+      "train_runtime": 2807.6117,
+      "train_tokens_per_second": 9604.83
+    },
+    {
+      "epoch": 0.20763422818791946,
+      "grad_norm": 0.8817240627429944,
+      "learning_rate": 9.73746725222356e-06,
+      "loss": 0.3849,
+      "num_input_tokens_seen": 27248856,
+      "step": 495,
+      "train_runtime": 2832.4551,
+      "train_tokens_per_second": 9620.225
+    },
+    {
+      "epoch": 0.20973154362416108,
+      "grad_norm": 0.8402576578478745,
+      "learning_rate": 9.732174127672321e-06,
+      "loss": 0.3966,
+      "num_input_tokens_seen": 27525360,
+      "step": 500,
+      "train_runtime": 2858.7591,
+      "train_tokens_per_second": 9628.429
+    },
+    {
+      "epoch": 0.2118288590604027,
+      "grad_norm": 0.9338307088436789,
+      "learning_rate": 9.726829642817204e-06,
+      "loss": 0.436,
+      "num_input_tokens_seen": 27806120,
+      "step": 505,
+      "train_runtime": 2889.8082,
+      "train_tokens_per_second": 9622.133
+    },
+    {
+      "epoch": 0.2139261744966443,
+      "grad_norm": 0.7790168231899691,
+      "learning_rate": 9.721433855664186e-06,
+      "loss": 0.4075,
+      "num_input_tokens_seen": 28081744,
+      "step": 510,
+      "train_runtime": 2920.9888,
+      "train_tokens_per_second": 9613.78
+    },
+    {
+      "epoch": 0.2160234899328859,
+      "grad_norm": 0.8626053092382174,
+      "learning_rate": 9.715986824776047e-06,
+      "loss": 0.3956,
+      "num_input_tokens_seen": 28390080,
+      "step": 515,
+      "train_runtime": 2946.9218,
+      "train_tokens_per_second": 9633.808
+    },
+    {
+      "epoch": 0.2181208053691275,
+      "grad_norm": 0.7188340010784988,
+      "learning_rate": 9.710488609271734e-06,
+      "loss": 0.381,
+      "num_input_tokens_seen": 28676720,
+      "step": 520,
+      "train_runtime": 2980.8642,
+      "train_tokens_per_second": 9620.271
+    },
+    {
+      "epoch": 0.22021812080536912,
+      "grad_norm": 0.7894577395077416,
+      "learning_rate": 9.704939268825726e-06,
+      "loss": 0.3985,
+      "num_input_tokens_seen": 28952552,
+      "step": 525,
+      "train_runtime": 3007.8727,
+      "train_tokens_per_second": 9625.591
+    },
+    {
+      "epoch": 0.22231543624161074,
+      "grad_norm": 0.7998696972875705,
+      "learning_rate": 9.69933886366738e-06,
+      "loss": 0.398,
+      "num_input_tokens_seen": 29225608,
+      "step": 530,
+      "train_runtime": 3028.5983,
+      "train_tokens_per_second": 9649.879
+    },
+    {
+      "epoch": 0.22441275167785235,
+      "grad_norm": 0.78729985672101,
+      "learning_rate": 9.693687454580283e-06,
+      "loss": 0.3996,
+      "num_input_tokens_seen": 29500056,
+      "step": 535,
+      "train_runtime": 3056.0257,
+      "train_tokens_per_second": 9653.078
+    },
+    {
+      "epoch": 0.22651006711409397,
+      "grad_norm": 0.7441010843456842,
+      "learning_rate": 9.687985102901588e-06,
+      "loss": 0.4183,
+      "num_input_tokens_seen": 29803616,
+      "step": 540,
+      "train_runtime": 3085.4567,
+      "train_tokens_per_second": 9659.386
+    },
+    {
+      "epoch": 0.22860738255033558,
+      "grad_norm": 0.8094923818610809,
+      "learning_rate": 9.682231870521347e-06,
+      "loss": 0.4198,
+      "num_input_tokens_seen": 30099616,
+      "step": 545,
+      "train_runtime": 3112.4363,
+      "train_tokens_per_second": 9670.757
+    },
+    {
+      "epoch": 0.23070469798657717,
+      "grad_norm": 0.8551144358409505,
+      "learning_rate": 9.676427819881846e-06,
+      "loss": 0.3944,
+      "num_input_tokens_seen": 30351744,
+      "step": 550,
+      "train_runtime": 3141.7855,
+      "train_tokens_per_second": 9660.667
+    },
+    {
+      "epoch": 0.23280201342281878,
+      "grad_norm": 0.7780031078407866,
+      "learning_rate": 9.670573013976924e-06,
+      "loss": 0.3805,
+      "num_input_tokens_seen": 30629064,
+      "step": 555,
+      "train_runtime": 3173.0421,
+      "train_tokens_per_second": 9652.902
+    },
+    {
+      "epoch": 0.2348993288590604,
+      "grad_norm": 0.8012478465294108,
+      "learning_rate": 9.664667516351282e-06,
+      "loss": 0.3994,
+      "num_input_tokens_seen": 30909904,
+      "step": 560,
+      "train_runtime": 3204.3607,
+      "train_tokens_per_second": 9646.2
+    },
+    {
+      "epoch": 0.236996644295302,
+      "grad_norm": 0.7603329817823411,
+      "learning_rate": 9.65871139109981e-06,
+      "loss": 0.3971,
+      "num_input_tokens_seen": 31200880,
+      "step": 565,
+      "train_runtime": 3228.842,
+      "train_tokens_per_second": 9663.18
+    },
+    {
+      "epoch": 0.23909395973154363,
+      "grad_norm": 0.8270792329538266,
+      "learning_rate": 9.652704702866868e-06,
+      "loss": 0.412,
+      "num_input_tokens_seen": 31442088,
+      "step": 570,
+      "train_runtime": 3250.2482,
+      "train_tokens_per_second": 9673.75
+    },
+    {
+      "epoch": 0.24119127516778524,
+      "grad_norm": 0.7286474076735678,
+      "learning_rate": 9.646647516845612e-06,
+      "loss": 0.4325,
+      "num_input_tokens_seen": 31743640,
+      "step": 575,
+      "train_runtime": 3279.7791,
+      "train_tokens_per_second": 9678.591
+    },
+    {
+      "epoch": 0.24328859060402686,
+      "grad_norm": 0.833846504170195,
+      "learning_rate": 9.64053989877726e-06,
+      "loss": 0.4311,
+      "num_input_tokens_seen": 32028000,
+      "step": 580,
+      "train_runtime": 3306.8807,
+      "train_tokens_per_second": 9685.26
+    },
+    {
+      "epoch": 0.24538590604026847,
+      "grad_norm": 0.731819806098127,
+      "learning_rate": 9.634381914950399e-06,
+      "loss": 0.3939,
+      "num_input_tokens_seen": 32306576,
+      "step": 585,
+      "train_runtime": 3341.3251,
+      "train_tokens_per_second": 9668.792
+    },
+    {
+      "epoch": 0.24748322147651006,
+      "grad_norm": 0.7925053430903917,
+      "learning_rate": 9.62817363220025e-06,
+      "loss": 0.389,
+      "num_input_tokens_seen": 32535072,
+      "step": 590,
+      "train_runtime": 3365.0361,
+      "train_tokens_per_second": 9668.565
+    },
+    {
+      "epoch": 0.24958053691275167,
+      "grad_norm": 0.7815796625895076,
+      "learning_rate": 9.621915117907956e-06,
+      "loss": 0.3778,
+      "num_input_tokens_seen": 32815176,
+      "step": 595,
+      "train_runtime": 3399.9667,
+      "train_tokens_per_second": 9651.617
+    },
+    {
+      "epoch": 0.2516778523489933,
+      "grad_norm": 0.8029824505935105,
+      "learning_rate": 9.615606439999843e-06,
+      "loss": 0.3886,
+      "num_input_tokens_seen": 33114040,
+      "step": 600,
+      "train_runtime": 3428.5293,
+      "train_tokens_per_second": 9658.38
+    },
+    {
+      "epoch": 0.2537751677852349,
+      "grad_norm": 1.6105088172818112,
+      "learning_rate": 9.60924766694668e-06,
+      "loss": 0.4092,
+      "num_input_tokens_seen": 33376096,
+      "step": 605,
+      "train_runtime": 3460.8179,
+      "train_tokens_per_second": 9643.991
+    },
+    {
+      "epoch": 0.2558724832214765,
+      "grad_norm": 0.8443823201914711,
+      "learning_rate": 9.602838867762941e-06,
+      "loss": 0.4228,
+      "num_input_tokens_seen": 33642840,
+      "step": 610,
+      "train_runtime": 3489.2216,
+      "train_tokens_per_second": 9641.933
+    },
+    {
+      "epoch": 0.25796979865771813,
+      "grad_norm": 0.7533217621462183,
+      "learning_rate": 9.59638011200606e-06,
+      "loss": 0.4072,
+      "num_input_tokens_seen": 33898736,
+      "step": 615,
+      "train_runtime": 3513.4267,
+      "train_tokens_per_second": 9648.34
+    },
+    {
+      "epoch": 0.2600671140939597,
+      "grad_norm": 0.8230507010928672,
+      "learning_rate": 9.589871469775667e-06,
+      "loss": 0.3962,
+      "num_input_tokens_seen": 34151760,
+      "step": 620,
+      "train_runtime": 3543.1073,
+      "train_tokens_per_second": 9638.929
+    },
+    {
+      "epoch": 0.26216442953020136,
+      "grad_norm": 1.0079573646577087,
+      "learning_rate": 9.583313011712832e-06,
+      "loss": 0.4172,
+      "num_input_tokens_seen": 34406504,
+      "step": 625,
+      "train_runtime": 3571.9213,
+      "train_tokens_per_second": 9632.492
+    },
+    {
+      "epoch": 0.26426174496644295,
+      "grad_norm": 0.8160943428226843,
+      "learning_rate": 9.576704808999295e-06,
+      "loss": 0.3887,
+      "num_input_tokens_seen": 34662216,
+      "step": 630,
+      "train_runtime": 3598.7365,
+      "train_tokens_per_second": 9631.774
+    },
+    {
+      "epoch": 0.2663590604026846,
+      "grad_norm": 0.7903991504279025,
+      "learning_rate": 9.5700469333567e-06,
+      "loss": 0.3773,
+      "num_input_tokens_seen": 34944128,
+      "step": 635,
+      "train_runtime": 3631.8181,
+      "train_tokens_per_second": 9621.663
+    },
+    {
+      "epoch": 0.2684563758389262,
+      "grad_norm": 0.855272292791432,
+      "learning_rate": 9.56333945704581e-06,
+      "loss": 0.3859,
+      "num_input_tokens_seen": 35197296,
+      "step": 640,
+      "train_runtime": 3660.9751,
+      "train_tokens_per_second": 9614.186
+    },
+    {
+      "epoch": 0.27055369127516776,
+      "grad_norm": 1.4554982828155416,
+      "learning_rate": 9.556582452865728e-06,
+      "loss": 0.4082,
+      "num_input_tokens_seen": 35479312,
+      "step": 645,
+      "train_runtime": 3693.0365,
+      "train_tokens_per_second": 9607.084
+    },
+    {
+      "epoch": 0.2726510067114094,
+      "grad_norm": 0.823015401579161,
+      "learning_rate": 9.549775994153098e-06,
+      "loss": 0.4143,
+      "num_input_tokens_seen": 35748560,
+      "step": 650,
+      "train_runtime": 3718.824,
+      "train_tokens_per_second": 9612.867
+    },
+    {
+      "epoch": 0.274748322147651,
+      "grad_norm": 0.8223472524338282,
+      "learning_rate": 9.542920154781323e-06,
+      "loss": 0.3763,
+      "num_input_tokens_seen": 36011616,
+      "step": 655,
+      "train_runtime": 3743.6837,
+      "train_tokens_per_second": 9619.3
+    },
+    {
+      "epoch": 0.27684563758389263,
+      "grad_norm": 0.7427855246793537,
+      "learning_rate": 9.536015009159749e-06,
+      "loss": 0.3824,
+      "num_input_tokens_seen": 36289696,
+      "step": 660,
+      "train_runtime": 3772.3727,
+      "train_tokens_per_second": 9619.86
+    },
+    {
+      "epoch": 0.2789429530201342,
+      "grad_norm": 0.8863793301662561,
+      "learning_rate": 9.529060632232867e-06,
+      "loss": 0.3862,
+      "num_input_tokens_seen": 36559104,
+      "step": 665,
+      "train_runtime": 3804.0729,
+      "train_tokens_per_second": 9610.516
+    },
+    {
+      "epoch": 0.28104026845637586,
+      "grad_norm": 0.6982273118090992,
+      "learning_rate": 9.522057099479495e-06,
+      "loss": 0.3682,
+      "num_input_tokens_seen": 36821376,
+      "step": 670,
+      "train_runtime": 3829.0408,
+      "train_tokens_per_second": 9616.345
+    },
+    {
+      "epoch": 0.28313758389261745,
+      "grad_norm": 0.8428208517748849,
+      "learning_rate": 9.51500448691196e-06,
+      "loss": 0.3991,
+      "num_input_tokens_seen": 37108256,
+      "step": 675,
+      "train_runtime": 3862.1144,
+      "train_tokens_per_second": 9608.275
+    },
+    {
+      "epoch": 0.28523489932885904,
+      "grad_norm": 1.0103110839367158,
+      "learning_rate": 9.507902871075278e-06,
+      "loss": 0.4353,
+      "num_input_tokens_seen": 37382456,
+      "step": 680,
+      "train_runtime": 3890.1772,
+      "train_tokens_per_second": 9609.448
+    },
+    {
+      "epoch": 0.2873322147651007,
+      "grad_norm": 0.7729195427869149,
+      "learning_rate": 9.500752329046307e-06,
+      "loss": 0.3922,
+      "num_input_tokens_seen": 37672400,
+      "step": 685,
+      "train_runtime": 3923.7507,
+      "train_tokens_per_second": 9601.12
+    },
+    {
+      "epoch": 0.28942953020134227,
+      "grad_norm": 0.8225371738362072,
+      "learning_rate": 9.493552938432937e-06,
+      "loss": 0.3967,
+      "num_input_tokens_seen": 37952152,
+      "step": 690,
+      "train_runtime": 3953.6103,
+      "train_tokens_per_second": 9599.366
+    },
+    {
+      "epoch": 0.2915268456375839,
+      "grad_norm": 0.8987197678171884,
+      "learning_rate": 9.486304777373219e-06,
+      "loss": 0.3978,
+      "num_input_tokens_seen": 38222352,
+      "step": 695,
+      "train_runtime": 3984.9907,
+      "train_tokens_per_second": 9591.579
+    },
+    {
+      "epoch": 0.2936241610738255,
+      "grad_norm": 0.7877668860664033,
+      "learning_rate": 9.479007924534537e-06,
+      "loss": 0.3999,
+      "num_input_tokens_seen": 38503200,
+      "step": 700,
+      "train_runtime": 4015.6294,
+      "train_tokens_per_second": 9588.335
+    },
+    {
+      "epoch": 0.29572147651006714,
+      "grad_norm": 0.735033067084273,
+      "learning_rate": 9.471662459112747e-06,
+      "loss": 0.4023,
+      "num_input_tokens_seen": 38783440,
+      "step": 705,
+      "train_runtime": 4038.7263,
+      "train_tokens_per_second": 9602.889
+    },
+    {
+      "epoch": 0.2978187919463087,
+      "grad_norm": 0.7241118255134049,
+      "learning_rate": 9.464268460831318e-06,
+      "loss": 0.3711,
+      "num_input_tokens_seen": 39080848,
+      "step": 710,
+      "train_runtime": 4067.5534,
+      "train_tokens_per_second": 9607.95
+    },
+    {
+      "epoch": 0.2999161073825503,
+      "grad_norm": 0.8409130654833741,
+      "learning_rate": 9.456826009940469e-06,
+      "loss": 0.3884,
+      "num_input_tokens_seen": 39364744,
+      "step": 715,
+      "train_runtime": 4093.1609,
+      "train_tokens_per_second": 9617.199
+    },
+    {
+      "epoch": 0.30201342281879195,
+      "grad_norm": 0.7765662763324089,
+      "learning_rate": 9.44933518721629e-06,
+      "loss": 0.3964,
+      "num_input_tokens_seen": 39654184,
+      "step": 720,
+      "train_runtime": 4123.4382,
+      "train_tokens_per_second": 9616.777
+    },
+    {
+      "epoch": 0.30411073825503354,
+      "grad_norm": 0.8043157629933612,
+      "learning_rate": 9.441796073959875e-06,
+      "loss": 0.3923,
+      "num_input_tokens_seen": 39914496,
+      "step": 725,
+      "train_runtime": 4153.6183,
+      "train_tokens_per_second": 9609.573
+    },
+    {
+      "epoch": 0.3062080536912752,
+      "grad_norm": 0.659513961380237,
+      "learning_rate": 9.434208751996434e-06,
+      "loss": 0.3904,
+      "num_input_tokens_seen": 40201888,
+      "step": 730,
+      "train_runtime": 4184.2511,
+      "train_tokens_per_second": 9607.905
+    },
+    {
+      "epoch": 0.30830536912751677,
+      "grad_norm": 0.6660766156728498,
+      "learning_rate": 9.426573303674412e-06,
+      "loss": 0.3883,
+      "num_input_tokens_seen": 40486104,
+      "step": 735,
+      "train_runtime": 4211.7281,
+      "train_tokens_per_second": 9612.706
+    },
+    {
+      "epoch": 0.3104026845637584,
+      "grad_norm": 0.8730869298418855,
+      "learning_rate": 9.418889811864584e-06,
+      "loss": 0.4092,
+      "num_input_tokens_seen": 40731112,
+      "step": 740,
+      "train_runtime": 4240.1627,
+      "train_tokens_per_second": 9606.026
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.7803544849018209,
+      "learning_rate": 9.41115835995916e-06,
+      "loss": 0.3974,
+      "num_input_tokens_seen": 41016992,
+      "step": 745,
+      "train_runtime": 4262.2114,
+      "train_tokens_per_second": 9623.406
+    },
+    {
+      "epoch": 0.3145973154362416,
+      "grad_norm": 0.9242813007920841,
+      "learning_rate": 9.40337903187089e-06,
+      "loss": 0.4142,
+      "num_input_tokens_seen": 41300392,
+      "step": 750,
+      "train_runtime": 4287.3586,
+      "train_tokens_per_second": 9633.062
+    },
+    {
+      "epoch": 0.31669463087248323,
+      "grad_norm": 0.8941900657785933,
+      "learning_rate": 9.395551912032138e-06,
+      "loss": 0.4137,
+      "num_input_tokens_seen": 41569272,
+      "step": 755,
+      "train_runtime": 4310.6981,
+      "train_tokens_per_second": 9643.281
+    },
+    {
+      "epoch": 0.3187919463087248,
+      "grad_norm": 0.9305586964699512,
+      "learning_rate": 9.387677085393969e-06,
+      "loss": 0.4008,
+      "num_input_tokens_seen": 41817352,
+      "step": 760,
+      "train_runtime": 4336.1492,
+      "train_tokens_per_second": 9643.891
+    },
+    {
+      "epoch": 0.32088926174496646,
+      "grad_norm": 0.8194576519732724,
+      "learning_rate": 9.37975463742524e-06,
+      "loss": 0.4096,
+      "num_input_tokens_seen": 42087160,
+      "step": 765,
+      "train_runtime": 4363.0964,
+      "train_tokens_per_second": 9646.168
+    },
+    {
+      "epoch": 0.32298657718120805,
+      "grad_norm": 0.863065268283902,
+      "learning_rate": 9.371784654111656e-06,
+      "loss": 0.4045,
+      "num_input_tokens_seen": 42358792,
+      "step": 770,
+      "train_runtime": 4388.453,
+      "train_tokens_per_second": 9652.329
+    },
+    {
+      "epoch": 0.3250838926174497,
+      "grad_norm": 0.8559808675730278,
+      "learning_rate": 9.363767221954843e-06,
+      "loss": 0.4043,
+      "num_input_tokens_seen": 42636032,
+      "step": 775,
+      "train_runtime": 4413.8258,
+      "train_tokens_per_second": 9659.654
+    },
+    {
+      "epoch": 0.3271812080536913,
+      "grad_norm": 0.7827333393818555,
+      "learning_rate": 9.355702427971414e-06,
+      "loss": 0.3542,
+      "num_input_tokens_seen": 42896064,
+      "step": 780,
+      "train_runtime": 4442.7231,
+      "train_tokens_per_second": 9655.354
+    },
+    {
+      "epoch": 0.32927852348993286,
+      "grad_norm": 0.7250038471470447,
+      "learning_rate": 9.347590359692015e-06,
+      "loss": 0.38,
+      "num_input_tokens_seen": 43165240,
+      "step": 785,
+      "train_runtime": 4473.2477,
+      "train_tokens_per_second": 9649.642
+    },
+    {
+      "epoch": 0.3313758389261745,
+      "grad_norm": 0.6725302183170192,
+      "learning_rate": 9.339431105160384e-06,
+      "loss": 0.3709,
+      "num_input_tokens_seen": 43445080,
+      "step": 790,
+      "train_runtime": 4503.4914,
+      "train_tokens_per_second": 9646.977
+    },
+    {
+      "epoch": 0.3334731543624161,
+      "grad_norm": 0.7446238473601807,
+      "learning_rate": 9.331224752932388e-06,
+      "loss": 0.403,
+      "num_input_tokens_seen": 43736112,
+      "step": 795,
+      "train_runtime": 4528.9518,
+      "train_tokens_per_second": 9657.005
+    },
+    {
+      "epoch": 0.33557046979865773,
+      "grad_norm": 0.735298064228247,
+      "learning_rate": 9.32297139207507e-06,
+      "loss": 0.3927,
+      "num_input_tokens_seen": 44013560,
+      "step": 800,
+      "train_runtime": 4558.4593,
+      "train_tokens_per_second": 9655.359
+    },
+    {
+      "epoch": 0.3376677852348993,
+      "grad_norm": 0.7306235971578608,
+      "learning_rate": 9.314671112165675e-06,
+      "loss": 0.4019,
+      "num_input_tokens_seen": 44282976,
+      "step": 805,
+      "train_runtime": 4593.4967,
+      "train_tokens_per_second": 9640.363
+    },
+    {
+      "epoch": 0.33976510067114096,
+      "grad_norm": 0.8147612011867227,
+      "learning_rate": 9.306324003290677e-06,
+      "loss": 0.3787,
+      "num_input_tokens_seen": 44552888,
+      "step": 810,
+      "train_runtime": 4622.0662,
+      "train_tokens_per_second": 9639.171
+    },
+    {
+      "epoch": 0.34186241610738255,
+      "grad_norm": 0.7793655835629372,
+      "learning_rate": 9.297930156044812e-06,
+      "loss": 0.3922,
+      "num_input_tokens_seen": 44823976,
+      "step": 815,
+      "train_runtime": 4647.9267,
+      "train_tokens_per_second": 9643.865
+    },
+    {
+      "epoch": 0.34395973154362414,
+      "grad_norm": 0.700782877954151,
+      "learning_rate": 9.289489661530082e-06,
+      "loss": 0.3874,
+      "num_input_tokens_seen": 45118848,
+      "step": 820,
+      "train_runtime": 4676.4097,
+      "train_tokens_per_second": 9648.181
+    },
+    {
+      "epoch": 0.3460570469798658,
+      "grad_norm": 0.7057511236760925,
+      "learning_rate": 9.281002611354775e-06,
+      "loss": 0.3849,
+      "num_input_tokens_seen": 45398128,
+      "step": 825,
+      "train_runtime": 4701.1123,
+      "train_tokens_per_second": 9656.891
+    },
+    {
+      "epoch": 0.34815436241610737,
+      "grad_norm": 0.7295185906831485,
+      "learning_rate": 9.272469097632465e-06,
+      "loss": 0.4048,
+      "num_input_tokens_seen": 45690872,
+      "step": 830,
+      "train_runtime": 4728.7859,
+      "train_tokens_per_second": 9662.284
+    },
+    {
+      "epoch": 0.350251677852349,
+      "grad_norm": 0.7527631832969672,
+      "learning_rate": 9.263889212981015e-06,
+      "loss": 0.3943,
+      "num_input_tokens_seen": 45949720,
+      "step": 835,
+      "train_runtime": 4759.9232,
+      "train_tokens_per_second": 9653.458
+    },
+    {
+      "epoch": 0.3523489932885906,
+      "grad_norm": 0.7407325767345979,
+      "learning_rate": 9.255263050521573e-06,
+      "loss": 0.4044,
+      "num_input_tokens_seen": 46242520,
+      "step": 840,
+      "train_runtime": 4788.6171,
+      "train_tokens_per_second": 9656.759
+    },
+    {
+      "epoch": 0.35444630872483224,
+      "grad_norm": 0.7125397718217822,
+      "learning_rate": 9.246590703877559e-06,
+      "loss": 0.3808,
+      "num_input_tokens_seen": 46530208,
+      "step": 845,
+      "train_runtime": 4815.9697,
+      "train_tokens_per_second": 9661.649
+    },
+    {
+      "epoch": 0.3565436241610738,
+      "grad_norm": 0.8073694469326297,
+      "learning_rate": 9.237872267173646e-06,
+      "loss": 0.4045,
+      "num_input_tokens_seen": 46807360,
+      "step": 850,
+      "train_runtime": 4842.1361,
+      "train_tokens_per_second": 9666.676
+    },
+    {
+      "epoch": 0.3586409395973154,
+      "grad_norm": 0.7212333508918212,
+      "learning_rate": 9.229107835034751e-06,
+      "loss": 0.385,
+      "num_input_tokens_seen": 47072240,
+      "step": 855,
+      "train_runtime": 4873.2583,
+      "train_tokens_per_second": 9659.295
+    },
+    {
+      "epoch": 0.36073825503355705,
+      "grad_norm": 0.7704305706522574,
+      "learning_rate": 9.220297502584992e-06,
+      "loss": 0.3896,
+      "num_input_tokens_seen": 47354896,
+      "step": 860,
+      "train_runtime": 4899.881,
+      "train_tokens_per_second": 9664.499
+    },
+    {
+      "epoch": 0.36283557046979864,
+      "grad_norm": 0.8010389025967832,
+      "learning_rate": 9.211441365446661e-06,
+      "loss": 0.3754,
+      "num_input_tokens_seen": 47638696,
+      "step": 865,
+      "train_runtime": 4928.2002,
+      "train_tokens_per_second": 9666.551
+    },
+    {
+      "epoch": 0.3649328859060403,
+      "grad_norm": 0.8040745157324941,
+      "learning_rate": 9.202539519739195e-06,
+      "loss": 0.3821,
+      "num_input_tokens_seen": 47916672,
+      "step": 870,
+      "train_runtime": 4954.4371,
+      "train_tokens_per_second": 9671.466
+    },
+    {
+      "epoch": 0.36703020134228187,
+      "grad_norm": 0.7671022816918951,
+      "learning_rate": 9.19359206207812e-06,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 48183816,
+      "step": 875,
+      "train_runtime": 4981.0015,
+      "train_tokens_per_second": 9673.52
+    },
+    {
+      "epoch": 0.3691275167785235,
+      "grad_norm": 0.7578572593506936,
+      "learning_rate": 9.184599089574011e-06,
+      "loss": 0.3978,
+      "num_input_tokens_seen": 48453928,
+      "step": 880,
+      "train_runtime": 5007.4675,
+      "train_tokens_per_second": 9676.334
+    },
+    {
+      "epoch": 0.3712248322147651,
+      "grad_norm": 0.8270141617173937,
+      "learning_rate": 9.175560699831431e-06,
+      "loss": 0.357,
+      "num_input_tokens_seen": 48713176,
+      "step": 885,
+      "train_runtime": 5033.4748,
+      "train_tokens_per_second": 9677.842
+    },
+    {
+      "epoch": 0.3733221476510067,
+      "grad_norm": 0.7450013467019271,
+      "learning_rate": 9.16647699094788e-06,
+      "loss": 0.4029,
+      "num_input_tokens_seen": 48967112,
+      "step": 890,
+      "train_runtime": 5057.6175,
+      "train_tokens_per_second": 9681.854
+    },
+    {
+      "epoch": 0.37541946308724833,
+      "grad_norm": 0.6772066720558229,
+      "learning_rate": 9.157348061512728e-06,
+      "loss": 0.3901,
+      "num_input_tokens_seen": 49255736,
+      "step": 895,
+      "train_runtime": 5086.0481,
+      "train_tokens_per_second": 9684.481
+    },
+    {
+      "epoch": 0.3775167785234899,
+      "grad_norm": 0.7589136031085396,
+      "learning_rate": 9.148174010606132e-06,
+      "loss": 0.3716,
+      "num_input_tokens_seen": 49526144,
+      "step": 900,
+      "train_runtime": 5107.9693,
+      "train_tokens_per_second": 9695.858
+    },
+    {
+      "epoch": 0.37961409395973156,
+      "grad_norm": 0.7107398210556067,
+      "learning_rate": 9.138954937797983e-06,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 49811896,
+      "step": 905,
+      "train_runtime": 5137.998,
+      "train_tokens_per_second": 9694.806
+    },
+    {
+      "epoch": 0.38171140939597314,
+      "grad_norm": 0.6871535286194669,
+      "learning_rate": 9.12969094314681e-06,
+      "loss": 0.374,
+      "num_input_tokens_seen": 50075336,
+      "step": 910,
+      "train_runtime": 5170.7756,
+      "train_tokens_per_second": 9684.299
+    },
+    {
+      "epoch": 0.3838087248322148,
+      "grad_norm": 0.7541190054504451,
+      "learning_rate": 9.120382127198696e-06,
+      "loss": 0.4129,
+      "num_input_tokens_seen": 50363680,
+      "step": 915,
+      "train_runtime": 5197.8966,
+      "train_tokens_per_second": 9689.242
+    },
+    {
+      "epoch": 0.3859060402684564,
+      "grad_norm": 0.7505494005023116,
+      "learning_rate": 9.111028590986193e-06,
+      "loss": 0.4031,
+      "num_input_tokens_seen": 50600184,
+      "step": 920,
+      "train_runtime": 5227.8426,
+      "train_tokens_per_second": 9678.98
+    },
+    {
+      "epoch": 0.38800335570469796,
+      "grad_norm": 0.6931499203152481,
+      "learning_rate": 9.101630436027216e-06,
+      "loss": 0.3682,
+      "num_input_tokens_seen": 50876168,
+      "step": 925,
+      "train_runtime": 5253.3801,
+      "train_tokens_per_second": 9684.464
+    },
+    {
+      "epoch": 0.3901006711409396,
+      "grad_norm": 0.7484643259987901,
+      "learning_rate": 9.092187764323948e-06,
+      "loss": 0.4125,
+      "num_input_tokens_seen": 51168248,
+      "step": 930,
+      "train_runtime": 5273.9026,
+      "train_tokens_per_second": 9702.16
+    },
+    {
+      "epoch": 0.3921979865771812,
+      "grad_norm": 1.2733102114734989,
+      "learning_rate": 9.082700678361733e-06,
+      "loss": 0.3973,
+      "num_input_tokens_seen": 51438960,
+      "step": 935,
+      "train_runtime": 5297.919,
+      "train_tokens_per_second": 9709.276
+    },
+    {
+      "epoch": 0.39429530201342283,
+      "grad_norm": 0.7416209916966714,
+      "learning_rate": 9.073169281107963e-06,
+      "loss": 0.3839,
+      "num_input_tokens_seen": 51696632,
+      "step": 940,
+      "train_runtime": 5330.331,
+      "train_tokens_per_second": 9698.578
+    },
+    {
+      "epoch": 0.3963926174496644,
+      "grad_norm": 0.7858684870919205,
+      "learning_rate": 9.063593676010954e-06,
+      "loss": 0.3955,
+      "num_input_tokens_seen": 51952184,
+      "step": 945,
+      "train_runtime": 5358.6256,
+      "train_tokens_per_second": 9695.058
+    },
+    {
+      "epoch": 0.39848993288590606,
+      "grad_norm": 0.6978057860876561,
+      "learning_rate": 9.053973966998831e-06,
+      "loss": 0.3792,
+      "num_input_tokens_seen": 52248968,
+      "step": 950,
+      "train_runtime": 5378.2386,
+      "train_tokens_per_second": 9714.885
+    },
+    {
+      "epoch": 0.40058724832214765,
+      "grad_norm": 0.8399641555413809,
+      "learning_rate": 9.044310258478402e-06,
+      "loss": 0.4232,
+      "num_input_tokens_seen": 52509408,
+      "step": 955,
+      "train_runtime": 5405.7838,
+      "train_tokens_per_second": 9713.561
+    },
+    {
+      "epoch": 0.40268456375838924,
+      "grad_norm": 0.7106422151261867,
+      "learning_rate": 9.034602655334013e-06,
+      "loss": 0.3827,
+      "num_input_tokens_seen": 52784304,
+      "step": 960,
+      "train_runtime": 5433.5979,
+      "train_tokens_per_second": 9714.43
+    },
+    {
+      "epoch": 0.4047818791946309,
+      "grad_norm": 0.670163260535576,
+      "learning_rate": 9.024851262926425e-06,
+      "loss": 0.3881,
+      "num_input_tokens_seen": 53052976,
+      "step": 965,
+      "train_runtime": 5461.5174,
+      "train_tokens_per_second": 9713.963
+    },
+    {
+      "epoch": 0.40687919463087246,
+      "grad_norm": 0.7367005448874322,
+      "learning_rate": 9.015056187091654e-06,
+      "loss": 0.3945,
+      "num_input_tokens_seen": 53323896,
+      "step": 970,
+      "train_runtime": 5491.5785,
+      "train_tokens_per_second": 9710.122
+    },
+    {
+      "epoch": 0.4089765100671141,
+      "grad_norm": 0.7350527058270782,
+      "learning_rate": 9.005217534139837e-06,
+      "loss": 0.3708,
+      "num_input_tokens_seen": 53591360,
+      "step": 975,
+      "train_runtime": 5516.3968,
+      "train_tokens_per_second": 9714.921
+    },
+    {
+      "epoch": 0.4110738255033557,
+      "grad_norm": 0.681737207965053,
+      "learning_rate": 8.99533541085407e-06,
+      "loss": 0.3742,
+      "num_input_tokens_seen": 53870160,
+      "step": 980,
+      "train_runtime": 5549.4529,
+      "train_tokens_per_second": 9707.292
+    },
+    {
+      "epoch": 0.41317114093959734,
+      "grad_norm": 0.7937119102350605,
+      "learning_rate": 8.985409924489252e-06,
+      "loss": 0.3899,
+      "num_input_tokens_seen": 54158864,
+      "step": 985,
+      "train_runtime": 5575.7689,
+      "train_tokens_per_second": 9713.255
+    },
+    {
+      "epoch": 0.4152684563758389,
+      "grad_norm": 0.7739462732817929,
+      "learning_rate": 8.97544118277092e-06,
+      "loss": 0.4225,
+      "num_input_tokens_seen": 54426272,
+      "step": 990,
+      "train_runtime": 5607.0617,
+      "train_tokens_per_second": 9706.737
+    },
+    {
+      "epoch": 0.4173657718120805,
+      "grad_norm": 1.1883156817273215,
+      "learning_rate": 8.965429293894078e-06,
+      "loss": 0.3942,
+      "num_input_tokens_seen": 54717896,
+      "step": 995,
+      "train_runtime": 5640.1067,
+      "train_tokens_per_second": 9701.571
+    },
+    {
+      "epoch": 0.41946308724832215,
+      "grad_norm": 0.783492129574589,
+      "learning_rate": 8.955374366522025e-06,
+      "loss": 0.3793,
+      "num_input_tokens_seen": 54964168,
+      "step": 1000,
+      "train_runtime": 5664.1797,
+      "train_tokens_per_second": 9703.818
+    },
+    {
+      "epoch": 0.42156040268456374,
+      "grad_norm": 0.7396886784890098,
+      "learning_rate": 8.945276509785177e-06,
+      "loss": 0.3859,
+      "num_input_tokens_seen": 55231656,
+      "step": 1005,
+      "train_runtime": 5693.9471,
+      "train_tokens_per_second": 9700.065
+    },
+    {
+      "epoch": 0.4236577181208054,
+      "grad_norm": 0.7351523436136801,
+      "learning_rate": 8.935135833279878e-06,
+      "loss": 0.3851,
+      "num_input_tokens_seen": 55505168,
+      "step": 1010,
+      "train_runtime": 5727.3985,
+      "train_tokens_per_second": 9691.166
+    },
+    {
+      "epoch": 0.42575503355704697,
+      "grad_norm": 0.7224007387282775,
+      "learning_rate": 8.924952447067216e-06,
+      "loss": 0.3617,
+      "num_input_tokens_seen": 55792296,
+      "step": 1015,
+      "train_runtime": 5751.0315,
+      "train_tokens_per_second": 9701.268
+    },
+    {
+      "epoch": 0.4278523489932886,
+      "grad_norm": 0.6949732830135107,
+      "learning_rate": 8.914726461671824e-06,
+      "loss": 0.3833,
+      "num_input_tokens_seen": 56055616,
+      "step": 1020,
+      "train_runtime": 5778.6711,
+      "train_tokens_per_second": 9700.434
+    },
+    {
+      "epoch": 0.4299496644295302,
+      "grad_norm": 0.741041743508846,
+      "learning_rate": 8.904457988080682e-06,
+      "loss": 0.3832,
+      "num_input_tokens_seen": 56314760,
+      "step": 1025,
+      "train_runtime": 5810.6171,
+      "train_tokens_per_second": 9691.7
+    },
+    {
+      "epoch": 0.4320469798657718,
+      "grad_norm": 0.703436315629902,
+      "learning_rate": 8.894147137741913e-06,
+      "loss": 0.3977,
+      "num_input_tokens_seen": 56607416,
+      "step": 1030,
+      "train_runtime": 5844.1045,
+      "train_tokens_per_second": 9686.243
+    },
+    {
+      "epoch": 0.4341442953020134,
+      "grad_norm": 0.814772672664591,
+      "learning_rate": 8.883794022563575e-06,
+      "loss": 0.3818,
+      "num_input_tokens_seen": 56879368,
+      "step": 1035,
+      "train_runtime": 5873.32,
+      "train_tokens_per_second": 9684.364
+    },
+    {
+      "epoch": 0.436241610738255,
+      "grad_norm": 0.6986565920830929,
+      "learning_rate": 8.873398754912437e-06,
+      "loss": 0.3781,
+      "num_input_tokens_seen": 57131800,
+      "step": 1040,
+      "train_runtime": 5903.3276,
+      "train_tokens_per_second": 9677.898
+    },
+    {
+      "epoch": 0.43833892617449666,
+      "grad_norm": 0.6933055378583148,
+      "learning_rate": 8.862961447612777e-06,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 57439864,
+      "step": 1045,
+      "train_runtime": 5927.6493,
+      "train_tokens_per_second": 9690.159
+    },
+    {
+      "epoch": 0.44043624161073824,
+      "grad_norm": 1.15222914243957,
+      "learning_rate": 8.852482213945142e-06,
+      "loss": 0.386,
+      "num_input_tokens_seen": 57728312,
+      "step": 1050,
+      "train_runtime": 5953.3963,
+      "train_tokens_per_second": 9696.702
+    },
+    {
+      "epoch": 0.4425335570469799,
+      "grad_norm": 0.7367475114835332,
+      "learning_rate": 8.84196116764512e-06,
+      "loss": 0.3733,
+      "num_input_tokens_seen": 58020072,
+      "step": 1055,
+      "train_runtime": 5976.7623,
+      "train_tokens_per_second": 9707.609
+    },
+    {
+      "epoch": 0.4446308724832215,
+      "grad_norm": 0.6924526989611216,
+      "learning_rate": 8.831398422902118e-06,
+      "loss": 0.3709,
+      "num_input_tokens_seen": 58279480,
+      "step": 1060,
+      "train_runtime": 6000.5957,
+      "train_tokens_per_second": 9712.282
+    },
+    {
+      "epoch": 0.44672818791946306,
+      "grad_norm": 0.7209430555754343,
+      "learning_rate": 8.820794094358105e-06,
+      "loss": 0.3759,
+      "num_input_tokens_seen": 58545144,
+      "step": 1065,
+      "train_runtime": 6033.3925,
+      "train_tokens_per_second": 9703.52
+    },
+    {
+      "epoch": 0.4488255033557047,
+      "grad_norm": 0.7828840805620894,
+      "learning_rate": 8.810148297106383e-06,
+      "loss": 0.4011,
+      "num_input_tokens_seen": 58799160,
+      "step": 1070,
+      "train_runtime": 6069.0135,
+      "train_tokens_per_second": 9688.421
+    },
+    {
+      "epoch": 0.4509228187919463,
+      "grad_norm": 0.7503336279875492,
+      "learning_rate": 8.79946114669033e-06,
+      "loss": 0.3879,
+      "num_input_tokens_seen": 59055296,
+      "step": 1075,
+      "train_runtime": 6098.5779,
+      "train_tokens_per_second": 9683.454
+    },
+    {
+      "epoch": 0.45302013422818793,
+      "grad_norm": 0.7082110986455266,
+      "learning_rate": 8.788732759102146e-06,
+      "loss": 0.388,
+      "num_input_tokens_seen": 59350440,
+      "step": 1080,
+      "train_runtime": 6125.0259,
+      "train_tokens_per_second": 9689.827
+    },
+    {
+      "epoch": 0.4551174496644295,
+      "grad_norm": 0.6814374216575967,
+      "learning_rate": 8.777963250781599e-06,
+      "loss": 0.3558,
+      "num_input_tokens_seen": 59637312,
+      "step": 1085,
+      "train_runtime": 6153.7512,
+      "train_tokens_per_second": 9691.213
+    },
+    {
+      "epoch": 0.45721476510067116,
+      "grad_norm": 0.8445623203916546,
+      "learning_rate": 8.767152738614751e-06,
+      "loss": 0.3896,
+      "num_input_tokens_seen": 59904976,
+      "step": 1090,
+      "train_runtime": 6182.9193,
+      "train_tokens_per_second": 9688.785
+    },
+    {
+      "epoch": 0.45931208053691275,
+      "grad_norm": 0.7577998624127624,
+      "learning_rate": 8.756301339932705e-06,
+      "loss": 0.382,
+      "num_input_tokens_seen": 60178608,
+      "step": 1095,
+      "train_runtime": 6208.5658,
+      "train_tokens_per_second": 9692.836
+    },
+    {
+      "epoch": 0.46140939597315433,
+      "grad_norm": 0.7214470874336397,
+      "learning_rate": 8.745409172510315e-06,
+      "loss": 0.3689,
+      "num_input_tokens_seen": 60422376,
+      "step": 1100,
+      "train_runtime": 6232.909,
+      "train_tokens_per_second": 9694.089
+    },
+    {
+      "epoch": 0.463506711409396,
+      "grad_norm": 0.7336174256751595,
+      "learning_rate": 8.734476354564924e-06,
+      "loss": 0.4005,
+      "num_input_tokens_seen": 60683256,
+      "step": 1105,
+      "train_runtime": 6261.1969,
+      "train_tokens_per_second": 9691.958
+    },
+    {
+      "epoch": 0.46560402684563756,
+      "grad_norm": 0.7096137844814188,
+      "learning_rate": 8.723503004755064e-06,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 60974624,
+      "step": 1110,
+      "train_runtime": 6290.815,
+      "train_tokens_per_second": 9692.643
+    },
+    {
+      "epoch": 0.4677013422818792,
+      "grad_norm": 0.8021829038101091,
+      "learning_rate": 8.71248924217918e-06,
+      "loss": 0.3722,
+      "num_input_tokens_seen": 61262752,
+      "step": 1115,
+      "train_runtime": 6315.5819,
+      "train_tokens_per_second": 9700.254
+    },
+    {
+      "epoch": 0.4697986577181208,
+      "grad_norm": 0.7460572916989756,
+      "learning_rate": 8.701435186374333e-06,
+      "loss": 0.3603,
+      "num_input_tokens_seen": 61516168,
+      "step": 1120,
+      "train_runtime": 6348.7575,
+      "train_tokens_per_second": 9689.481
+    },
+    {
+      "epoch": 0.47189597315436244,
+      "grad_norm": 0.7926367505002561,
+      "learning_rate": 8.69034095731491e-06,
+      "loss": 0.3632,
+      "num_input_tokens_seen": 61766648,
+      "step": 1125,
+      "train_runtime": 6372.1561,
+      "train_tokens_per_second": 9693.21
+    },
+    {
+      "epoch": 0.473993288590604,
+      "grad_norm": 0.7185094872054525,
+      "learning_rate": 8.679206675411305e-06,
+      "loss": 0.3782,
+      "num_input_tokens_seen": 62040064,
+      "step": 1130,
+      "train_runtime": 6394.5221,
+      "train_tokens_per_second": 9702.064
+    },
+    {
+      "epoch": 0.47609060402684567,
+      "grad_norm": 0.7479313112887285,
+      "learning_rate": 8.66803246150863e-06,
+      "loss": 0.3856,
+      "num_input_tokens_seen": 62311976,
+      "step": 1135,
+      "train_runtime": 6422.9114,
+      "train_tokens_per_second": 9701.516
+    },
+    {
+      "epoch": 0.47818791946308725,
+      "grad_norm": 0.76927597286747,
+      "learning_rate": 8.656818436885395e-06,
+      "loss": 0.4017,
+      "num_input_tokens_seen": 62573384,
+      "step": 1140,
+      "train_runtime": 6457.6273,
+      "train_tokens_per_second": 9689.841
+    },
+    {
+      "epoch": 0.48028523489932884,
+      "grad_norm": 0.6622334400575087,
+      "learning_rate": 8.645564723252188e-06,
+      "loss": 0.3695,
+      "num_input_tokens_seen": 62832704,
+      "step": 1145,
+      "train_runtime": 6485.9391,
+      "train_tokens_per_second": 9687.526
+    },
+    {
+      "epoch": 0.4823825503355705,
+      "grad_norm": 0.7505737333811565,
+      "learning_rate": 8.634271442750364e-06,
+      "loss": 0.3587,
+      "num_input_tokens_seen": 63097216,
+      "step": 1150,
+      "train_runtime": 6515.2264,
+      "train_tokens_per_second": 9684.578
+    },
+    {
+      "epoch": 0.48447986577181207,
+      "grad_norm": 0.7540299766106143,
+      "learning_rate": 8.622938717950714e-06,
+      "loss": 0.3727,
+      "num_input_tokens_seen": 63369088,
+      "step": 1155,
+      "train_runtime": 6540.7726,
+      "train_tokens_per_second": 9688.319
+    },
+    {
+      "epoch": 0.4865771812080537,
+      "grad_norm": 0.8137301796576052,
+      "learning_rate": 8.61156667185213e-06,
+      "loss": 0.3862,
+      "num_input_tokens_seen": 63627712,
+      "step": 1160,
+      "train_runtime": 6568.0613,
+      "train_tokens_per_second": 9687.442
+    },
+    {
+      "epoch": 0.4886744966442953,
+      "grad_norm": 0.727553323595707,
+      "learning_rate": 8.600155427880282e-06,
+      "loss": 0.3575,
+      "num_input_tokens_seen": 63914064,
+      "step": 1165,
+      "train_runtime": 6597.2169,
+      "train_tokens_per_second": 9688.034
+    },
+    {
+      "epoch": 0.49077181208053694,
+      "grad_norm": 0.7652954918777157,
+      "learning_rate": 8.588705109886264e-06,
+      "loss": 0.3868,
+      "num_input_tokens_seen": 64162848,
+      "step": 1170,
+      "train_runtime": 6618.3478,
+      "train_tokens_per_second": 9694.693
+    },
+    {
+      "epoch": 0.4928691275167785,
+      "grad_norm": 0.7152149626740316,
+      "learning_rate": 8.57721584214526e-06,
+      "loss": 0.3891,
+      "num_input_tokens_seen": 64450688,
+      "step": 1175,
+      "train_runtime": 6645.0064,
+      "train_tokens_per_second": 9699.116
+    },
+    {
+      "epoch": 0.4949664429530201,
+      "grad_norm": 0.7970935746191729,
+      "learning_rate": 8.565687749355194e-06,
+      "loss": 0.3577,
+      "num_input_tokens_seen": 64726704,
+      "step": 1180,
+      "train_runtime": 6676.3116,
+      "train_tokens_per_second": 9694.979
+    },
+    {
+      "epoch": 0.49706375838926176,
+      "grad_norm": 0.8159345438834689,
+      "learning_rate": 8.554120956635375e-06,
+      "loss": 0.3884,
+      "num_input_tokens_seen": 65013152,
+      "step": 1185,
+      "train_runtime": 6710.6391,
+      "train_tokens_per_second": 9688.072
+    },
+    {
+      "epoch": 0.49916107382550334,
+      "grad_norm": 0.7663106676121332,
+      "learning_rate": 8.542515589525134e-06,
+      "loss": 0.389,
+      "num_input_tokens_seen": 65258696,
+      "step": 1190,
+      "train_runtime": 6741.7669,
+      "train_tokens_per_second": 9679.761
+    },
+    {
+      "epoch": 0.5012583892617449,
+      "grad_norm": 0.7741516724130791,
+      "learning_rate": 8.53087177398247e-06,
+      "loss": 0.3713,
+      "num_input_tokens_seen": 65551056,
+      "step": 1195,
+      "train_runtime": 6768.5209,
+      "train_tokens_per_second": 9684.694
+    },
+    {
+      "epoch": 0.5033557046979866,
+      "grad_norm": 0.6610080019588243,
+      "learning_rate": 8.519189636382678e-06,
+      "loss": 0.3808,
+      "num_input_tokens_seen": 65829400,
+      "step": 1200,
+      "train_runtime": 6796.8018,
+      "train_tokens_per_second": 9685.349
+    },
+    {
+      "epoch": 0.5054530201342282,
+      "grad_norm": 0.7184483405410419,
+      "learning_rate": 8.507469303516984e-06,
+      "loss": 0.4043,
+      "num_input_tokens_seen": 66135056,
+      "step": 1205,
+      "train_runtime": 6830.9847,
+      "train_tokens_per_second": 9681.628
+    },
+    {
+      "epoch": 0.5075503355704698,
+      "grad_norm": 0.7365013270296795,
+      "learning_rate": 8.495710902591153e-06,
+      "loss": 0.3926,
+      "num_input_tokens_seen": 66422216,
+      "step": 1210,
+      "train_runtime": 6859.893,
+      "train_tokens_per_second": 9682.69
+    },
+    {
+      "epoch": 0.5096476510067114,
+      "grad_norm": 0.6200867212998569,
+      "learning_rate": 8.483914561224127e-06,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 66697632,
+      "step": 1215,
+      "train_runtime": 6882.4619,
+      "train_tokens_per_second": 9690.955
+    },
+    {
+      "epoch": 0.511744966442953,
+      "grad_norm": 0.6766639593917874,
+      "learning_rate": 8.47208040744663e-06,
+      "loss": 0.3513,
+      "num_input_tokens_seen": 66991224,
+      "step": 1220,
+      "train_runtime": 6907.1632,
+      "train_tokens_per_second": 9698.804
+    },
+    {
+      "epoch": 0.5138422818791947,
+      "grad_norm": 0.6905233290078246,
+      "learning_rate": 8.460208569699779e-06,
+      "loss": 0.3932,
+      "num_input_tokens_seen": 67269920,
+      "step": 1225,
+      "train_runtime": 6939.0255,
+      "train_tokens_per_second": 9694.433
+    },
+    {
+      "epoch": 0.5159395973154363,
+      "grad_norm": 0.6720820193650875,
+      "learning_rate": 8.448299176833692e-06,
+      "loss": 0.3974,
+      "num_input_tokens_seen": 67517144,
+      "step": 1230,
+      "train_runtime": 6968.9454,
+      "train_tokens_per_second": 9688.287
+    },
+    {
+      "epoch": 0.5180369127516778,
+      "grad_norm": 0.8920513343405098,
+      "learning_rate": 8.436352358106089e-06,
+      "loss": 0.4041,
+      "num_input_tokens_seen": 67777104,
+      "step": 1235,
+      "train_runtime": 6990.9768,
+      "train_tokens_per_second": 9694.94
+    },
+    {
+      "epoch": 0.5201342281879194,
+      "grad_norm": 0.8701135534199622,
+      "learning_rate": 8.424368243180888e-06,
+      "loss": 0.3945,
+      "num_input_tokens_seen": 68028560,
+      "step": 1240,
+      "train_runtime": 7018.8562,
+      "train_tokens_per_second": 9692.257
+    },
+    {
+      "epoch": 0.522231543624161,
+      "grad_norm": 0.7282145739086415,
+      "learning_rate": 8.412346962126798e-06,
+      "loss": 0.404,
+      "num_input_tokens_seen": 68313968,
+      "step": 1245,
+      "train_runtime": 7040.6306,
+      "train_tokens_per_second": 9702.82
+    },
+    {
+      "epoch": 0.5243288590604027,
+      "grad_norm": 0.7329602028919602,
+      "learning_rate": 8.400288645415913e-06,
+      "loss": 0.3652,
+      "num_input_tokens_seen": 68564256,
+      "step": 1250,
+      "train_runtime": 7068.5177,
+      "train_tokens_per_second": 9699.948
+    },
+    {
+      "epoch": 0.5264261744966443,
+      "grad_norm": 0.7515132410103944,
+      "learning_rate": 8.388193423922282e-06,
+      "loss": 0.3732,
+      "num_input_tokens_seen": 68838864,
+      "step": 1255,
+      "train_runtime": 7092.208,
+      "train_tokens_per_second": 9706.267
+    },
+    {
+      "epoch": 0.5285234899328859,
+      "grad_norm": 0.7273795907965857,
+      "learning_rate": 8.376061428920502e-06,
+      "loss": 0.3653,
+      "num_input_tokens_seen": 69106600,
+      "step": 1260,
+      "train_runtime": 7113.882,
+      "train_tokens_per_second": 9714.33
+    },
+    {
+      "epoch": 0.5306208053691275,
+      "grad_norm": 0.7087131743213766,
+      "learning_rate": 8.363892792084291e-06,
+      "loss": 0.3793,
+      "num_input_tokens_seen": 69376448,
+      "step": 1265,
+      "train_runtime": 7143.4291,
+      "train_tokens_per_second": 9711.925
+    },
+    {
+      "epoch": 0.5327181208053692,
+      "grad_norm": 0.7225088416260722,
+      "learning_rate": 8.351687645485053e-06,
+      "loss": 0.3742,
+      "num_input_tokens_seen": 69646952,
+      "step": 1270,
+      "train_runtime": 7173.0521,
+      "train_tokens_per_second": 9709.528
+    },
+    {
+      "epoch": 0.5348154362416108,
+      "grad_norm": 0.6762469504443263,
+      "learning_rate": 8.33944612159045e-06,
+      "loss": 0.3909,
+      "num_input_tokens_seen": 69930000,
+      "step": 1275,
+      "train_runtime": 7204.9085,
+      "train_tokens_per_second": 9705.883
+    },
+    {
+      "epoch": 0.5369127516778524,
+      "grad_norm": 0.8390779346862972,
+      "learning_rate": 8.32716835326296e-06,
+      "loss": 0.4012,
+      "num_input_tokens_seen": 70230744,
+      "step": 1280,
+      "train_runtime": 7237.0484,
+      "train_tokens_per_second": 9704.335
+    },
+    {
+      "epoch": 0.5390100671140939,
+      "grad_norm": 0.7858531305559264,
+      "learning_rate": 8.314854473758438e-06,
+      "loss": 0.391,
+      "num_input_tokens_seen": 70504184,
+      "step": 1285,
+      "train_runtime": 7258.1303,
+      "train_tokens_per_second": 9713.822
+    },
+    {
+      "epoch": 0.5411073825503355,
+      "grad_norm": 0.7012840552938745,
+      "learning_rate": 8.302504616724672e-06,
+      "loss": 0.3631,
+      "num_input_tokens_seen": 70764312,
+      "step": 1290,
+      "train_runtime": 7284.3648,
+      "train_tokens_per_second": 9714.548
+    },
+    {
+      "epoch": 0.5432046979865772,
+      "grad_norm": 0.7317573426572039,
+      "learning_rate": 8.290118916199924e-06,
+      "loss": 0.3867,
+      "num_input_tokens_seen": 71054544,
+      "step": 1295,
+      "train_runtime": 7314.0669,
+      "train_tokens_per_second": 9714.779
+    },
+    {
+      "epoch": 0.5453020134228188,
+      "grad_norm": 0.6135958259155392,
+      "learning_rate": 8.277697506611489e-06,
+      "loss": 0.3684,
+      "num_input_tokens_seen": 71338848,
+      "step": 1300,
+      "train_runtime": 7342.224,
+      "train_tokens_per_second": 9716.245
+    },
+    {
+      "epoch": 0.5473993288590604,
+      "grad_norm": 0.6458462016270743,
+      "learning_rate": 8.265240522774213e-06,
+      "loss": 0.3641,
+      "num_input_tokens_seen": 71611304,
+      "step": 1305,
+      "train_runtime": 7371.7243,
+      "train_tokens_per_second": 9714.322
+    },
+    {
+      "epoch": 0.549496644295302,
+      "grad_norm": 0.6621726881218407,
+      "learning_rate": 8.252748099889065e-06,
+      "loss": 0.3808,
+      "num_input_tokens_seen": 71917312,
+      "step": 1310,
+      "train_runtime": 7401.1996,
+      "train_tokens_per_second": 9716.98
+    },
+    {
+      "epoch": 0.5515939597315436,
+      "grad_norm": 0.7266452291668771,
+      "learning_rate": 8.24022037354163e-06,
+      "loss": 0.3871,
+      "num_input_tokens_seen": 72165744,
+      "step": 1315,
+      "train_runtime": 7430.9372,
+      "train_tokens_per_second": 9711.527
+    },
+    {
+      "epoch": 0.5536912751677853,
+      "grad_norm": 0.7492688477848389,
+      "learning_rate": 8.227657479700669e-06,
+      "loss": 0.4087,
+      "num_input_tokens_seen": 72451344,
+      "step": 1320,
+      "train_runtime": 7456.2071,
+      "train_tokens_per_second": 9716.917
+    },
+    {
+      "epoch": 0.5557885906040269,
+      "grad_norm": 0.7326845727125353,
+      "learning_rate": 8.215059554716624e-06,
+      "loss": 0.3912,
+      "num_input_tokens_seen": 72741944,
+      "step": 1325,
+      "train_runtime": 7483.4809,
+      "train_tokens_per_second": 9720.335
+    },
+    {
+      "epoch": 0.5578859060402684,
+      "grad_norm": 0.7139320924878949,
+      "learning_rate": 8.202426735320148e-06,
+      "loss": 0.3748,
+      "num_input_tokens_seen": 73017880,
+      "step": 1330,
+      "train_runtime": 7514.2357,
+      "train_tokens_per_second": 9717.273
+    },
+    {
+      "epoch": 0.55998322147651,
+      "grad_norm": 0.7341355688717331,
+      "learning_rate": 8.189759158620618e-06,
+      "loss": 0.3955,
+      "num_input_tokens_seen": 73296736,
+      "step": 1335,
+      "train_runtime": 7541.4181,
+      "train_tokens_per_second": 9719.225
+    },
+    {
+      "epoch": 0.5620805369127517,
+      "grad_norm": 0.785050287546258,
+      "learning_rate": 8.177056962104645e-06,
+      "loss": 0.3889,
+      "num_input_tokens_seen": 73562992,
+      "step": 1340,
+      "train_runtime": 7575.2989,
+      "train_tokens_per_second": 9710.903
+    },
+    {
+      "epoch": 0.5641778523489933,
+      "grad_norm": 0.7224760265762157,
+      "learning_rate": 8.164320283634585e-06,
+      "loss": 0.3758,
+      "num_input_tokens_seen": 73835624,
+      "step": 1345,
+      "train_runtime": 7607.5571,
+      "train_tokens_per_second": 9705.563
+    },
+    {
+      "epoch": 0.5662751677852349,
+      "grad_norm": 0.6790823540960922,
+      "learning_rate": 8.15154926144704e-06,
+      "loss": 0.365,
+      "num_input_tokens_seen": 74115312,
+      "step": 1350,
+      "train_runtime": 7637.4504,
+      "train_tokens_per_second": 9704.196
+    },
+    {
+      "epoch": 0.5683724832214765,
+      "grad_norm": 0.7823531653263242,
+      "learning_rate": 8.138744034151365e-06,
+      "loss": 0.3849,
+      "num_input_tokens_seen": 74375360,
+      "step": 1355,
+      "train_runtime": 7657.5454,
+      "train_tokens_per_second": 9712.689
+    },
+    {
+      "epoch": 0.5704697986577181,
+      "grad_norm": 0.667769876183174,
+      "learning_rate": 8.125904740728146e-06,
+      "loss": 0.4002,
+      "num_input_tokens_seen": 74651816,
+      "step": 1360,
+      "train_runtime": 7684.5878,
+      "train_tokens_per_second": 9714.485
+    },
+    {
+      "epoch": 0.5725671140939598,
+      "grad_norm": 0.776858939198694,
+      "learning_rate": 8.113031520527714e-06,
+      "loss": 0.3786,
+      "num_input_tokens_seen": 74915792,
+      "step": 1365,
+      "train_runtime": 7716.7196,
+      "train_tokens_per_second": 9708.243
+    },
+    {
+      "epoch": 0.5746644295302014,
+      "grad_norm": 0.7930792131274486,
+      "learning_rate": 8.100124513268616e-06,
+      "loss": 0.3779,
+      "num_input_tokens_seen": 75186856,
+      "step": 1370,
+      "train_runtime": 7746.7323,
+      "train_tokens_per_second": 9705.622
+    },
+    {
+      "epoch": 0.576761744966443,
+      "grad_norm": 0.7012605649266624,
+      "learning_rate": 8.087183859036107e-06,
+      "loss": 0.3686,
+      "num_input_tokens_seen": 75469264,
+      "step": 1375,
+      "train_runtime": 7776.9998,
+      "train_tokens_per_second": 9704.162
+    },
+    {
+      "epoch": 0.5788590604026845,
+      "grad_norm": 0.6817915051023459,
+      "learning_rate": 8.074209698280627e-06,
+      "loss": 0.3487,
+      "num_input_tokens_seen": 75728952,
+      "step": 1380,
+      "train_runtime": 7798.1468,
+      "train_tokens_per_second": 9711.147
+    },
+    {
+      "epoch": 0.5809563758389261,
+      "grad_norm": 0.8389656454469276,
+      "learning_rate": 8.061202171816273e-06,
+      "loss": 0.372,
+      "num_input_tokens_seen": 76005320,
+      "step": 1385,
+      "train_runtime": 7822.1325,
+      "train_tokens_per_second": 9716.701
+    },
+    {
+      "epoch": 0.5830536912751678,
+      "grad_norm": 0.7581774277933364,
+      "learning_rate": 8.04816142081928e-06,
+      "loss": 0.3461,
+      "num_input_tokens_seen": 76239992,
+      "step": 1390,
+      "train_runtime": 7849.7433,
+      "train_tokens_per_second": 9712.418
+    },
+    {
+      "epoch": 0.5851510067114094,
+      "grad_norm": 0.7026810669368263,
+      "learning_rate": 8.035087586826477e-06,
+      "loss": 0.3639,
+      "num_input_tokens_seen": 76524048,
+      "step": 1395,
+      "train_runtime": 7875.4589,
+      "train_tokens_per_second": 9716.773
+    },
+    {
+      "epoch": 0.587248322147651,
+      "grad_norm": 0.6303234473092231,
+      "learning_rate": 8.021980811733761e-06,
+      "loss": 0.3697,
+      "num_input_tokens_seen": 76825256,
+      "step": 1400,
+      "train_runtime": 7909.8336,
+      "train_tokens_per_second": 9712.626
+    },
+    {
+      "epoch": 0.5893456375838926,
+      "grad_norm": 0.7178791581054851,
+      "learning_rate": 8.008841237794551e-06,
+      "loss": 0.3697,
+      "num_input_tokens_seen": 77096800,
+      "step": 1405,
+      "train_runtime": 7935.2107,
+      "train_tokens_per_second": 9715.785
+    },
+    {
+      "epoch": 0.5914429530201343,
+      "grad_norm": 0.7171728645738804,
+      "learning_rate": 7.995669007618247e-06,
+      "loss": 0.3488,
+      "num_input_tokens_seen": 77378672,
+      "step": 1410,
+      "train_runtime": 7961.7074,
+      "train_tokens_per_second": 9718.854
+    },
+    {
+      "epoch": 0.5935402684563759,
+      "grad_norm": 0.7195961345962036,
+      "learning_rate": 7.982464264168682e-06,
+      "loss": 0.3835,
+      "num_input_tokens_seen": 77663136,
+      "step": 1415,
+      "train_runtime": 7994.7656,
+      "train_tokens_per_second": 9714.248
+    },
+    {
+      "epoch": 0.5956375838926175,
+      "grad_norm": 0.7110067371777979,
+      "learning_rate": 7.969227150762563e-06,
+      "loss": 0.3834,
+      "num_input_tokens_seen": 77927448,
+      "step": 1420,
+      "train_runtime": 8021.9355,
+      "train_tokens_per_second": 9714.295
+    },
+    {
+      "epoch": 0.597734899328859,
+      "grad_norm": 0.6602393728929685,
+      "learning_rate": 7.955957811067932e-06,
+      "loss": 0.3817,
+      "num_input_tokens_seen": 78224808,
+      "step": 1425,
+      "train_runtime": 8048.229,
+      "train_tokens_per_second": 9719.506
+    },
+    {
+      "epoch": 0.5998322147651006,
+      "grad_norm": 0.7182899225737743,
+      "learning_rate": 7.94265638910259e-06,
+      "loss": 0.3651,
+      "num_input_tokens_seen": 78505328,
+      "step": 1430,
+      "train_runtime": 8080.0668,
+      "train_tokens_per_second": 9715.926
+    },
+    {
+      "epoch": 0.6019295302013423,
+      "grad_norm": 0.7085963962506443,
+      "learning_rate": 7.92932302923254e-06,
+      "loss": 0.359,
+      "num_input_tokens_seen": 78786832,
+      "step": 1435,
+      "train_runtime": 8106.2786,
+      "train_tokens_per_second": 9719.236
+    },
+    {
+      "epoch": 0.6040268456375839,
+      "grad_norm": 0.6522494517592689,
+      "learning_rate": 7.915957876170427e-06,
+      "loss": 0.3692,
+      "num_input_tokens_seen": 79058464,
+      "step": 1440,
+      "train_runtime": 8133.448,
+      "train_tokens_per_second": 9720.166
+    },
+    {
+      "epoch": 0.6061241610738255,
+      "grad_norm": 0.743070916137933,
+      "learning_rate": 7.902561074973952e-06,
+      "loss": 0.3556,
+      "num_input_tokens_seen": 79309560,
+      "step": 1445,
+      "train_runtime": 8164.7392,
+      "train_tokens_per_second": 9713.667
+    },
+    {
+      "epoch": 0.6082214765100671,
+      "grad_norm": 0.7299813503209149,
+      "learning_rate": 7.889132771044315e-06,
+      "loss": 0.3666,
+      "num_input_tokens_seen": 79590016,
+      "step": 1450,
+      "train_runtime": 8197.4273,
+      "train_tokens_per_second": 9709.146
+    },
+    {
+      "epoch": 0.6103187919463087,
+      "grad_norm": 0.6481563995410097,
+      "learning_rate": 7.875673110124625e-06,
+      "loss": 0.362,
+      "num_input_tokens_seen": 79867768,
+      "step": 1455,
+      "train_runtime": 8221.9184,
+      "train_tokens_per_second": 9714.006
+    },
+    {
+      "epoch": 0.6124161073825504,
+      "grad_norm": 0.6994104104603921,
+      "learning_rate": 7.862182238298318e-06,
+      "loss": 0.3659,
+      "num_input_tokens_seen": 80150352,
+      "step": 1460,
+      "train_runtime": 8246.6794,
+      "train_tokens_per_second": 9719.106
+    },
+    {
+      "epoch": 0.614513422818792,
+      "grad_norm": 0.7979216860876888,
+      "learning_rate": 7.848660301987585e-06,
+      "loss": 0.3711,
+      "num_input_tokens_seen": 80440752,
+      "step": 1465,
+      "train_runtime": 8273.687,
+      "train_tokens_per_second": 9722.479
+    },
+    {
+      "epoch": 0.6166107382550335,
+      "grad_norm": 0.7208639094907003,
+      "learning_rate": 7.835107447951761e-06,
+      "loss": 0.3977,
+      "num_input_tokens_seen": 80728528,
+      "step": 1470,
+      "train_runtime": 8300.6297,
+      "train_tokens_per_second": 9725.591
+    },
+    {
+      "epoch": 0.6187080536912751,
+      "grad_norm": 0.6865683707927888,
+      "learning_rate": 7.821523823285758e-06,
+      "loss": 0.3696,
+      "num_input_tokens_seen": 81000232,
+      "step": 1475,
+      "train_runtime": 8330.7463,
+      "train_tokens_per_second": 9723.046
+    },
+    {
+      "epoch": 0.6208053691275168,
+      "grad_norm": 0.6595489152088321,
+      "learning_rate": 7.80790957541844e-06,
+      "loss": 0.3884,
+      "num_input_tokens_seen": 81264824,
+      "step": 1480,
+      "train_runtime": 8358.1197,
+      "train_tokens_per_second": 9722.86
+    },
+    {
+      "epoch": 0.6229026845637584,
+      "grad_norm": 0.7308438680168452,
+      "learning_rate": 7.79426485211105e-06,
+      "loss": 0.3472,
+      "num_input_tokens_seen": 81508736,
+      "step": 1485,
+      "train_runtime": 8386.6944,
+      "train_tokens_per_second": 9718.816
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.7472145159504431,
+      "learning_rate": 7.780589801455586e-06,
+      "loss": 0.4002,
+      "num_input_tokens_seen": 81781976,
+      "step": 1490,
+      "train_runtime": 8412.2231,
+      "train_tokens_per_second": 9721.803
+    },
+    {
+      "epoch": 0.6270973154362416,
+      "grad_norm": 0.8196528607221002,
+      "learning_rate": 7.766884571873208e-06,
+      "loss": 0.352,
+      "num_input_tokens_seen": 82033120,
+      "step": 1495,
+      "train_runtime": 8437.1862,
+      "train_tokens_per_second": 9722.805
+    },
+    {
+      "epoch": 0.6291946308724832,
+      "grad_norm": 0.7908391441474241,
+      "learning_rate": 7.753149312112613e-06,
+      "loss": 0.3815,
+      "num_input_tokens_seen": 82314312,
+      "step": 1500,
+      "train_runtime": 8461.5189,
+      "train_tokens_per_second": 9728.078
+    },
+    {
+      "epoch": 0.6312919463087249,
+      "grad_norm": 0.6870932097166096,
+      "learning_rate": 7.739384171248436e-06,
+      "loss": 0.3823,
+      "num_input_tokens_seen": 82583560,
+      "step": 1505,
+      "train_runtime": 8487.3968,
+      "train_tokens_per_second": 9730.14
+    },
+    {
+      "epoch": 0.6333892617449665,
+      "grad_norm": 0.8070159985818102,
+      "learning_rate": 7.725589298679619e-06,
+      "loss": 0.3928,
+      "num_input_tokens_seen": 82847040,
+      "step": 1510,
+      "train_runtime": 8515.9565,
+      "train_tokens_per_second": 9728.448
+    },
+    {
+      "epoch": 0.635486577181208,
+      "grad_norm": 0.689834606027195,
+      "learning_rate": 7.711764844127796e-06,
+      "loss": 0.3496,
+      "num_input_tokens_seen": 83146816,
+      "step": 1515,
+      "train_runtime": 8546.7969,
+      "train_tokens_per_second": 9728.418
+    },
+    {
+      "epoch": 0.6375838926174496,
+      "grad_norm": 0.7215096623557021,
+      "learning_rate": 7.697910957635672e-06,
+      "loss": 0.3982,
+      "num_input_tokens_seen": 83410984,
+      "step": 1520,
+      "train_runtime": 8575.5052,
+      "train_tokens_per_second": 9726.655
+    },
+    {
+      "epoch": 0.6396812080536913,
+      "grad_norm": 0.7859989810988971,
+      "learning_rate": 7.684027789565384e-06,
+      "loss": 0.3748,
+      "num_input_tokens_seen": 83692280,
+      "step": 1525,
+      "train_runtime": 8601.9511,
+      "train_tokens_per_second": 9729.453
+    },
+    {
+      "epoch": 0.6417785234899329,
+      "grad_norm": 0.7164631663582786,
+      "learning_rate": 7.670115490596872e-06,
+      "loss": 0.3838,
+      "num_input_tokens_seen": 83955256,
+      "step": 1530,
+      "train_runtime": 8627.5517,
+      "train_tokens_per_second": 9731.064
+    },
+    {
+      "epoch": 0.6438758389261745,
+      "grad_norm": 0.7181349024505007,
+      "learning_rate": 7.656174211726253e-06,
+      "loss": 0.3771,
+      "num_input_tokens_seen": 84225232,
+      "step": 1535,
+      "train_runtime": 8658.4718,
+      "train_tokens_per_second": 9727.494
+    },
+    {
+      "epoch": 0.6459731543624161,
+      "grad_norm": 0.794644715611002,
+      "learning_rate": 7.642204104264171e-06,
+      "loss": 0.3852,
+      "num_input_tokens_seen": 84477872,
+      "step": 1540,
+      "train_runtime": 8681.3662,
+      "train_tokens_per_second": 9730.942
+    },
+    {
+      "epoch": 0.6480704697986577,
+      "grad_norm": 0.6529237819630043,
+      "learning_rate": 7.6282053198341585e-06,
+      "loss": 0.3553,
+      "num_input_tokens_seen": 84746400,
+      "step": 1545,
+      "train_runtime": 8704.7845,
+      "train_tokens_per_second": 9735.611
+    },
+    {
+      "epoch": 0.6501677852348994,
+      "grad_norm": 0.7908171178842476,
+      "learning_rate": 7.614178010370996e-06,
+      "loss": 0.3772,
+      "num_input_tokens_seen": 84997120,
+      "step": 1550,
+      "train_runtime": 8730.733,
+      "train_tokens_per_second": 9735.393
+    },
+    {
+      "epoch": 0.652265100671141,
+      "grad_norm": 0.6902471246215945,
+      "learning_rate": 7.600122328119053e-06,
+      "loss": 0.3644,
+      "num_input_tokens_seen": 85267232,
+      "step": 1555,
+      "train_runtime": 8756.0409,
+      "train_tokens_per_second": 9738.103
+    },
+    {
+      "epoch": 0.6543624161073825,
+      "grad_norm": 0.6587103714983763,
+      "learning_rate": 7.586038425630642e-06,
+      "loss": 0.3598,
+      "num_input_tokens_seen": 85542392,
+      "step": 1560,
+      "train_runtime": 8786.9228,
+      "train_tokens_per_second": 9735.193
+    },
+    {
+      "epoch": 0.6564597315436241,
+      "grad_norm": 0.7608622963093746,
+      "learning_rate": 7.5719264557643645e-06,
+      "loss": 0.3685,
+      "num_input_tokens_seen": 85811368,
+      "step": 1565,
+      "train_runtime": 8814.396,
+      "train_tokens_per_second": 9735.366
+    },
+    {
+      "epoch": 0.6585570469798657,
+      "grad_norm": 0.6436169560166048,
+      "learning_rate": 7.55778657168345e-06,
+      "loss": 0.3763,
+      "num_input_tokens_seen": 86112880,
+      "step": 1570,
+      "train_runtime": 8849.2641,
+      "train_tokens_per_second": 9731.078
+    },
+    {
+      "epoch": 0.6606543624161074,
+      "grad_norm": 0.7523192850220879,
+      "learning_rate": 7.543618926854086e-06,
+      "loss": 0.385,
+      "num_input_tokens_seen": 86391840,
+      "step": 1575,
+      "train_runtime": 8883.1538,
+      "train_tokens_per_second": 9725.357
+    },
+    {
+      "epoch": 0.662751677852349,
+      "grad_norm": 0.6689997677451208,
+      "learning_rate": 7.529423675043765e-06,
+      "loss": 0.3729,
+      "num_input_tokens_seen": 86649376,
+      "step": 1580,
+      "train_runtime": 8913.694,
+      "train_tokens_per_second": 9720.928
+    },
+    {
+      "epoch": 0.6648489932885906,
+      "grad_norm": 0.5624453214430509,
+      "learning_rate": 7.5152009703196105e-06,
+      "loss": 0.3649,
+      "num_input_tokens_seen": 86945312,
+      "step": 1585,
+      "train_runtime": 8937.1128,
+      "train_tokens_per_second": 9728.568
+    },
+    {
+      "epoch": 0.6669463087248322,
+      "grad_norm": 0.721144076554704,
+      "learning_rate": 7.500950967046701e-06,
+      "loss": 0.3587,
+      "num_input_tokens_seen": 87212032,
+      "step": 1590,
+      "train_runtime": 8964.4332,
+      "train_tokens_per_second": 9728.672
+    },
+    {
+      "epoch": 0.6690436241610739,
+      "grad_norm": 0.6741568886738529,
+      "learning_rate": 7.486673819886398e-06,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 87481144,
+      "step": 1595,
+      "train_runtime": 8992.065,
+      "train_tokens_per_second": 9728.705
+    },
+    {
+      "epoch": 0.6711409395973155,
+      "grad_norm": 0.7383168186382499,
+      "learning_rate": 7.47236968379467e-06,
+      "loss": 0.3784,
+      "num_input_tokens_seen": 87730032,
+      "step": 1600,
+      "train_runtime": 9024.3571,
+      "train_tokens_per_second": 9721.472
+    },
+    {
+      "epoch": 0.673238255033557,
+      "grad_norm": 0.763895571613687,
+      "learning_rate": 7.458038714020406e-06,
+      "loss": 0.3651,
+      "num_input_tokens_seen": 88042536,
+      "step": 1605,
+      "train_runtime": 9049.1696,
+      "train_tokens_per_second": 9729.35
+    },
+    {
+      "epoch": 0.6753355704697986,
+      "grad_norm": 0.7349174573179577,
+      "learning_rate": 7.4436810661037295e-06,
+      "loss": 0.3545,
+      "num_input_tokens_seen": 88308024,
+      "step": 1610,
+      "train_runtime": 9075.0817,
+      "train_tokens_per_second": 9730.824
+    },
+    {
+      "epoch": 0.6774328859060402,
+      "grad_norm": 0.6457803648061802,
+      "learning_rate": 7.429296895874318e-06,
+      "loss": 0.3775,
+      "num_input_tokens_seen": 88576904,
+      "step": 1615,
+      "train_runtime": 9105.0993,
+      "train_tokens_per_second": 9728.274
+    },
+    {
+      "epoch": 0.6795302013422819,
+      "grad_norm": 0.6477251523289795,
+      "learning_rate": 7.4148863594497085e-06,
+      "loss": 0.3516,
+      "num_input_tokens_seen": 88850344,
+      "step": 1620,
+      "train_runtime": 9140.1037,
+      "train_tokens_per_second": 9720.934
+    },
+    {
+      "epoch": 0.6816275167785235,
+      "grad_norm": 0.673118402002902,
+      "learning_rate": 7.400449613233595e-06,
+      "loss": 0.3877,
+      "num_input_tokens_seen": 89123528,
+      "step": 1625,
+      "train_runtime": 9172.7591,
+      "train_tokens_per_second": 9716.109
+    },
+    {
+      "epoch": 0.6837248322147651,
+      "grad_norm": 0.7319946699000737,
+      "learning_rate": 7.385986813914141e-06,
+      "loss": 0.3537,
+      "num_input_tokens_seen": 89363512,
+      "step": 1630,
+      "train_runtime": 9196.8336,
+      "train_tokens_per_second": 9716.77
+    },
+    {
+      "epoch": 0.6858221476510067,
+      "grad_norm": 0.7321662510611724,
+      "learning_rate": 7.371498118462276e-06,
+      "loss": 0.3582,
+      "num_input_tokens_seen": 89645744,
+      "step": 1635,
+      "train_runtime": 9227.223,
+      "train_tokens_per_second": 9715.355
+    },
+    {
+      "epoch": 0.6879194630872483,
+      "grad_norm": 0.7769919931074649,
+      "learning_rate": 7.3569836841299905e-06,
+      "loss": 0.3959,
+      "num_input_tokens_seen": 89923776,
+      "step": 1640,
+      "train_runtime": 9257.8357,
+      "train_tokens_per_second": 9713.261
+    },
+    {
+      "epoch": 0.69001677852349,
+      "grad_norm": 0.7178109328373244,
+      "learning_rate": 7.342443668448627e-06,
+      "loss": 0.3787,
+      "num_input_tokens_seen": 90192584,
+      "step": 1645,
+      "train_runtime": 9285.7826,
+      "train_tokens_per_second": 9712.976
+    },
+    {
+      "epoch": 0.6921140939597316,
+      "grad_norm": 0.6852600583839311,
+      "learning_rate": 7.327878229227182e-06,
+      "loss": 0.3419,
+      "num_input_tokens_seen": 90473912,
+      "step": 1650,
+      "train_runtime": 9310.6897,
+      "train_tokens_per_second": 9717.208
+    },
+    {
+      "epoch": 0.6942114093959731,
+      "grad_norm": 0.6924248385354302,
+      "learning_rate": 7.313287524550574e-06,
+      "loss": 0.3825,
+      "num_input_tokens_seen": 90760912,
+      "step": 1655,
+      "train_runtime": 9345.4626,
+      "train_tokens_per_second": 9711.762
+    },
+    {
+      "epoch": 0.6963087248322147,
+      "grad_norm": 0.6796185641243762,
+      "learning_rate": 7.298671712777947e-06,
+      "loss": 0.3851,
+      "num_input_tokens_seen": 91023552,
+      "step": 1660,
+      "train_runtime": 9367.5063,
+      "train_tokens_per_second": 9716.946
+    },
+    {
+      "epoch": 0.6984060402684564,
+      "grad_norm": 0.7678229468133488,
+      "learning_rate": 7.284030952540937e-06,
+      "loss": 0.3995,
+      "num_input_tokens_seen": 91319648,
+      "step": 1665,
+      "train_runtime": 9395.7494,
+      "train_tokens_per_second": 9719.251
+    },
+    {
+      "epoch": 0.700503355704698,
+      "grad_norm": 0.6944094996949246,
+      "learning_rate": 7.269365402741959e-06,
+      "loss": 0.3527,
+      "num_input_tokens_seen": 91603112,
+      "step": 1670,
+      "train_runtime": 9429.3174,
+      "train_tokens_per_second": 9714.713
+    },
+    {
+      "epoch": 0.7026006711409396,
+      "grad_norm": 0.767547366713501,
+      "learning_rate": 7.2546752225524775e-06,
+      "loss": 0.3367,
+      "num_input_tokens_seen": 91849176,
+      "step": 1675,
+      "train_runtime": 9457.6167,
+      "train_tokens_per_second": 9711.662
+    },
+    {
+      "epoch": 0.7046979865771812,
+      "grad_norm": 0.7039681780175793,
+      "learning_rate": 7.2399605714112865e-06,
+      "loss": 0.3832,
+      "num_input_tokens_seen": 92105848,
+      "step": 1680,
+      "train_runtime": 9479.8315,
+      "train_tokens_per_second": 9715.979
+    },
+    {
+      "epoch": 0.7067953020134228,
+      "grad_norm": 0.8455706420054568,
+      "learning_rate": 7.225221609022766e-06,
+      "loss": 0.3918,
+      "num_input_tokens_seen": 92379504,
+      "step": 1685,
+      "train_runtime": 9512.7091,
+      "train_tokens_per_second": 9711.167
+    },
+    {
+      "epoch": 0.7088926174496645,
+      "grad_norm": 0.6404824593507985,
+      "learning_rate": 7.210458495355165e-06,
+      "loss": 0.4009,
+      "num_input_tokens_seen": 92687048,
+      "step": 1690,
+      "train_runtime": 9546.8991,
+      "train_tokens_per_second": 9708.602
+    },
+    {
+      "epoch": 0.7109899328859061,
+      "grad_norm": 0.6863294035673944,
+      "learning_rate": 7.195671390638847e-06,
+      "loss": 0.3944,
+      "num_input_tokens_seen": 92980456,
+      "step": 1695,
+      "train_runtime": 9569.5701,
+      "train_tokens_per_second": 9716.263
+    },
+    {
+      "epoch": 0.7130872483221476,
+      "grad_norm": 0.6748098546311956,
+      "learning_rate": 7.180860455364567e-06,
+      "loss": 0.3776,
+      "num_input_tokens_seen": 93247472,
+      "step": 1700,
+      "train_runtime": 9595.1917,
+      "train_tokens_per_second": 9718.146
+    },
+    {
+      "epoch": 0.7151845637583892,
+      "grad_norm": 0.6554662888858104,
+      "learning_rate": 7.166025850281724e-06,
+      "loss": 0.3656,
+      "num_input_tokens_seen": 93518192,
+      "step": 1705,
+      "train_runtime": 9622.2498,
+      "train_tokens_per_second": 9718.953
+    },
+    {
+      "epoch": 0.7172818791946308,
+      "grad_norm": 0.7758466574705513,
+      "learning_rate": 7.151167736396611e-06,
+      "loss": 0.3741,
+      "num_input_tokens_seen": 93797608,
+      "step": 1710,
+      "train_runtime": 9658.0399,
+      "train_tokens_per_second": 9711.868
+    },
+    {
+      "epoch": 0.7193791946308725,
+      "grad_norm": 0.6748492530435837,
+      "learning_rate": 7.136286274970677e-06,
+      "loss": 0.3826,
+      "num_input_tokens_seen": 94096920,
+      "step": 1715,
+      "train_runtime": 9695.9182,
+      "train_tokens_per_second": 9704.797
+    },
+    {
+      "epoch": 0.7214765100671141,
+      "grad_norm": 0.6964261755024105,
+      "learning_rate": 7.121381627518764e-06,
+      "loss": 0.3833,
+      "num_input_tokens_seen": 94398320,
+      "step": 1720,
+      "train_runtime": 9725.7629,
+      "train_tokens_per_second": 9706.007
+    },
+    {
+      "epoch": 0.7235738255033557,
+      "grad_norm": 0.7646393327335587,
+      "learning_rate": 7.106453955807372e-06,
+      "loss": 0.3645,
+      "num_input_tokens_seen": 94667672,
+      "step": 1725,
+      "train_runtime": 9758.0965,
+      "train_tokens_per_second": 9701.449
+    },
+    {
+      "epoch": 0.7256711409395973,
+      "grad_norm": 0.6750946491250904,
+      "learning_rate": 7.091503421852887e-06,
+      "loss": 0.4054,
+      "num_input_tokens_seen": 94954944,
+      "step": 1730,
+      "train_runtime": 9785.0617,
+      "train_tokens_per_second": 9704.072
+    },
+    {
+      "epoch": 0.727768456375839,
+      "grad_norm": 0.6740981028138281,
+      "learning_rate": 7.076530187919828e-06,
+      "loss": 0.3534,
+      "num_input_tokens_seen": 95244664,
+      "step": 1735,
+      "train_runtime": 9807.3601,
+      "train_tokens_per_second": 9711.55
+    },
+    {
+      "epoch": 0.7298657718120806,
+      "grad_norm": 0.932891462452153,
+      "learning_rate": 7.06153441651909e-06,
+      "loss": 0.3622,
+      "num_input_tokens_seen": 95519032,
+      "step": 1740,
+      "train_runtime": 9836.2069,
+      "train_tokens_per_second": 9710.962
+    },
+    {
+      "epoch": 0.7319630872483222,
+      "grad_norm": 0.7880527100970997,
+      "learning_rate": 7.046516270406174e-06,
+      "loss": 0.359,
+      "num_input_tokens_seen": 95765576,
+      "step": 1745,
+      "train_runtime": 9865.3325,
+      "train_tokens_per_second": 9707.283
+    },
+    {
+      "epoch": 0.7340604026845637,
+      "grad_norm": 0.6469597817307424,
+      "learning_rate": 7.031475912579424e-06,
+      "loss": 0.3682,
+      "num_input_tokens_seen": 96055472,
+      "step": 1750,
+      "train_runtime": 9887.9216,
+      "train_tokens_per_second": 9714.425
+    },
+    {
+      "epoch": 0.7361577181208053,
+      "grad_norm": 0.8690449749716439,
+      "learning_rate": 7.01641350627826e-06,
+      "loss": 0.3732,
+      "num_input_tokens_seen": 96316056,
+      "step": 1755,
+      "train_runtime": 9915.9181,
+      "train_tokens_per_second": 9713.277
+    },
+    {
+      "epoch": 0.738255033557047,
+      "grad_norm": 0.7234758994802252,
+      "learning_rate": 7.001329214981398e-06,
+      "loss": 0.3781,
+      "num_input_tokens_seen": 96590168,
+      "step": 1760,
+      "train_runtime": 9940.0676,
+      "train_tokens_per_second": 9717.255
+    },
+    {
+      "epoch": 0.7403523489932886,
+      "grad_norm": 0.713443491880628,
+      "learning_rate": 6.986223202405088e-06,
+      "loss": 0.3888,
+      "num_input_tokens_seen": 96885608,
+      "step": 1765,
+      "train_runtime": 9964.0763,
+      "train_tokens_per_second": 9723.491
+    },
+    {
+      "epoch": 0.7424496644295302,
+      "grad_norm": 0.7887091601090002,
+      "learning_rate": 6.971095632501321e-06,
+      "loss": 0.3782,
+      "num_input_tokens_seen": 97140216,
+      "step": 1770,
+      "train_runtime": 9991.4431,
+      "train_tokens_per_second": 9722.341
+    },
+    {
+      "epoch": 0.7445469798657718,
+      "grad_norm": 0.6939358911083898,
+      "learning_rate": 6.955946669456071e-06,
+      "loss": 0.399,
+      "num_input_tokens_seen": 97408064,
+      "step": 1775,
+      "train_runtime": 10018.8812,
+      "train_tokens_per_second": 9722.449
+    },
+    {
+      "epoch": 0.7466442953020134,
+      "grad_norm": 0.6958636104926587,
+      "learning_rate": 6.940776477687487e-06,
+      "loss": 0.3511,
+      "num_input_tokens_seen": 97670904,
+      "step": 1780,
+      "train_runtime": 10046.1204,
+      "train_tokens_per_second": 9722.251
+    },
+    {
+      "epoch": 0.7487416107382551,
+      "grad_norm": 0.7965355664555905,
+      "learning_rate": 6.925585221844137e-06,
+      "loss": 0.3814,
+      "num_input_tokens_seen": 97932384,
+      "step": 1785,
+      "train_runtime": 10070.6402,
+      "train_tokens_per_second": 9724.544
+    },
+    {
+      "epoch": 0.7508389261744967,
+      "grad_norm": 0.6409039420275505,
+      "learning_rate": 6.910373066803194e-06,
+      "loss": 0.3783,
+      "num_input_tokens_seen": 98189384,
+      "step": 1790,
+      "train_runtime": 10101.5837,
+      "train_tokens_per_second": 9720.197
+    },
+    {
+      "epoch": 0.7529362416107382,
+      "grad_norm": 0.6946925710032521,
+      "learning_rate": 6.895140177668666e-06,
+      "loss": 0.372,
+      "num_input_tokens_seen": 98472024,
+      "step": 1795,
+      "train_runtime": 10128.4113,
+      "train_tokens_per_second": 9722.356
+    },
+    {
+      "epoch": 0.7550335570469798,
+      "grad_norm": 0.6690257528007673,
+      "learning_rate": 6.879886719769595e-06,
+      "loss": 0.3894,
+      "num_input_tokens_seen": 98769360,
+      "step": 1800,
+      "train_runtime": 10157.6609,
+      "train_tokens_per_second": 9723.632
+    },
+    {
+      "epoch": 0.7571308724832215,
+      "grad_norm": 0.6737685363234319,
+      "learning_rate": 6.864612858658264e-06,
+      "loss": 0.3429,
+      "num_input_tokens_seen": 99048720,
+      "step": 1805,
+      "train_runtime": 10184.5946,
+      "train_tokens_per_second": 9725.347
+    },
+    {
+      "epoch": 0.7592281879194631,
+      "grad_norm": 0.7482844960863255,
+      "learning_rate": 6.849318760108401e-06,
+      "loss": 0.3648,
+      "num_input_tokens_seen": 99309584,
+      "step": 1810,
+      "train_runtime": 10213.2135,
+      "train_tokens_per_second": 9723.637
+    },
+    {
+      "epoch": 0.7613255033557047,
+      "grad_norm": 0.7302929459755721,
+      "learning_rate": 6.83400459011338e-06,
+      "loss": 0.3624,
+      "num_input_tokens_seen": 99584352,
+      "step": 1815,
+      "train_runtime": 10239.8118,
+      "train_tokens_per_second": 9725.213
+    },
+    {
+      "epoch": 0.7634228187919463,
+      "grad_norm": 0.6989565170185783,
+      "learning_rate": 6.81867051488442e-06,
+      "loss": 0.3753,
+      "num_input_tokens_seen": 99835712,
+      "step": 1820,
+      "train_runtime": 10269.7371,
+      "train_tokens_per_second": 9721.35
+    },
+    {
+      "epoch": 0.7655201342281879,
+      "grad_norm": 0.6677525697122539,
+      "learning_rate": 6.8033167008487784e-06,
+      "loss": 0.3295,
+      "num_input_tokens_seen": 100130768,
+      "step": 1825,
+      "train_runtime": 10296.7577,
+      "train_tokens_per_second": 9724.495
+    },
+    {
+      "epoch": 0.7676174496644296,
+      "grad_norm": 0.7082327951640518,
+      "learning_rate": 6.787943314647945e-06,
+      "loss": 0.3662,
+      "num_input_tokens_seen": 100405104,
+      "step": 1830,
+      "train_runtime": 10328.2572,
+      "train_tokens_per_second": 9721.398
+    },
+    {
+      "epoch": 0.7697147651006712,
+      "grad_norm": 0.7146760268832543,
+      "learning_rate": 6.7725505231358365e-06,
+      "loss": 0.3585,
+      "num_input_tokens_seen": 100675416,
+      "step": 1835,
+      "train_runtime": 10357.6251,
+      "train_tokens_per_second": 9719.932
+    },
+    {
+      "epoch": 0.7718120805369127,
+      "grad_norm": 0.6624123390794864,
+      "learning_rate": 6.757138493376985e-06,
+      "loss": 0.3504,
+      "num_input_tokens_seen": 100923008,
+      "step": 1840,
+      "train_runtime": 10385.1543,
+      "train_tokens_per_second": 9718.008
+    },
+    {
+      "epoch": 0.7739093959731543,
+      "grad_norm": 0.6686262762584677,
+      "learning_rate": 6.741707392644723e-06,
+      "loss": 0.3487,
+      "num_input_tokens_seen": 101196688,
+      "step": 1845,
+      "train_runtime": 10417.5304,
+      "train_tokens_per_second": 9714.077
+    },
+    {
+      "epoch": 0.7760067114093959,
+      "grad_norm": 0.7897900174110379,
+      "learning_rate": 6.726257388419365e-06,
+      "loss": 0.3744,
+      "num_input_tokens_seen": 101477392,
+      "step": 1850,
+      "train_runtime": 10445.19,
+      "train_tokens_per_second": 9715.227
+    },
+    {
+      "epoch": 0.7781040268456376,
+      "grad_norm": 0.6950246520946964,
+      "learning_rate": 6.7107886483863975e-06,
+      "loss": 0.3574,
+      "num_input_tokens_seen": 101764488,
+      "step": 1855,
+      "train_runtime": 10468.8937,
+      "train_tokens_per_second": 9720.654
+    },
+    {
+      "epoch": 0.7802013422818792,
+      "grad_norm": 0.7619950254673009,
+      "learning_rate": 6.69530134043465e-06,
+      "loss": 0.3891,
+      "num_input_tokens_seen": 102027040,
+      "step": 1860,
+      "train_runtime": 10504.4518,
+      "train_tokens_per_second": 9712.743
+    },
+    {
+      "epoch": 0.7822986577181208,
+      "grad_norm": 0.7050565220976114,
+      "learning_rate": 6.679795632654482e-06,
+      "loss": 0.331,
+      "num_input_tokens_seen": 102285608,
+      "step": 1865,
+      "train_runtime": 10530.3464,
+      "train_tokens_per_second": 9713.413
+    },
+    {
+      "epoch": 0.7843959731543624,
+      "grad_norm": 0.679229301048062,
+      "learning_rate": 6.664271693335954e-06,
+      "loss": 0.3382,
+      "num_input_tokens_seen": 102572960,
+      "step": 1870,
+      "train_runtime": 10561.64,
+      "train_tokens_per_second": 9711.84
+    },
+    {
+      "epoch": 0.7864932885906041,
+      "grad_norm": 0.7660508917976222,
+      "learning_rate": 6.648729690966999e-06,
+      "loss": 0.3334,
+      "num_input_tokens_seen": 102851128,
+      "step": 1875,
+      "train_runtime": 10590.859,
+      "train_tokens_per_second": 9711.311
+    },
+    {
+      "epoch": 0.7885906040268457,
+      "grad_norm": 0.6246973856894367,
+      "learning_rate": 6.6331697942315955e-06,
+      "loss": 0.349,
+      "num_input_tokens_seen": 103139024,
+      "step": 1880,
+      "train_runtime": 10618.2121,
+      "train_tokens_per_second": 9713.408
+    },
+    {
+      "epoch": 0.7906879194630873,
+      "grad_norm": 0.6911856929074752,
+      "learning_rate": 6.617592172007938e-06,
+      "loss": 0.4236,
+      "num_input_tokens_seen": 103439352,
+      "step": 1885,
+      "train_runtime": 10644.1378,
+      "train_tokens_per_second": 9717.964
+    },
+    {
+      "epoch": 0.7927852348993288,
+      "grad_norm": 0.8819901591922512,
+      "learning_rate": 6.601996993366606e-06,
+      "loss": 0.3704,
+      "num_input_tokens_seen": 103715904,
+      "step": 1890,
+      "train_runtime": 10667.1617,
+      "train_tokens_per_second": 9722.915
+    },
+    {
+      "epoch": 0.7948825503355704,
+      "grad_norm": 0.6208782802831756,
+      "learning_rate": 6.586384427568723e-06,
+      "loss": 0.3695,
+      "num_input_tokens_seen": 103990824,
+      "step": 1895,
+      "train_runtime": 10698.8738,
+      "train_tokens_per_second": 9719.792
+    },
+    {
+      "epoch": 0.7969798657718121,
+      "grad_norm": 0.6717545452671818,
+      "learning_rate": 6.570754644064124e-06,
+      "loss": 0.3766,
+      "num_input_tokens_seen": 104267464,
+      "step": 1900,
+      "train_runtime": 10723.9214,
+      "train_tokens_per_second": 9722.886
+    },
+    {
+      "epoch": 0.7990771812080537,
+      "grad_norm": 0.8321841052071693,
+      "learning_rate": 6.555107812489513e-06,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 104519568,
+      "step": 1905,
+      "train_runtime": 10752.2194,
+      "train_tokens_per_second": 9720.744
+    },
+    {
+      "epoch": 0.8011744966442953,
+      "grad_norm": 0.820295708355053,
+      "learning_rate": 6.539444102666626e-06,
+      "loss": 0.3805,
+      "num_input_tokens_seen": 104780952,
+      "step": 1910,
+      "train_runtime": 10783.3573,
+      "train_tokens_per_second": 9716.914
+    },
+    {
+      "epoch": 0.8032718120805369,
+      "grad_norm": 0.8015745251510984,
+      "learning_rate": 6.523763684600387e-06,
+      "loss": 0.3241,
+      "num_input_tokens_seen": 105042656,
+      "step": 1915,
+      "train_runtime": 10806.365,
+      "train_tokens_per_second": 9720.443
+    },
+    {
+      "epoch": 0.8053691275167785,
+      "grad_norm": 0.7245992014677467,
+      "learning_rate": 6.508066728477062e-06,
+      "loss": 0.3625,
+      "num_input_tokens_seen": 105295752,
+      "step": 1920,
+      "train_runtime": 10830.4734,
+      "train_tokens_per_second": 9722.174
+    },
+    {
+      "epoch": 0.8074664429530202,
+      "grad_norm": 0.7007879994424163,
+      "learning_rate": 6.492353404662406e-06,
+      "loss": 0.3623,
+      "num_input_tokens_seen": 105575056,
+      "step": 1925,
+      "train_runtime": 10858.416,
+      "train_tokens_per_second": 9722.878
+    },
+    {
+      "epoch": 0.8095637583892618,
+      "grad_norm": 0.6939127294445337,
+      "learning_rate": 6.4766238836998265e-06,
+      "loss": 0.3433,
+      "num_input_tokens_seen": 105874016,
+      "step": 1930,
+      "train_runtime": 10882.7733,
+      "train_tokens_per_second": 9728.588
+    },
+    {
+      "epoch": 0.8116610738255033,
+      "grad_norm": 0.6928527557606368,
+      "learning_rate": 6.4608783363085195e-06,
+      "loss": 0.3673,
+      "num_input_tokens_seen": 106171584,
+      "step": 1935,
+      "train_runtime": 10908.0676,
+      "train_tokens_per_second": 9733.308
+    },
+    {
+      "epoch": 0.8137583892617449,
+      "grad_norm": 0.8873593639411321,
+      "learning_rate": 6.445116933381626e-06,
+      "loss": 0.3458,
+      "num_input_tokens_seen": 106442272,
+      "step": 1940,
+      "train_runtime": 10930.0687,
+      "train_tokens_per_second": 9738.482
+    },
+    {
+      "epoch": 0.8158557046979866,
+      "grad_norm": 0.7159989495212568,
+      "learning_rate": 6.429339845984377e-06,
+      "loss": 0.3701,
+      "num_input_tokens_seen": 106708408,
+      "step": 1945,
+      "train_runtime": 10957.1325,
+      "train_tokens_per_second": 9738.717
+    },
+    {
+      "epoch": 0.8179530201342282,
+      "grad_norm": 0.6367610781501775,
+      "learning_rate": 6.413547245352228e-06,
+      "loss": 0.3546,
+      "num_input_tokens_seen": 106995160,
+      "step": 1950,
+      "train_runtime": 10986.6373,
+      "train_tokens_per_second": 9738.663
+    },
+    {
+      "epoch": 0.8200503355704698,
+      "grad_norm": 0.642480091043729,
+      "learning_rate": 6.397739302889008e-06,
+      "loss": 0.3552,
+      "num_input_tokens_seen": 107291640,
+      "step": 1955,
+      "train_runtime": 11015.9281,
+      "train_tokens_per_second": 9739.682
+    },
+    {
+      "epoch": 0.8221476510067114,
+      "grad_norm": 0.6272143479541751,
+      "learning_rate": 6.381916190165059e-06,
+      "loss": 0.3614,
+      "num_input_tokens_seen": 107583848,
+      "step": 1960,
+      "train_runtime": 11043.1603,
+      "train_tokens_per_second": 9742.125
+    },
+    {
+      "epoch": 0.824244966442953,
+      "grad_norm": 0.81206849031903,
+      "learning_rate": 6.366078078915372e-06,
+      "loss": 0.3659,
+      "num_input_tokens_seen": 107841856,
+      "step": 1965,
+      "train_runtime": 11066.7245,
+      "train_tokens_per_second": 9744.695
+    },
+    {
+      "epoch": 0.8263422818791947,
+      "grad_norm": 0.7932668914935843,
+      "learning_rate": 6.3502251410377195e-06,
+      "loss": 0.3681,
+      "num_input_tokens_seen": 108091336,
+      "step": 1970,
+      "train_runtime": 11096.1179,
+      "train_tokens_per_second": 9741.365
+    },
+    {
+      "epoch": 0.8284395973154363,
+      "grad_norm": 0.7216621566378123,
+      "learning_rate": 6.334357548590799e-06,
+      "loss": 0.3607,
+      "num_input_tokens_seen": 108356096,
+      "step": 1975,
+      "train_runtime": 11125.3707,
+      "train_tokens_per_second": 9739.549
+    },
+    {
+      "epoch": 0.8305369127516778,
+      "grad_norm": 0.6876142986274044,
+      "learning_rate": 6.318475473792358e-06,
+      "loss": 0.3495,
+      "num_input_tokens_seen": 108626832,
+      "step": 1980,
+      "train_runtime": 11151.8455,
+      "train_tokens_per_second": 9740.705
+    },
+    {
+      "epoch": 0.8326342281879194,
+      "grad_norm": 0.6771829324863735,
+      "learning_rate": 6.302579089017328e-06,
+      "loss": 0.396,
+      "num_input_tokens_seen": 108933800,
+      "step": 1985,
+      "train_runtime": 11173.9259,
+      "train_tokens_per_second": 9748.928
+    },
+    {
+      "epoch": 0.834731543624161,
+      "grad_norm": 0.692238192474028,
+      "learning_rate": 6.286668566795952e-06,
+      "loss": 0.345,
+      "num_input_tokens_seen": 109213784,
+      "step": 1990,
+      "train_runtime": 11199.7461,
+      "train_tokens_per_second": 9751.452
+    },
+    {
+      "epoch": 0.8368288590604027,
+      "grad_norm": 0.7144811981042052,
+      "learning_rate": 6.27074407981191e-06,
+      "loss": 0.3468,
+      "num_input_tokens_seen": 109450928,
+      "step": 1995,
+      "train_runtime": 11223.662,
+      "train_tokens_per_second": 9751.802
+    },
+    {
+      "epoch": 0.8389261744966443,
+      "grad_norm": 0.6463160600596939,
+      "learning_rate": 6.254805800900454e-06,
+      "loss": 0.3741,
+      "num_input_tokens_seen": 109743848,
+      "step": 2000,
+      "train_runtime": 11244.7283,
+      "train_tokens_per_second": 9759.582
+    },
+    {
+      "epoch": 0.8410234899328859,
+      "grad_norm": 0.678436279790263,
+      "learning_rate": 6.23885390304652e-06,
+      "loss": 0.3831,
+      "num_input_tokens_seen": 110025936,
+      "step": 2005,
+      "train_runtime": 11272.136,
+      "train_tokens_per_second": 9760.877
+    },
+    {
+      "epoch": 0.8431208053691275,
+      "grad_norm": 0.6556593851267769,
+      "learning_rate": 6.22288855938286e-06,
+      "loss": 0.3691,
+      "num_input_tokens_seen": 110297504,
+      "step": 2010,
+      "train_runtime": 11298.7387,
+      "train_tokens_per_second": 9761.931
+    },
+    {
+      "epoch": 0.8452181208053692,
+      "grad_norm": 0.7424955585876786,
+      "learning_rate": 6.206909943188154e-06,
+      "loss": 0.3609,
+      "num_input_tokens_seen": 110564488,
+      "step": 2015,
+      "train_runtime": 11324.6206,
+      "train_tokens_per_second": 9763.196
+    },
+    {
+      "epoch": 0.8473154362416108,
+      "grad_norm": 0.7972893547035088,
+      "learning_rate": 6.1909182278851406e-06,
+      "loss": 0.3829,
+      "num_input_tokens_seen": 110854656,
+      "step": 2020,
+      "train_runtime": 11354.6445,
+      "train_tokens_per_second": 9762.935
+    },
+    {
+      "epoch": 0.8494127516778524,
+      "grad_norm": 0.7337561070261962,
+      "learning_rate": 6.174913587038725e-06,
+      "loss": 0.3551,
+      "num_input_tokens_seen": 111132032,
+      "step": 2025,
+      "train_runtime": 11386.6252,
+      "train_tokens_per_second": 9759.874
+    },
+    {
+      "epoch": 0.8515100671140939,
+      "grad_norm": 0.7178990055985761,
+      "learning_rate": 6.1588961943541e-06,
+      "loss": 0.3848,
+      "num_input_tokens_seen": 111406864,
+      "step": 2030,
+      "train_runtime": 11416.4668,
+      "train_tokens_per_second": 9758.436
+    },
+    {
+      "epoch": 0.8536073825503355,
+      "grad_norm": 0.6209802055246124,
+      "learning_rate": 6.142866223674859e-06,
+      "loss": 0.3583,
+      "num_input_tokens_seen": 111692968,
+      "step": 2035,
+      "train_runtime": 11444.0604,
+      "train_tokens_per_second": 9759.907
+    },
+    {
+      "epoch": 0.8557046979865772,
+      "grad_norm": 0.6755535402517975,
+      "learning_rate": 6.126823848981109e-06,
+      "loss": 0.3625,
+      "num_input_tokens_seen": 111961968,
+      "step": 2040,
+      "train_runtime": 11469.0058,
+      "train_tokens_per_second": 9762.134
+    },
+    {
+      "epoch": 0.8578020134228188,
+      "grad_norm": 0.7619799713462062,
+      "learning_rate": 6.110769244387584e-06,
+      "loss": 0.3594,
+      "num_input_tokens_seen": 112232208,
+      "step": 2045,
+      "train_runtime": 11501.55,
+      "train_tokens_per_second": 9758.007
+    },
+    {
+      "epoch": 0.8598993288590604,
+      "grad_norm": 0.757477195260574,
+      "learning_rate": 6.094702584141755e-06,
+      "loss": 0.3755,
+      "num_input_tokens_seen": 112500120,
+      "step": 2050,
+      "train_runtime": 11527.9631,
+      "train_tokens_per_second": 9758.89
+    },
+    {
+      "epoch": 0.861996644295302,
+      "grad_norm": 0.7596572013313776,
+      "learning_rate": 6.078624042621938e-06,
+      "loss": 0.3441,
+      "num_input_tokens_seen": 112774944,
+      "step": 2055,
+      "train_runtime": 11560.9707,
+      "train_tokens_per_second": 9754.799
+    },
+    {
+      "epoch": 0.8640939597315436,
+      "grad_norm": 0.6571576240197052,
+      "learning_rate": 6.062533794335399e-06,
+      "loss": 0.3632,
+      "num_input_tokens_seen": 113058040,
+      "step": 2060,
+      "train_runtime": 11588.9087,
+      "train_tokens_per_second": 9755.711
+    },
+    {
+      "epoch": 0.8661912751677853,
+      "grad_norm": 0.6903295610799248,
+      "learning_rate": 6.046432013916467e-06,
+      "loss": 0.3568,
+      "num_input_tokens_seen": 113325752,
+      "step": 2065,
+      "train_runtime": 11621.7548,
+      "train_tokens_per_second": 9751.174
+    },
+    {
+      "epoch": 0.8682885906040269,
+      "grad_norm": 0.6757375524286993,
+      "learning_rate": 6.030318876124629e-06,
+      "loss": 0.3827,
+      "num_input_tokens_seen": 113602096,
+      "step": 2070,
+      "train_runtime": 11654.9548,
+      "train_tokens_per_second": 9747.107
+    },
+    {
+      "epoch": 0.8703859060402684,
+      "grad_norm": 0.6389124067341382,
+      "learning_rate": 6.014194555842642e-06,
+      "loss": 0.3644,
+      "num_input_tokens_seen": 113866280,
+      "step": 2075,
+      "train_runtime": 11683.4191,
+      "train_tokens_per_second": 9745.972
+    },
+    {
+      "epoch": 0.87248322147651,
+      "grad_norm": 0.7340436795920906,
+      "learning_rate": 5.998059228074629e-06,
+      "loss": 0.3486,
+      "num_input_tokens_seen": 114155624,
+      "step": 2080,
+      "train_runtime": 11708.3345,
+      "train_tokens_per_second": 9749.946
+    },
+    {
+      "epoch": 0.8745805369127517,
+      "grad_norm": 0.8125020822426324,
+      "learning_rate": 5.981913067944187e-06,
+      "loss": 0.3713,
+      "num_input_tokens_seen": 114405504,
+      "step": 2085,
+      "train_runtime": 11737.0069,
+      "train_tokens_per_second": 9747.417
+    },
+    {
+      "epoch": 0.8766778523489933,
+      "grad_norm": 0.6455806774383137,
+      "learning_rate": 5.965756250692475e-06,
+      "loss": 0.366,
+      "num_input_tokens_seen": 114709744,
+      "step": 2090,
+      "train_runtime": 11765.6242,
+      "train_tokens_per_second": 9749.567
+    },
+    {
+      "epoch": 0.8787751677852349,
+      "grad_norm": 0.6779737877596309,
+      "learning_rate": 5.949588951676319e-06,
+      "loss": 0.3559,
+      "num_input_tokens_seen": 114985664,
+      "step": 2095,
+      "train_runtime": 11795.9909,
+      "train_tokens_per_second": 9747.86
+    },
+    {
+      "epoch": 0.8808724832214765,
+      "grad_norm": 0.6953288853371343,
+      "learning_rate": 5.933411346366315e-06,
+      "loss": 0.388,
+      "num_input_tokens_seen": 115270440,
+      "step": 2100,
+      "train_runtime": 11822.576,
+      "train_tokens_per_second": 9750.027
+    },
+    {
+      "epoch": 0.8829697986577181,
+      "grad_norm": 0.6752129470017262,
+      "learning_rate": 5.9172236103449065e-06,
+      "loss": 0.3525,
+      "num_input_tokens_seen": 115558176,
+      "step": 2105,
+      "train_runtime": 11850.2412,
+      "train_tokens_per_second": 9751.546
+    },
+    {
+      "epoch": 0.8850671140939598,
+      "grad_norm": 0.6359500192138051,
+      "learning_rate": 5.9010259193045e-06,
+      "loss": 0.3532,
+      "num_input_tokens_seen": 115838936,
+      "step": 2110,
+      "train_runtime": 11877.8167,
+      "train_tokens_per_second": 9752.545
+    },
+    {
+      "epoch": 0.8871644295302014,
+      "grad_norm": 0.697112795475774,
+      "learning_rate": 5.884818449045544e-06,
+      "loss": 0.3762,
+      "num_input_tokens_seen": 116134264,
+      "step": 2115,
+      "train_runtime": 11902.1869,
+      "train_tokens_per_second": 9757.389
+    },
+    {
+      "epoch": 0.889261744966443,
+      "grad_norm": 0.6931041051087072,
+      "learning_rate": 5.868601375474622e-06,
+      "loss": 0.3599,
+      "num_input_tokens_seen": 116438928,
+      "step": 2120,
+      "train_runtime": 11932.8149,
+      "train_tokens_per_second": 9757.876
+    },
+    {
+      "epoch": 0.8913590604026845,
+      "grad_norm": 0.6779366731701167,
+      "learning_rate": 5.852374874602552e-06,
+      "loss": 0.3508,
+      "num_input_tokens_seen": 116699808,
+      "step": 2125,
+      "train_runtime": 11957.2239,
+      "train_tokens_per_second": 9759.774
+    },
+    {
+      "epoch": 0.8934563758389261,
+      "grad_norm": 0.8177494726982232,
+      "learning_rate": 5.8361391225424655e-06,
+      "loss": 0.3792,
+      "num_input_tokens_seen": 116976528,
+      "step": 2130,
+      "train_runtime": 11988.5557,
+      "train_tokens_per_second": 9757.349
+    },
+    {
+      "epoch": 0.8955536912751678,
+      "grad_norm": 0.7249373619627936,
+      "learning_rate": 5.819894295507907e-06,
+      "loss": 0.3592,
+      "num_input_tokens_seen": 117283256,
+      "step": 2135,
+      "train_runtime": 12014.0886,
+      "train_tokens_per_second": 9762.143
+    },
+    {
+      "epoch": 0.8976510067114094,
+      "grad_norm": 0.7361206275749945,
+      "learning_rate": 5.80364056981091e-06,
+      "loss": 0.3805,
+      "num_input_tokens_seen": 117566056,
+      "step": 2140,
+      "train_runtime": 12039.8273,
+      "train_tokens_per_second": 9764.763
+    },
+    {
+      "epoch": 0.899748322147651,
+      "grad_norm": 0.6873698697925683,
+      "learning_rate": 5.78737812186009e-06,
+      "loss": 0.371,
+      "num_input_tokens_seen": 117852440,
+      "step": 2145,
+      "train_runtime": 12071.9214,
+      "train_tokens_per_second": 9762.525
+    },
+    {
+      "epoch": 0.9018456375838926,
+      "grad_norm": 0.7657116567605219,
+      "learning_rate": 5.771107128158731e-06,
+      "loss": 0.3676,
+      "num_input_tokens_seen": 118131472,
+      "step": 2150,
+      "train_runtime": 12101.8753,
+      "train_tokens_per_second": 9761.419
+    },
+    {
+      "epoch": 0.9039429530201343,
+      "grad_norm": 0.6478079187275864,
+      "learning_rate": 5.754827765302866e-06,
+      "loss": 0.3691,
+      "num_input_tokens_seen": 118414352,
+      "step": 2155,
+      "train_runtime": 12135.3749,
+      "train_tokens_per_second": 9757.783
+    },
+    {
+      "epoch": 0.9060402684563759,
+      "grad_norm": 0.6928796857910786,
+      "learning_rate": 5.738540209979363e-06,
+      "loss": 0.3787,
+      "num_input_tokens_seen": 118711200,
+      "step": 2160,
+      "train_runtime": 12166.7128,
+      "train_tokens_per_second": 9757.048
+    },
+    {
+      "epoch": 0.9081375838926175,
+      "grad_norm": 0.7152953100821898,
+      "learning_rate": 5.722244638964005e-06,
+      "loss": 0.3739,
+      "num_input_tokens_seen": 118992520,
+      "step": 2165,
+      "train_runtime": 12198.8831,
+      "train_tokens_per_second": 9754.378
+    },
+    {
+      "epoch": 0.910234899328859,
+      "grad_norm": 0.6405178758288701,
+      "learning_rate": 5.705941229119576e-06,
+      "loss": 0.3517,
+      "num_input_tokens_seen": 119291728,
+      "step": 2170,
+      "train_runtime": 12230.625,
+      "train_tokens_per_second": 9753.527
+    },
+    {
+      "epoch": 0.9123322147651006,
+      "grad_norm": 0.6978775933292674,
+      "learning_rate": 5.6896301573939336e-06,
+      "loss": 0.3531,
+      "num_input_tokens_seen": 119538056,
+      "step": 2175,
+      "train_runtime": 12255.7144,
+      "train_tokens_per_second": 9753.659
+    },
+    {
+      "epoch": 0.9144295302013423,
+      "grad_norm": 0.652542439099466,
+      "learning_rate": 5.673311600818096e-06,
+      "loss": 0.3552,
+      "num_input_tokens_seen": 119813840,
+      "step": 2180,
+      "train_runtime": 12282.0699,
+      "train_tokens_per_second": 9755.183
+    },
+    {
+      "epoch": 0.9165268456375839,
+      "grad_norm": 0.660364217231358,
+      "learning_rate": 5.656985736504314e-06,
+      "loss": 0.3556,
+      "num_input_tokens_seen": 120097928,
+      "step": 2185,
+      "train_runtime": 12306.2132,
+      "train_tokens_per_second": 9759.129
+    },
+    {
+      "epoch": 0.9186241610738255,
+      "grad_norm": 0.7200920433069488,
+      "learning_rate": 5.6406527416441605e-06,
+      "loss": 0.3495,
+      "num_input_tokens_seen": 120377784,
+      "step": 2190,
+      "train_runtime": 12332.8521,
+      "train_tokens_per_second": 9760.742
+    },
+    {
+      "epoch": 0.9207214765100671,
+      "grad_norm": 0.7305690282507485,
+      "learning_rate": 5.62431279350659e-06,
+      "loss": 0.3307,
+      "num_input_tokens_seen": 120659448,
+      "step": 2195,
+      "train_runtime": 12359.8251,
+      "train_tokens_per_second": 9762.229
+    },
+    {
+      "epoch": 0.9228187919463087,
+      "grad_norm": 0.7312908538870769,
+      "learning_rate": 5.6079660694360305e-06,
+      "loss": 0.3525,
+      "num_input_tokens_seen": 120910272,
+      "step": 2200,
+      "train_runtime": 12389.2877,
+      "train_tokens_per_second": 9759.259
+    },
+    {
+      "epoch": 0.9249161073825504,
+      "grad_norm": 0.7045468501169404,
+      "learning_rate": 5.591612746850448e-06,
+      "loss": 0.3622,
+      "num_input_tokens_seen": 121167584,
+      "step": 2205,
+      "train_runtime": 12420.9679,
+      "train_tokens_per_second": 9755.084
+    },
+    {
+      "epoch": 0.927013422818792,
+      "grad_norm": 0.7042585969665877,
+      "learning_rate": 5.5752530032394245e-06,
+      "loss": 0.3566,
+      "num_input_tokens_seen": 121464672,
+      "step": 2210,
+      "train_runtime": 12450.7215,
+      "train_tokens_per_second": 9755.633
+    },
+    {
+      "epoch": 0.9291107382550335,
+      "grad_norm": 0.6908580729615781,
+      "learning_rate": 5.558887016162238e-06,
+      "loss": 0.3563,
+      "num_input_tokens_seen": 121733560,
+      "step": 2215,
+      "train_runtime": 12479.6857,
+      "train_tokens_per_second": 9754.537
+    },
+    {
+      "epoch": 0.9312080536912751,
+      "grad_norm": 0.8032507386041652,
+      "learning_rate": 5.542514963245924e-06,
+      "loss": 0.3781,
+      "num_input_tokens_seen": 121973784,
+      "step": 2220,
+      "train_runtime": 12504.1687,
+      "train_tokens_per_second": 9754.65
+    },
+    {
+      "epoch": 0.9333053691275168,
+      "grad_norm": 0.7651340961152648,
+      "learning_rate": 5.526137022183356e-06,
+      "loss": 0.3784,
+      "num_input_tokens_seen": 122225480,
+      "step": 2225,
+      "train_runtime": 12533.3442,
+      "train_tokens_per_second": 9752.025
+    },
+    {
+      "epoch": 0.9354026845637584,
+      "grad_norm": 0.7608671435182088,
+      "learning_rate": 5.50975337073131e-06,
+      "loss": 0.3686,
+      "num_input_tokens_seen": 122485480,
+      "step": 2230,
+      "train_runtime": 12566.9453,
+      "train_tokens_per_second": 9746.639
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 0.6622476866965753,
+      "learning_rate": 5.493364186708544e-06,
+      "loss": 0.3611,
+      "num_input_tokens_seen": 122762392,
+      "step": 2235,
+      "train_runtime": 12597.985,
+      "train_tokens_per_second": 9744.605
+    },
+    {
+      "epoch": 0.9395973154362416,
+      "grad_norm": 0.8583780797999162,
+      "learning_rate": 5.476969647993861e-06,
+      "loss": 0.3516,
+      "num_input_tokens_seen": 123046920,
+      "step": 2240,
+      "train_runtime": 12621.5906,
+      "train_tokens_per_second": 9748.923
+    },
+    {
+      "epoch": 0.9416946308724832,
+      "grad_norm": 0.7099771472433337,
+      "learning_rate": 5.460569932524181e-06,
+      "loss": 0.3455,
+      "num_input_tokens_seen": 123330960,
+      "step": 2245,
+      "train_runtime": 12653.1151,
+      "train_tokens_per_second": 9747.083
+    },
+    {
+      "epoch": 0.9437919463087249,
+      "grad_norm": 0.6907116449434365,
+      "learning_rate": 5.444165218292612e-06,
+      "loss": 0.3306,
+      "num_input_tokens_seen": 123601096,
+      "step": 2250,
+      "train_runtime": 12687.1994,
+      "train_tokens_per_second": 9742.189
+    },
+    {
+      "epoch": 0.9458892617449665,
+      "grad_norm": 0.638952460576574,
+      "learning_rate": 5.42775568334651e-06,
+      "loss": 0.3468,
+      "num_input_tokens_seen": 123876808,
+      "step": 2255,
+      "train_runtime": 12714.8428,
+      "train_tokens_per_second": 9742.693
+    },
+    {
+      "epoch": 0.947986577181208,
+      "grad_norm": 0.6898841022866466,
+      "learning_rate": 5.411341505785559e-06,
+      "loss": 0.358,
+      "num_input_tokens_seen": 124125848,
+      "step": 2260,
+      "train_runtime": 12749.9525,
+      "train_tokens_per_second": 9735.397
+    },
+    {
+      "epoch": 0.9500838926174496,
+      "grad_norm": 0.6486668790612827,
+      "learning_rate": 5.3949228637598235e-06,
+      "loss": 0.336,
+      "num_input_tokens_seen": 124400608,
+      "step": 2265,
+      "train_runtime": 12775.484,
+      "train_tokens_per_second": 9737.448
+    },
+    {
+      "epoch": 0.9521812080536913,
+      "grad_norm": 0.6734858795997746,
+      "learning_rate": 5.378499935467834e-06,
+      "loss": 0.3522,
+      "num_input_tokens_seen": 124676784,
+      "step": 2270,
+      "train_runtime": 12798.3802,
+      "train_tokens_per_second": 9741.607
+    },
+    {
+      "epoch": 0.9542785234899329,
+      "grad_norm": 0.692280828595565,
+      "learning_rate": 5.362072899154629e-06,
+      "loss": 0.3487,
+      "num_input_tokens_seen": 124946680,
+      "step": 2275,
+      "train_runtime": 12825.879,
+      "train_tokens_per_second": 9741.764
+    },
+    {
+      "epoch": 0.9563758389261745,
+      "grad_norm": 0.7283372074062345,
+      "learning_rate": 5.345641933109841e-06,
+      "loss": 0.3777,
+      "num_input_tokens_seen": 125246288,
+      "step": 2280,
+      "train_runtime": 12854.4762,
+      "train_tokens_per_second": 9743.399
+    },
+    {
+      "epoch": 0.9584731543624161,
+      "grad_norm": 0.6583437157923052,
+      "learning_rate": 5.329207215665751e-06,
+      "loss": 0.3571,
+      "num_input_tokens_seen": 125520136,
+      "step": 2285,
+      "train_runtime": 12881.4935,
+      "train_tokens_per_second": 9744.222
+    },
+    {
+      "epoch": 0.9605704697986577,
+      "grad_norm": 0.6575022495089723,
+      "learning_rate": 5.312768925195359e-06,
+      "loss": 0.349,
+      "num_input_tokens_seen": 125794104,
+      "step": 2290,
+      "train_runtime": 12910.5285,
+      "train_tokens_per_second": 9743.529
+    },
+    {
+      "epoch": 0.9626677852348994,
+      "grad_norm": 0.652927941615282,
+      "learning_rate": 5.296327240110437e-06,
+      "loss": 0.3402,
+      "num_input_tokens_seen": 126065880,
+      "step": 2295,
+      "train_runtime": 12939.6521,
+      "train_tokens_per_second": 9742.602
+    },
+    {
+      "epoch": 0.964765100671141,
+      "grad_norm": 0.6896101956662979,
+      "learning_rate": 5.279882338859609e-06,
+      "loss": 0.3541,
+      "num_input_tokens_seen": 126358456,
+      "step": 2300,
+      "train_runtime": 12967.0758,
+      "train_tokens_per_second": 9744.561
+    },
+    {
+      "epoch": 0.9668624161073825,
+      "grad_norm": 0.6713321951264678,
+      "learning_rate": 5.2634343999263985e-06,
+      "loss": 0.3632,
+      "num_input_tokens_seen": 126624952,
+      "step": 2305,
+      "train_runtime": 12993.1614,
+      "train_tokens_per_second": 9745.507
+    },
+    {
+      "epoch": 0.9689597315436241,
+      "grad_norm": 0.6033121271445016,
+      "learning_rate": 5.246983601827302e-06,
+      "loss": 0.3681,
+      "num_input_tokens_seen": 126888696,
+      "step": 2310,
+      "train_runtime": 13021.9148,
+      "train_tokens_per_second": 9744.243
+    },
+    {
+      "epoch": 0.9710570469798657,
+      "grad_norm": 0.6666113189092139,
+      "learning_rate": 5.230530123109849e-06,
+      "loss": 0.3547,
+      "num_input_tokens_seen": 127170608,
+      "step": 2315,
+      "train_runtime": 13047.0358,
+      "train_tokens_per_second": 9747.088
+    },
+    {
+      "epoch": 0.9731543624161074,
+      "grad_norm": 0.7380294257052984,
+      "learning_rate": 5.214074142350657e-06,
+      "loss": 0.3773,
+      "num_input_tokens_seen": 127453496,
+      "step": 2320,
+      "train_runtime": 13074.8018,
+      "train_tokens_per_second": 9748.025
+    },
+    {
+      "epoch": 0.975251677852349,
+      "grad_norm": 0.7529049555698176,
+      "learning_rate": 5.197615838153506e-06,
+      "loss": 0.3512,
+      "num_input_tokens_seen": 127718840,
+      "step": 2325,
+      "train_runtime": 13103.1005,
+      "train_tokens_per_second": 9747.223
+    },
+    {
+      "epoch": 0.9773489932885906,
+      "grad_norm": 0.7344867116638953,
+      "learning_rate": 5.1811553891473874e-06,
+      "loss": 0.3522,
+      "num_input_tokens_seen": 127969776,
+      "step": 2330,
+      "train_runtime": 13130.4459,
+      "train_tokens_per_second": 9746.034
+    },
+    {
+      "epoch": 0.9794463087248322,
+      "grad_norm": 0.6710530330440179,
+      "learning_rate": 5.164692973984577e-06,
+      "loss": 0.3317,
+      "num_input_tokens_seen": 128267536,
+      "step": 2335,
+      "train_runtime": 13160.7147,
+      "train_tokens_per_second": 9746.244
+    },
+    {
+      "epoch": 0.9815436241610739,
+      "grad_norm": 0.6611945786937319,
+      "learning_rate": 5.148228771338686e-06,
+      "loss": 0.3294,
+      "num_input_tokens_seen": 128562888,
+      "step": 2340,
+      "train_runtime": 13197.1456,
+      "train_tokens_per_second": 9741.719
+    },
+    {
+      "epoch": 0.9836409395973155,
+      "grad_norm": 0.6876531532820319,
+      "learning_rate": 5.131762959902724e-06,
+      "loss": 0.3568,
+      "num_input_tokens_seen": 128842592,
+      "step": 2345,
+      "train_runtime": 13223.2988,
+      "train_tokens_per_second": 9743.604
+    },
+    {
+      "epoch": 0.985738255033557,
+      "grad_norm": 0.6991375966968375,
+      "learning_rate": 5.1152957183871685e-06,
+      "loss": 0.3764,
+      "num_input_tokens_seen": 129122664,
+      "step": 2350,
+      "train_runtime": 13256.0064,
+      "train_tokens_per_second": 9740.691
+    },
+    {
+      "epoch": 0.9878355704697986,
+      "grad_norm": 0.7078250401990933,
+      "learning_rate": 5.0988272255180125e-06,
+      "loss": 0.3716,
+      "num_input_tokens_seen": 129390832,
+      "step": 2355,
+      "train_runtime": 13286.4447,
+      "train_tokens_per_second": 9738.559
+    },
+    {
+      "epoch": 0.9899328859060402,
+      "grad_norm": 0.9953038634969839,
+      "learning_rate": 5.082357660034833e-06,
+      "loss": 0.347,
+      "num_input_tokens_seen": 129653448,
+      "step": 2360,
+      "train_runtime": 13311.6807,
+      "train_tokens_per_second": 9739.826
+    },
+    {
+      "epoch": 0.9920302013422819,
+      "grad_norm": 0.7201538760123838,
+      "learning_rate": 5.065887200688847e-06,
+      "loss": 0.3475,
+      "num_input_tokens_seen": 129919288,
+      "step": 2365,
+      "train_runtime": 13341.9381,
+      "train_tokens_per_second": 9737.662
+    },
+    {
+      "epoch": 0.9941275167785235,
+      "grad_norm": 0.7869930986152237,
+      "learning_rate": 5.049416026240973e-06,
+      "loss": 0.3469,
+      "num_input_tokens_seen": 130174440,
+      "step": 2370,
+      "train_runtime": 13365.3829,
+      "train_tokens_per_second": 9739.672
+    },
+    {
+      "epoch": 0.9962248322147651,
+      "grad_norm": 0.6388028010124842,
+      "learning_rate": 5.032944315459892e-06,
+      "loss": 0.3631,
+      "num_input_tokens_seen": 130445928,
+      "step": 2375,
+      "train_runtime": 13398.1327,
+      "train_tokens_per_second": 9736.127
+    },
+    {
+      "epoch": 0.9983221476510067,
+      "grad_norm": 0.6303257522899145,
+      "learning_rate": 5.0164722471201075e-06,
+      "loss": 0.3429,
+      "num_input_tokens_seen": 130728448,
+      "step": 2380,
+      "train_runtime": 13432.8966,
+      "train_tokens_per_second": 9731.963
+    },
+    {
+      "epoch": 1.0004194630872483,
+      "grad_norm": 0.6567018376901954,
+      "learning_rate": 5e-06,
+      "loss": 0.3219,
+      "num_input_tokens_seen": 130994144,
+      "step": 2385,
+      "train_runtime": 13459.4309,
+      "train_tokens_per_second": 9732.517
+    },
+    {
+      "epoch": 1.0025167785234899,
+      "grad_norm": 0.7589074870617772,
+      "learning_rate": 4.983527752879893e-06,
+      "loss": 0.2976,
+      "num_input_tokens_seen": 131256680,
+      "step": 2390,
+      "train_runtime": 13482.8795,
+      "train_tokens_per_second": 9735.063
+    },
+    {
+      "epoch": 1.0046140939597314,
+      "grad_norm": 0.7424601837666054,
+      "learning_rate": 4.967055684540109e-06,
+      "loss": 0.3204,
+      "num_input_tokens_seen": 131526592,
+      "step": 2395,
+      "train_runtime": 13508.9237,
+      "train_tokens_per_second": 9736.275
+    },
+    {
+      "epoch": 1.0067114093959733,
+      "grad_norm": 0.7508933430428305,
+      "learning_rate": 4.950583973759029e-06,
+      "loss": 0.3081,
+      "num_input_tokens_seen": 131808000,
+      "step": 2400,
+      "train_runtime": 13534.3493,
+      "train_tokens_per_second": 9738.776
+    },
+    {
+      "epoch": 1.0088087248322148,
+      "grad_norm": 1.1268979473534804,
+      "learning_rate": 4.934112799311155e-06,
+      "loss": 0.3224,
+      "num_input_tokens_seen": 132082072,
+      "step": 2405,
+      "train_runtime": 13556.5969,
+      "train_tokens_per_second": 9743.011
+    },
+    {
+      "epoch": 1.0109060402684564,
+      "grad_norm": 0.721797413073044,
+      "learning_rate": 4.9176423399651675e-06,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 132358432,
+      "step": 2410,
+      "train_runtime": 13584.9751,
+      "train_tokens_per_second": 9743.001
+    },
+    {
+      "epoch": 1.013003355704698,
+      "grad_norm": 0.6834673966178163,
+      "learning_rate": 4.901172774481988e-06,
+      "loss": 0.3157,
+      "num_input_tokens_seen": 132652912,
+      "step": 2415,
+      "train_runtime": 13611.4443,
+      "train_tokens_per_second": 9745.69
+    },
+    {
+      "epoch": 1.0151006711409396,
+      "grad_norm": 0.6411622725513136,
+      "learning_rate": 4.884704281612832e-06,
+      "loss": 0.2856,
+      "num_input_tokens_seen": 132913856,
+      "step": 2420,
+      "train_runtime": 13637.2246,
+      "train_tokens_per_second": 9746.401
+    },
+    {
+      "epoch": 1.0171979865771812,
+      "grad_norm": 0.7609301600954341,
+      "learning_rate": 4.8682370400972774e-06,
+      "loss": 0.3145,
+      "num_input_tokens_seen": 133190432,
+      "step": 2425,
+      "train_runtime": 13663.1492,
+      "train_tokens_per_second": 9748.15
+    },
+    {
+      "epoch": 1.0192953020134228,
+      "grad_norm": 0.7370749849359561,
+      "learning_rate": 4.851771228661316e-06,
+      "loss": 0.3268,
+      "num_input_tokens_seen": 133478560,
+      "step": 2430,
+      "train_runtime": 13686.238,
+      "train_tokens_per_second": 9752.758
+    },
+    {
+      "epoch": 1.0213926174496644,
+      "grad_norm": 0.7253848595859842,
+      "learning_rate": 4.8353070260154244e-06,
+      "loss": 0.3132,
+      "num_input_tokens_seen": 133733920,
+      "step": 2435,
+      "train_runtime": 13715.0944,
+      "train_tokens_per_second": 9750.857
+    },
+    {
+      "epoch": 1.023489932885906,
+      "grad_norm": 0.7000802750939579,
+      "learning_rate": 4.818844610852613e-06,
+      "loss": 0.3145,
+      "num_input_tokens_seen": 134009288,
+      "step": 2440,
+      "train_runtime": 13744.5409,
+      "train_tokens_per_second": 9750.001
+    },
+    {
+      "epoch": 1.0255872483221478,
+      "grad_norm": 0.733052920354238,
+      "learning_rate": 4.802384161846496e-06,
+      "loss": 0.3215,
+      "num_input_tokens_seen": 134298360,
+      "step": 2445,
+      "train_runtime": 13778.873,
+      "train_tokens_per_second": 9746.687
+    },
+    {
+      "epoch": 1.0276845637583893,
+      "grad_norm": 0.7435703075966702,
+      "learning_rate": 4.785925857649344e-06,
+      "loss": 0.3177,
+      "num_input_tokens_seen": 134589128,
+      "step": 2450,
+      "train_runtime": 13807.4058,
+      "train_tokens_per_second": 9747.604
+    },
+    {
+      "epoch": 1.029781879194631,
+      "grad_norm": 0.7015001945931422,
+      "learning_rate": 4.769469876890152e-06,
+      "loss": 0.3004,
+      "num_input_tokens_seen": 134851744,
+      "step": 2455,
+      "train_runtime": 13837.9203,
+      "train_tokens_per_second": 9745.087
+    },
+    {
+      "epoch": 1.0318791946308725,
+      "grad_norm": 0.6164168543059595,
+      "learning_rate": 4.753016398172697e-06,
+      "loss": 0.3095,
+      "num_input_tokens_seen": 135140728,
+      "step": 2460,
+      "train_runtime": 13862.8573,
+      "train_tokens_per_second": 9748.404
+    },
+    {
+      "epoch": 1.033976510067114,
+      "grad_norm": 0.7577280144964795,
+      "learning_rate": 4.736565600073602e-06,
+      "loss": 0.3281,
+      "num_input_tokens_seen": 135412664,
+      "step": 2465,
+      "train_runtime": 13893.5223,
+      "train_tokens_per_second": 9746.46
+    },
+    {
+      "epoch": 1.0360738255033557,
+      "grad_norm": 0.8701325948002333,
+      "learning_rate": 4.7201176611403935e-06,
+      "loss": 0.3223,
+      "num_input_tokens_seen": 135676848,
+      "step": 2470,
+      "train_runtime": 13921.1785,
+      "train_tokens_per_second": 9746.075
+    },
+    {
+      "epoch": 1.0381711409395973,
+      "grad_norm": 0.7302851726200538,
+      "learning_rate": 4.7036727598895645e-06,
+      "loss": 0.298,
+      "num_input_tokens_seen": 135943072,
+      "step": 2475,
+      "train_runtime": 13951.1496,
+      "train_tokens_per_second": 9744.22
+    },
+    {
+      "epoch": 1.0402684563758389,
+      "grad_norm": 0.692946074391214,
+      "learning_rate": 4.687231074804643e-06,
+      "loss": 0.3127,
+      "num_input_tokens_seen": 136205664,
+      "step": 2480,
+      "train_runtime": 13980.5777,
+      "train_tokens_per_second": 9742.492
+    },
+    {
+      "epoch": 1.0423657718120805,
+      "grad_norm": 0.8520589508914272,
+      "learning_rate": 4.6707927843342505e-06,
+      "loss": 0.3041,
+      "num_input_tokens_seen": 136512584,
+      "step": 2485,
+      "train_runtime": 14006.1119,
+      "train_tokens_per_second": 9746.644
+    },
+    {
+      "epoch": 1.044463087248322,
+      "grad_norm": 0.7329117323374387,
+      "learning_rate": 4.654358066890161e-06,
+      "loss": 0.3248,
+      "num_input_tokens_seen": 136782648,
+      "step": 2490,
+      "train_runtime": 14032.5053,
+      "train_tokens_per_second": 9747.557
+    },
+    {
+      "epoch": 1.0465604026845639,
+      "grad_norm": 0.7054244470263111,
+      "learning_rate": 4.637927100845374e-06,
+      "loss": 0.3117,
+      "num_input_tokens_seen": 137082912,
+      "step": 2495,
+      "train_runtime": 14060.9058,
+      "train_tokens_per_second": 9749.223
+    },
+    {
+      "epoch": 1.0486577181208054,
+      "grad_norm": 0.6660997252498075,
+      "learning_rate": 4.6215000645321696e-06,
+      "loss": 0.3006,
+      "num_input_tokens_seen": 137348120,
+      "step": 2500,
+      "train_runtime": 14085.6791,
+      "train_tokens_per_second": 9750.905
+    },
+    {
+      "epoch": 1.050755033557047,
+      "grad_norm": 0.8311960669954274,
+      "learning_rate": 4.6050771362401764e-06,
+      "loss": 0.3175,
+      "num_input_tokens_seen": 137623472,
+      "step": 2505,
+      "train_runtime": 14110.3859,
+      "train_tokens_per_second": 9753.346
+    },
+    {
+      "epoch": 1.0528523489932886,
+      "grad_norm": 0.7835417531528674,
+      "learning_rate": 4.588658494214443e-06,
+      "loss": 0.2994,
+      "num_input_tokens_seen": 137897016,
+      "step": 2510,
+      "train_runtime": 14136.0291,
+      "train_tokens_per_second": 9755.004
+    },
+    {
+      "epoch": 1.0549496644295302,
+      "grad_norm": 0.7596859748700587,
+      "learning_rate": 4.5722443166534904e-06,
+      "loss": 0.3303,
+      "num_input_tokens_seen": 138175592,
+      "step": 2515,
+      "train_runtime": 14167.7397,
+      "train_tokens_per_second": 9752.832
+    },
+    {
+      "epoch": 1.0570469798657718,
+      "grad_norm": 0.6665469843633857,
+      "learning_rate": 4.55583478170739e-06,
+      "loss": 0.3114,
+      "num_input_tokens_seen": 138442664,
+      "step": 2520,
+      "train_runtime": 14201.189,
+      "train_tokens_per_second": 9748.667
+    },
+    {
+      "epoch": 1.0591442953020134,
+      "grad_norm": 0.6470503974790873,
+      "learning_rate": 4.53943006747582e-06,
+      "loss": 0.2945,
+      "num_input_tokens_seen": 138713048,
+      "step": 2525,
+      "train_runtime": 14228.9529,
+      "train_tokens_per_second": 9748.648
+    },
+    {
+      "epoch": 1.061241610738255,
+      "grad_norm": 0.7739556999755998,
+      "learning_rate": 4.523030352006141e-06,
+      "loss": 0.2927,
+      "num_input_tokens_seen": 138983832,
+      "step": 2530,
+      "train_runtime": 14251.9364,
+      "train_tokens_per_second": 9751.926
+    },
+    {
+      "epoch": 1.0633389261744965,
+      "grad_norm": 0.6974372576404982,
+      "learning_rate": 4.506635813291458e-06,
+      "loss": 0.3324,
+      "num_input_tokens_seen": 139251408,
+      "step": 2535,
+      "train_runtime": 14273.7986,
+      "train_tokens_per_second": 9755.736
+    },
+    {
+      "epoch": 1.0654362416107384,
+      "grad_norm": 0.6304602381647292,
+      "learning_rate": 4.490246629268692e-06,
+      "loss": 0.2862,
+      "num_input_tokens_seen": 139537568,
+      "step": 2540,
+      "train_runtime": 14297.0093,
+      "train_tokens_per_second": 9759.913
+    },
+    {
+      "epoch": 1.06753355704698,
+      "grad_norm": 0.7109840326265371,
+      "learning_rate": 4.473862977816647e-06,
+      "loss": 0.3313,
+      "num_input_tokens_seen": 139809368,
+      "step": 2545,
+      "train_runtime": 14319.7553,
+      "train_tokens_per_second": 9763.391
+    },
+    {
+      "epoch": 1.0696308724832215,
+      "grad_norm": 0.6677686910113535,
+      "learning_rate": 4.457485036754076e-06,
+      "loss": 0.34,
+      "num_input_tokens_seen": 140078064,
+      "step": 2550,
+      "train_runtime": 14350.8619,
+      "train_tokens_per_second": 9760.951
+    },
+    {
+      "epoch": 1.0717281879194631,
+      "grad_norm": 0.6755168929848873,
+      "learning_rate": 4.441112983837763e-06,
+      "loss": 0.3042,
+      "num_input_tokens_seen": 140359384,
+      "step": 2555,
+      "train_runtime": 14379.7793,
+      "train_tokens_per_second": 9760.886
+    },
+    {
+      "epoch": 1.0738255033557047,
+      "grad_norm": 0.6297990180289057,
+      "learning_rate": 4.424746996760576e-06,
+      "loss": 0.279,
+      "num_input_tokens_seen": 140651920,
+      "step": 2560,
+      "train_runtime": 14410.0471,
+      "train_tokens_per_second": 9760.684
+    },
+    {
+      "epoch": 1.0759228187919463,
+      "grad_norm": 0.7223003533230274,
+      "learning_rate": 4.408387253149553e-06,
+      "loss": 0.3002,
+      "num_input_tokens_seen": 140946536,
+      "step": 2565,
+      "train_runtime": 14442.7524,
+      "train_tokens_per_second": 9758.98
+    },
+    {
+      "epoch": 1.0780201342281879,
+      "grad_norm": 0.6479782816830172,
+      "learning_rate": 4.392033930563971e-06,
+      "loss": 0.3025,
+      "num_input_tokens_seen": 141207464,
+      "step": 2570,
+      "train_runtime": 14471.0678,
+      "train_tokens_per_second": 9757.916
+    },
+    {
+      "epoch": 1.0801174496644295,
+      "grad_norm": 0.7462221539768958,
+      "learning_rate": 4.375687206493411e-06,
+      "loss": 0.3052,
+      "num_input_tokens_seen": 141481520,
+      "step": 2575,
+      "train_runtime": 14499.7179,
+      "train_tokens_per_second": 9757.536
+    },
+    {
+      "epoch": 1.082214765100671,
+      "grad_norm": 0.7499688931363782,
+      "learning_rate": 4.359347258355842e-06,
+      "loss": 0.3101,
+      "num_input_tokens_seen": 141776872,
+      "step": 2580,
+      "train_runtime": 14526.3224,
+      "train_tokens_per_second": 9759.998
+    },
+    {
+      "epoch": 1.0843120805369129,
+      "grad_norm": 0.7385326605583298,
+      "learning_rate": 4.3430142634956866e-06,
+      "loss": 0.3345,
+      "num_input_tokens_seen": 142064872,
+      "step": 2585,
+      "train_runtime": 14553.6012,
+      "train_tokens_per_second": 9761.493
+    },
+    {
+      "epoch": 1.0864093959731544,
+      "grad_norm": 0.8033364208827543,
+      "learning_rate": 4.326688399181907e-06,
+      "loss": 0.3042,
+      "num_input_tokens_seen": 142318544,
+      "step": 2590,
+      "train_runtime": 14580.8819,
+      "train_tokens_per_second": 9760.627
+    },
+    {
+      "epoch": 1.088506711409396,
+      "grad_norm": 0.8331136922782929,
+      "learning_rate": 4.310369842606066e-06,
+      "loss": 0.3058,
+      "num_input_tokens_seen": 142574008,
+      "step": 2595,
+      "train_runtime": 14608.2576,
+      "train_tokens_per_second": 9759.823
+    },
+    {
+      "epoch": 1.0906040268456376,
+      "grad_norm": 0.7183576625289043,
+      "learning_rate": 4.294058770880424e-06,
+      "loss": 0.3055,
+      "num_input_tokens_seen": 142847384,
+      "step": 2600,
+      "train_runtime": 14637.0992,
+      "train_tokens_per_second": 9759.269
+    },
+    {
+      "epoch": 1.0927013422818792,
+      "grad_norm": 0.8261147853515342,
+      "learning_rate": 4.277755361035995e-06,
+      "loss": 0.287,
+      "num_input_tokens_seen": 143110936,
+      "step": 2605,
+      "train_runtime": 14661.2479,
+      "train_tokens_per_second": 9761.17
+    },
+    {
+      "epoch": 1.0947986577181208,
+      "grad_norm": 0.7422361988519603,
+      "learning_rate": 4.261459790020638e-06,
+      "loss": 0.302,
+      "num_input_tokens_seen": 143386016,
+      "step": 2610,
+      "train_runtime": 14693.7591,
+      "train_tokens_per_second": 9758.294
+    },
+    {
+      "epoch": 1.0968959731543624,
+      "grad_norm": 0.7056190205816357,
+      "learning_rate": 4.2451722346971355e-06,
+      "loss": 0.3037,
+      "num_input_tokens_seen": 143670680,
+      "step": 2615,
+      "train_runtime": 14724.6567,
+      "train_tokens_per_second": 9757.15
+    },
+    {
+      "epoch": 1.098993288590604,
+      "grad_norm": 0.6892515643216794,
+      "learning_rate": 4.22889287184127e-06,
+      "loss": 0.2993,
+      "num_input_tokens_seen": 143959232,
+      "step": 2620,
+      "train_runtime": 14752.7582,
+      "train_tokens_per_second": 9758.123
+    },
+    {
+      "epoch": 1.1010906040268456,
+      "grad_norm": 0.880453189830704,
+      "learning_rate": 4.212621878139912e-06,
+      "loss": 0.2986,
+      "num_input_tokens_seen": 144227512,
+      "step": 2625,
+      "train_runtime": 14781.7474,
+      "train_tokens_per_second": 9757.135
+    },
+    {
+      "epoch": 1.1031879194630871,
+      "grad_norm": 0.7351937902047315,
+      "learning_rate": 4.196359430189092e-06,
+      "loss": 0.2975,
+      "num_input_tokens_seen": 144513160,
+      "step": 2630,
+      "train_runtime": 14810.8945,
+      "train_tokens_per_second": 9757.22
+    },
+    {
+      "epoch": 1.105285234899329,
+      "grad_norm": 0.7022062870001068,
+      "learning_rate": 4.180105704492096e-06,
+      "loss": 0.2917,
+      "num_input_tokens_seen": 144784408,
+      "step": 2635,
+      "train_runtime": 14839.5582,
+      "train_tokens_per_second": 9756.652
+    },
+    {
+      "epoch": 1.1073825503355705,
+      "grad_norm": 0.7557276421498977,
+      "learning_rate": 4.163860877457535e-06,
+      "loss": 0.3087,
+      "num_input_tokens_seen": 145040896,
+      "step": 2640,
+      "train_runtime": 14870.4584,
+      "train_tokens_per_second": 9753.626
+    },
+    {
+      "epoch": 1.1094798657718121,
+      "grad_norm": 0.8450696970313682,
+      "learning_rate": 4.14762512539745e-06,
+      "loss": 0.2937,
+      "num_input_tokens_seen": 145326632,
+      "step": 2645,
+      "train_runtime": 14897.6233,
+      "train_tokens_per_second": 9755.021
+    },
+    {
+      "epoch": 1.1115771812080537,
+      "grad_norm": 0.6855104087115983,
+      "learning_rate": 4.131398624525379e-06,
+      "loss": 0.3097,
+      "num_input_tokens_seen": 145594432,
+      "step": 2650,
+      "train_runtime": 14922.9632,
+      "train_tokens_per_second": 9756.402
+    },
+    {
+      "epoch": 1.1136744966442953,
+      "grad_norm": 0.702928049018474,
+      "learning_rate": 4.115181550954458e-06,
+      "loss": 0.3068,
+      "num_input_tokens_seen": 145884360,
+      "step": 2655,
+      "train_runtime": 14951.7522,
+      "train_tokens_per_second": 9757.008
+    },
+    {
+      "epoch": 1.1157718120805369,
+      "grad_norm": 0.7651304976839798,
+      "learning_rate": 4.098974080695501e-06,
+      "loss": 0.316,
+      "num_input_tokens_seen": 146178464,
+      "step": 2660,
+      "train_runtime": 14975.6531,
+      "train_tokens_per_second": 9761.074
+    },
+    {
+      "epoch": 1.1178691275167785,
+      "grad_norm": 0.7567540421813526,
+      "learning_rate": 4.082776389655095e-06,
+      "loss": 0.3014,
+      "num_input_tokens_seen": 146442200,
+      "step": 2665,
+      "train_runtime": 15001.6039,
+      "train_tokens_per_second": 9761.77
+    },
+    {
+      "epoch": 1.11996644295302,
+      "grad_norm": 0.6516948626060655,
+      "learning_rate": 4.066588653633688e-06,
+      "loss": 0.312,
+      "num_input_tokens_seen": 146716872,
+      "step": 2670,
+      "train_runtime": 15033.6781,
+      "train_tokens_per_second": 9759.213
+    },
+    {
+      "epoch": 1.1220637583892616,
+      "grad_norm": 0.6307873087047597,
+      "learning_rate": 4.050411048323682e-06,
+      "loss": 0.3025,
+      "num_input_tokens_seen": 147013632,
+      "step": 2675,
+      "train_runtime": 15062.6958,
+      "train_tokens_per_second": 9760.114
+    },
+    {
+      "epoch": 1.1241610738255035,
+      "grad_norm": 0.7929137756069601,
+      "learning_rate": 4.0342437493075285e-06,
+      "loss": 0.301,
+      "num_input_tokens_seen": 147294112,
+      "step": 2680,
+      "train_runtime": 15090.0632,
+      "train_tokens_per_second": 9761.0
+    },
+    {
+      "epoch": 1.126258389261745,
+      "grad_norm": 0.7172455366283252,
+      "learning_rate": 4.018086932055815e-06,
+      "loss": 0.3134,
+      "num_input_tokens_seen": 147583376,
+      "step": 2685,
+      "train_runtime": 15117.2915,
+      "train_tokens_per_second": 9762.554
+    },
+    {
+      "epoch": 1.1283557046979866,
+      "grad_norm": 0.6976001150385307,
+      "learning_rate": 4.001940771925371e-06,
+      "loss": 0.293,
+      "num_input_tokens_seen": 147822480,
+      "step": 2690,
+      "train_runtime": 15144.9216,
+      "train_tokens_per_second": 9760.531
+    },
+    {
+      "epoch": 1.1304530201342282,
+      "grad_norm": 0.8402803534076032,
+      "learning_rate": 3.98580544415736e-06,
+      "loss": 0.3171,
+      "num_input_tokens_seen": 148085096,
+      "step": 2695,
+      "train_runtime": 15175.4379,
+      "train_tokens_per_second": 9758.209
+    },
+    {
+      "epoch": 1.1325503355704698,
+      "grad_norm": 0.7563765142349673,
+      "learning_rate": 3.969681123875373e-06,
+      "loss": 0.3084,
+      "num_input_tokens_seen": 148314560,
+      "step": 2700,
+      "train_runtime": 15204.3378,
+      "train_tokens_per_second": 9754.753
+    },
+    {
+      "epoch": 1.1346476510067114,
+      "grad_norm": 0.6940784312713784,
+      "learning_rate": 3.953567986083535e-06,
+      "loss": 0.3083,
+      "num_input_tokens_seen": 148621416,
+      "step": 2705,
+      "train_runtime": 15232.0781,
+      "train_tokens_per_second": 9757.133
+    },
+    {
+      "epoch": 1.136744966442953,
+      "grad_norm": 0.6185787631310018,
+      "learning_rate": 3.937466205664602e-06,
+      "loss": 0.2994,
+      "num_input_tokens_seen": 148893168,
+      "step": 2710,
+      "train_runtime": 15255.2411,
+      "train_tokens_per_second": 9760.132
+    },
+    {
+      "epoch": 1.1388422818791946,
+      "grad_norm": 0.7543311377479143,
+      "learning_rate": 3.921375957378064e-06,
+      "loss": 0.308,
+      "num_input_tokens_seen": 149141168,
+      "step": 2715,
+      "train_runtime": 15285.5973,
+      "train_tokens_per_second": 9756.974
+    },
+    {
+      "epoch": 1.1409395973154361,
+      "grad_norm": 0.7307868152806485,
+      "learning_rate": 3.905297415858247e-06,
+      "loss": 0.3088,
+      "num_input_tokens_seen": 149380976,
+      "step": 2720,
+      "train_runtime": 15313.6551,
+      "train_tokens_per_second": 9754.756
+    },
+    {
+      "epoch": 1.1430369127516777,
+      "grad_norm": 0.7255552242009188,
+      "learning_rate": 3.889230755612417e-06,
+      "loss": 0.3074,
+      "num_input_tokens_seen": 149651248,
+      "step": 2725,
+      "train_runtime": 15340.9277,
+      "train_tokens_per_second": 9755.032
+    },
+    {
+      "epoch": 1.1451342281879195,
+      "grad_norm": 0.7225350697434612,
+      "learning_rate": 3.873176151018893e-06,
+      "loss": 0.2876,
+      "num_input_tokens_seen": 149921312,
+      "step": 2730,
+      "train_runtime": 15366.8382,
+      "train_tokens_per_second": 9756.159
+    },
+    {
+      "epoch": 1.1472315436241611,
+      "grad_norm": 0.6843269695691859,
+      "learning_rate": 3.857133776325142e-06,
+      "loss": 0.3037,
+      "num_input_tokens_seen": 150191784,
+      "step": 2735,
+      "train_runtime": 15392.1003,
+      "train_tokens_per_second": 9757.719
+    },
+    {
+      "epoch": 1.1493288590604027,
+      "grad_norm": 0.7566289625911207,
+      "learning_rate": 3.8411038056459e-06,
+      "loss": 0.3032,
+      "num_input_tokens_seen": 150463296,
+      "step": 2740,
+      "train_runtime": 15418.5837,
+      "train_tokens_per_second": 9758.568
+    },
+    {
+      "epoch": 1.1514261744966443,
+      "grad_norm": 0.8481639973499833,
+      "learning_rate": 3.825086412961275e-06,
+      "loss": 0.3108,
+      "num_input_tokens_seen": 150717640,
+      "step": 2745,
+      "train_runtime": 15445.7251,
+      "train_tokens_per_second": 9757.887
+    },
+    {
+      "epoch": 1.153523489932886,
+      "grad_norm": 0.7620364316160727,
+      "learning_rate": 3.809081772114861e-06,
+      "loss": 0.3017,
+      "num_input_tokens_seen": 150979832,
+      "step": 2750,
+      "train_runtime": 15474.1048,
+      "train_tokens_per_second": 9756.935
+    },
+    {
+      "epoch": 1.1556208053691275,
+      "grad_norm": 0.6704709049284874,
+      "learning_rate": 3.793090056811847e-06,
+      "loss": 0.2943,
+      "num_input_tokens_seen": 151247456,
+      "step": 2755,
+      "train_runtime": 15495.3179,
+      "train_tokens_per_second": 9760.849
+    },
+    {
+      "epoch": 1.157718120805369,
+      "grad_norm": 0.679466275188701,
+      "learning_rate": 3.777111440617142e-06,
+      "loss": 0.2863,
+      "num_input_tokens_seen": 151514496,
+      "step": 2760,
+      "train_runtime": 15521.1214,
+      "train_tokens_per_second": 9761.827
+    },
+    {
+      "epoch": 1.1598154362416107,
+      "grad_norm": 0.7414270839162481,
+      "learning_rate": 3.761146096953482e-06,
+      "loss": 0.3137,
+      "num_input_tokens_seen": 151774208,
+      "step": 2765,
+      "train_runtime": 15552.1703,
+      "train_tokens_per_second": 9759.037
+    },
+    {
+      "epoch": 1.1619127516778525,
+      "grad_norm": 0.7200599361238398,
+      "learning_rate": 3.745194199099549e-06,
+      "loss": 0.315,
+      "num_input_tokens_seen": 152063608,
+      "step": 2770,
+      "train_runtime": 15577.6752,
+      "train_tokens_per_second": 9761.637
+    },
+    {
+      "epoch": 1.164010067114094,
+      "grad_norm": 1.9120220105682426,
+      "learning_rate": 3.729255920188092e-06,
+      "loss": 0.2996,
+      "num_input_tokens_seen": 152333200,
+      "step": 2775,
+      "train_runtime": 15608.1794,
+      "train_tokens_per_second": 9759.831
+    },
+    {
+      "epoch": 1.1661073825503356,
+      "grad_norm": 0.7622473440278222,
+      "learning_rate": 3.7133314332040516e-06,
+      "loss": 0.3046,
+      "num_input_tokens_seen": 152617984,
+      "step": 2780,
+      "train_runtime": 15635.2215,
+      "train_tokens_per_second": 9761.165
+    },
+    {
+      "epoch": 1.1682046979865772,
+      "grad_norm": 0.6289498993120689,
+      "learning_rate": 3.6974209109826724e-06,
+      "loss": 0.2987,
+      "num_input_tokens_seen": 152919904,
+      "step": 2785,
+      "train_runtime": 15662.6535,
+      "train_tokens_per_second": 9763.346
+    },
+    {
+      "epoch": 1.1703020134228188,
+      "grad_norm": 0.6662519846422594,
+      "learning_rate": 3.6815245262076426e-06,
+      "loss": 0.3134,
+      "num_input_tokens_seen": 153204776,
+      "step": 2790,
+      "train_runtime": 15687.4305,
+      "train_tokens_per_second": 9766.085
+    },
+    {
+      "epoch": 1.1723993288590604,
+      "grad_norm": 0.6681675222839376,
+      "learning_rate": 3.6656424514092025e-06,
+      "loss": 0.2936,
+      "num_input_tokens_seen": 153476632,
+      "step": 2795,
+      "train_runtime": 15715.9488,
+      "train_tokens_per_second": 9765.661
+    },
+    {
+      "epoch": 1.174496644295302,
+      "grad_norm": 0.8307892187532807,
+      "learning_rate": 3.6497748589622826e-06,
+      "loss": 0.3042,
+      "num_input_tokens_seen": 153765256,
+      "step": 2800,
+      "train_runtime": 15740.4098,
+      "train_tokens_per_second": 9768.822
+    },
+    {
+      "epoch": 1.1765939597315436,
+      "grad_norm": 0.6569418879302218,
+      "learning_rate": 3.63392192108463e-06,
+      "loss": 0.299,
+      "num_input_tokens_seen": 154034536,
+      "step": 2805,
+      "train_runtime": 15770.0948,
+      "train_tokens_per_second": 9767.509
+    },
+    {
+      "epoch": 1.1786912751677852,
+      "grad_norm": 0.7115972350466219,
+      "learning_rate": 3.6180838098349417e-06,
+      "loss": 0.2976,
+      "num_input_tokens_seen": 154331000,
+      "step": 2810,
+      "train_runtime": 15794.0737,
+      "train_tokens_per_second": 9771.45
+    },
+    {
+      "epoch": 1.1807885906040267,
+      "grad_norm": 0.7451039553073958,
+      "learning_rate": 3.602260697110993e-06,
+      "loss": 0.3186,
+      "num_input_tokens_seen": 154632024,
+      "step": 2815,
+      "train_runtime": 15818.8522,
+      "train_tokens_per_second": 9775.173
+    },
+    {
+      "epoch": 1.1828859060402686,
+      "grad_norm": 0.7917556822784153,
+      "learning_rate": 3.5864527546477733e-06,
+      "loss": 0.3007,
+      "num_input_tokens_seen": 154911816,
+      "step": 2820,
+      "train_runtime": 15854.233,
+      "train_tokens_per_second": 9771.007
+    },
+    {
+      "epoch": 1.1849832214765101,
+      "grad_norm": 0.6549028259405517,
+      "learning_rate": 3.570660154015625e-06,
+      "loss": 0.3115,
+      "num_input_tokens_seen": 155175000,
+      "step": 2825,
+      "train_runtime": 15883.3897,
+      "train_tokens_per_second": 9769.64
+    },
+    {
+      "epoch": 1.1870805369127517,
+      "grad_norm": 0.7467595423085601,
+      "learning_rate": 3.5548830666183737e-06,
+      "loss": 0.3237,
+      "num_input_tokens_seen": 155443192,
+      "step": 2830,
+      "train_runtime": 15915.5814,
+      "train_tokens_per_second": 9766.73
+    },
+    {
+      "epoch": 1.1891778523489933,
+      "grad_norm": 0.7167487417919347,
+      "learning_rate": 3.5391216636914826e-06,
+      "loss": 0.3163,
+      "num_input_tokens_seen": 155733000,
+      "step": 2835,
+      "train_runtime": 15948.3787,
+      "train_tokens_per_second": 9764.817
+    },
+    {
+      "epoch": 1.191275167785235,
+      "grad_norm": 0.7313494703146947,
+      "learning_rate": 3.5233761163001755e-06,
+      "loss": 0.3143,
+      "num_input_tokens_seen": 156023960,
+      "step": 2840,
+      "train_runtime": 15974.9061,
+      "train_tokens_per_second": 9766.815
+    },
+    {
+      "epoch": 1.1933724832214765,
+      "grad_norm": 0.8326423793360515,
+      "learning_rate": 3.5076465953375954e-06,
+      "loss": 0.3271,
+      "num_input_tokens_seen": 156273480,
+      "step": 2845,
+      "train_runtime": 16006.5555,
+      "train_tokens_per_second": 9763.092
+    },
+    {
+      "epoch": 1.195469798657718,
+      "grad_norm": 0.7522688687370259,
+      "learning_rate": 3.4919332715229404e-06,
+      "loss": 0.3054,
+      "num_input_tokens_seen": 156559008,
+      "step": 2850,
+      "train_runtime": 16035.8073,
+      "train_tokens_per_second": 9763.089
+    },
+    {
+      "epoch": 1.1975671140939597,
+      "grad_norm": 0.7063068460413899,
+      "learning_rate": 3.4762363153996136e-06,
+      "loss": 0.3239,
+      "num_input_tokens_seen": 156838856,
+      "step": 2855,
+      "train_runtime": 16064.5828,
+      "train_tokens_per_second": 9763.021
+    },
+    {
+      "epoch": 1.1996644295302012,
+      "grad_norm": 0.7675351604290258,
+      "learning_rate": 3.460555897333375e-06,
+      "loss": 0.328,
+      "num_input_tokens_seen": 157131920,
+      "step": 2860,
+      "train_runtime": 16087.4748,
+      "train_tokens_per_second": 9767.345
+    },
+    {
+      "epoch": 1.201761744966443,
+      "grad_norm": 0.7118979482936135,
+      "learning_rate": 3.4448921875104898e-06,
+      "loss": 0.3047,
+      "num_input_tokens_seen": 157363024,
+      "step": 2865,
+      "train_runtime": 16112.7326,
+      "train_tokens_per_second": 9766.377
+    },
+    {
+      "epoch": 1.2038590604026846,
+      "grad_norm": 0.6558701208775267,
+      "learning_rate": 3.4292453559358795e-06,
+      "loss": 0.3131,
+      "num_input_tokens_seen": 157607424,
+      "step": 2870,
+      "train_runtime": 16138.0807,
+      "train_tokens_per_second": 9766.181
+    },
+    {
+      "epoch": 1.2059563758389262,
+      "grad_norm": 0.7459966393578854,
+      "learning_rate": 3.4136155724312774e-06,
+      "loss": 0.2945,
+      "num_input_tokens_seen": 157878944,
+      "step": 2875,
+      "train_runtime": 16166.3182,
+      "train_tokens_per_second": 9765.918
+    },
+    {
+      "epoch": 1.2080536912751678,
+      "grad_norm": 0.7107330733887335,
+      "learning_rate": 3.3980030066333947e-06,
+      "loss": 0.31,
+      "num_input_tokens_seen": 158155048,
+      "step": 2880,
+      "train_runtime": 16199.1399,
+      "train_tokens_per_second": 9763.176
+    },
+    {
+      "epoch": 1.2101510067114094,
+      "grad_norm": 0.6872330548436397,
+      "learning_rate": 3.3824078279920625e-06,
+      "loss": 0.3029,
+      "num_input_tokens_seen": 158438304,
+      "step": 2885,
+      "train_runtime": 16224.3411,
+      "train_tokens_per_second": 9765.469
+    },
+    {
+      "epoch": 1.212248322147651,
+      "grad_norm": 0.8857480991298772,
+      "learning_rate": 3.3668302057684053e-06,
+      "loss": 0.3284,
+      "num_input_tokens_seen": 158700368,
+      "step": 2890,
+      "train_runtime": 16257.1612,
+      "train_tokens_per_second": 9761.875
+    },
+    {
+      "epoch": 1.2143456375838926,
+      "grad_norm": 0.8942120331930039,
+      "learning_rate": 3.3512703090330023e-06,
+      "loss": 0.3135,
+      "num_input_tokens_seen": 158991528,
+      "step": 2895,
+      "train_runtime": 16289.4345,
+      "train_tokens_per_second": 9760.408
+    },
+    {
+      "epoch": 1.2164429530201342,
+      "grad_norm": 0.6844755509705026,
+      "learning_rate": 3.3357283066640472e-06,
+      "loss": 0.2896,
+      "num_input_tokens_seen": 159271952,
+      "step": 2900,
+      "train_runtime": 16317.7467,
+      "train_tokens_per_second": 9760.658
+    },
+    {
+      "epoch": 1.2185402684563758,
+      "grad_norm": 0.7199604269370643,
+      "learning_rate": 3.3202043673455194e-06,
+      "loss": 0.2813,
+      "num_input_tokens_seen": 159549168,
+      "step": 2905,
+      "train_runtime": 16347.1383,
+      "train_tokens_per_second": 9760.067
+    },
+    {
+      "epoch": 1.2206375838926173,
+      "grad_norm": 0.7421547160775688,
+      "learning_rate": 3.3046986595653528e-06,
+      "loss": 0.2939,
+      "num_input_tokens_seen": 159834544,
+      "step": 2910,
+      "train_runtime": 16380.7743,
+      "train_tokens_per_second": 9757.447
+    },
+    {
+      "epoch": 1.2227348993288591,
+      "grad_norm": 0.6543212847971928,
+      "learning_rate": 3.2892113516136055e-06,
+      "loss": 0.2968,
+      "num_input_tokens_seen": 160108616,
+      "step": 2915,
+      "train_runtime": 16412.8827,
+      "train_tokens_per_second": 9755.058
+    },
+    {
+      "epoch": 1.2248322147651007,
+      "grad_norm": 0.6847503293125466,
+      "learning_rate": 3.273742611580637e-06,
+      "loss": 0.302,
+      "num_input_tokens_seen": 160367928,
+      "step": 2920,
+      "train_runtime": 16442.9395,
+      "train_tokens_per_second": 9752.996
+    },
+    {
+      "epoch": 1.2269295302013423,
+      "grad_norm": 0.7740784771827695,
+      "learning_rate": 3.258292607355277e-06,
+      "loss": 0.2862,
+      "num_input_tokens_seen": 160660352,
+      "step": 2925,
+      "train_runtime": 16469.9082,
+      "train_tokens_per_second": 9754.781
+    },
+    {
+      "epoch": 1.229026845637584,
+      "grad_norm": 0.6956328708616384,
+      "learning_rate": 3.242861506623015e-06,
+      "loss": 0.3021,
+      "num_input_tokens_seen": 160940504,
+      "step": 2930,
+      "train_runtime": 16498.2678,
+      "train_tokens_per_second": 9754.994
+    },
+    {
+      "epoch": 1.2311241610738255,
+      "grad_norm": 0.6194378994158636,
+      "learning_rate": 3.227449476864164e-06,
+      "loss": 0.2791,
+      "num_input_tokens_seen": 161214096,
+      "step": 2935,
+      "train_runtime": 16529.5076,
+      "train_tokens_per_second": 9753.109
+    },
+    {
+      "epoch": 1.233221476510067,
+      "grad_norm": 0.6365025579098936,
+      "learning_rate": 3.2120566853520574e-06,
+      "loss": 0.2995,
+      "num_input_tokens_seen": 161491560,
+      "step": 2940,
+      "train_runtime": 16559.8304,
+      "train_tokens_per_second": 9752.006
+    },
+    {
+      "epoch": 1.2353187919463087,
+      "grad_norm": 0.6996945364273696,
+      "learning_rate": 3.1966832991512232e-06,
+      "loss": 0.2808,
+      "num_input_tokens_seen": 161763968,
+      "step": 2945,
+      "train_runtime": 16586.2215,
+      "train_tokens_per_second": 9752.913
+    },
+    {
+      "epoch": 1.2374161073825503,
+      "grad_norm": 0.703628756321934,
+      "learning_rate": 3.1813294851155807e-06,
+      "loss": 0.2951,
+      "num_input_tokens_seen": 162049552,
+      "step": 2950,
+      "train_runtime": 16614.3609,
+      "train_tokens_per_second": 9753.583
+    },
+    {
+      "epoch": 1.239513422818792,
+      "grad_norm": 0.6807889218031736,
+      "learning_rate": 3.1659954098866207e-06,
+      "loss": 0.2804,
+      "num_input_tokens_seen": 162334584,
+      "step": 2955,
+      "train_runtime": 16644.8359,
+      "train_tokens_per_second": 9752.85
+    },
+    {
+      "epoch": 1.2416107382550337,
+      "grad_norm": 0.6320274147785321,
+      "learning_rate": 3.150681239891601e-06,
+      "loss": 0.2814,
+      "num_input_tokens_seen": 162615336,
+      "step": 2960,
+      "train_runtime": 16673.071,
+      "train_tokens_per_second": 9753.172
+    },
+    {
+      "epoch": 1.2437080536912752,
+      "grad_norm": 0.7520747877694004,
+      "learning_rate": 3.135387141341738e-06,
+      "loss": 0.3228,
+      "num_input_tokens_seen": 162888392,
+      "step": 2965,
+      "train_runtime": 16699.3035,
+      "train_tokens_per_second": 9754.203
+    },
+    {
+      "epoch": 1.2458053691275168,
+      "grad_norm": 0.7123543895445158,
+      "learning_rate": 3.1201132802304056e-06,
+      "loss": 0.3037,
+      "num_input_tokens_seen": 163176384,
+      "step": 2970,
+      "train_runtime": 16731.05,
+      "train_tokens_per_second": 9752.908
+    },
+    {
+      "epoch": 1.2479026845637584,
+      "grad_norm": 0.954805444931472,
+      "learning_rate": 3.104859822331334e-06,
+      "loss": 0.2811,
+      "num_input_tokens_seen": 163475624,
+      "step": 2975,
+      "train_runtime": 16755.0356,
+      "train_tokens_per_second": 9756.806
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.630953471350359,
+      "learning_rate": 3.0896269331968066e-06,
+      "loss": 0.2768,
+      "num_input_tokens_seen": 163744288,
+      "step": 2980,
+      "train_runtime": 16785.2992,
+      "train_tokens_per_second": 9755.22
+    },
+    {
+      "epoch": 1.2520973154362416,
+      "grad_norm": 0.7420109597863583,
+      "learning_rate": 3.074414778155864e-06,
+      "loss": 0.3136,
+      "num_input_tokens_seen": 164022016,
+      "step": 2985,
+      "train_runtime": 16816.6559,
+      "train_tokens_per_second": 9753.545
+    },
+    {
+      "epoch": 1.2541946308724832,
+      "grad_norm": 0.735503370109786,
+      "learning_rate": 3.0592235223125135e-06,
+      "loss": 0.311,
+      "num_input_tokens_seen": 164287808,
+      "step": 2990,
+      "train_runtime": 16847.1623,
+      "train_tokens_per_second": 9751.661
+    },
+    {
+      "epoch": 1.2562919463087248,
+      "grad_norm": 0.633878564313551,
+      "learning_rate": 3.0440533305439312e-06,
+      "loss": 0.2986,
+      "num_input_tokens_seen": 164545552,
+      "step": 2995,
+      "train_runtime": 16877.5779,
+      "train_tokens_per_second": 9749.358
+    },
+    {
+      "epoch": 1.2583892617449663,
+      "grad_norm": 0.7715930571323115,
+      "learning_rate": 3.0289043674986794e-06,
+      "loss": 0.2984,
+      "num_input_tokens_seen": 164794064,
+      "step": 3000,
+      "train_runtime": 16906.3148,
+      "train_tokens_per_second": 9747.486
+    },
+    {
+      "epoch": 1.260486577181208,
+      "grad_norm": 0.6792970619840287,
+      "learning_rate": 3.013776797594915e-06,
+      "loss": 0.3261,
+      "num_input_tokens_seen": 165064848,
+      "step": 3005,
+      "train_runtime": 16933.6895,
+      "train_tokens_per_second": 9747.719
+    },
+    {
+      "epoch": 1.2625838926174497,
+      "grad_norm": 0.6425936079555361,
+      "learning_rate": 2.998670785018604e-06,
+      "loss": 0.3053,
+      "num_input_tokens_seen": 165317704,
+      "step": 3010,
+      "train_runtime": 16962.6464,
+      "train_tokens_per_second": 9745.985
+    },
+    {
+      "epoch": 1.2646812080536913,
+      "grad_norm": 0.7555088832151043,
+      "learning_rate": 2.983586493721741e-06,
+      "loss": 0.3046,
+      "num_input_tokens_seen": 165604096,
+      "step": 3015,
+      "train_runtime": 16996.0626,
+      "train_tokens_per_second": 9743.674
+    },
+    {
+      "epoch": 1.266778523489933,
+      "grad_norm": 0.6851578167394686,
+      "learning_rate": 2.9685240874205767e-06,
+      "loss": 0.3048,
+      "num_input_tokens_seen": 165871440,
+      "step": 3020,
+      "train_runtime": 17018.9201,
+      "train_tokens_per_second": 9746.296
+    },
+    {
+      "epoch": 1.2688758389261745,
+      "grad_norm": 0.7676265429853713,
+      "learning_rate": 2.9534837295938268e-06,
+      "loss": 0.3079,
+      "num_input_tokens_seen": 166130400,
+      "step": 3025,
+      "train_runtime": 17047.8027,
+      "train_tokens_per_second": 9744.974
+    },
+    {
+      "epoch": 1.270973154362416,
+      "grad_norm": 0.816206722108229,
+      "learning_rate": 2.9384655834809115e-06,
+      "loss": 0.3084,
+      "num_input_tokens_seen": 166379376,
+      "step": 3030,
+      "train_runtime": 17072.7652,
+      "train_tokens_per_second": 9745.309
+    },
+    {
+      "epoch": 1.2730704697986577,
+      "grad_norm": 0.6817052612299781,
+      "learning_rate": 2.923469812080173e-06,
+      "loss": 0.2882,
+      "num_input_tokens_seen": 166655248,
+      "step": 3035,
+      "train_runtime": 17101.9373,
+      "train_tokens_per_second": 9744.817
+    },
+    {
+      "epoch": 1.2751677852348993,
+      "grad_norm": 0.6476603820409995,
+      "learning_rate": 2.908496578147115e-06,
+      "loss": 0.3006,
+      "num_input_tokens_seen": 166931784,
+      "step": 3040,
+      "train_runtime": 17133.1302,
+      "train_tokens_per_second": 9743.216
+    },
+    {
+      "epoch": 1.277265100671141,
+      "grad_norm": 0.6637994169306357,
+      "learning_rate": 2.8935460441926298e-06,
+      "loss": 0.3113,
+      "num_input_tokens_seen": 167225728,
+      "step": 3045,
+      "train_runtime": 17160.199,
+      "train_tokens_per_second": 9744.976
+    },
+    {
+      "epoch": 1.2793624161073827,
+      "grad_norm": 0.7528698626610816,
+      "learning_rate": 2.878618372481238e-06,
+      "loss": 0.3015,
+      "num_input_tokens_seen": 167467040,
+      "step": 3050,
+      "train_runtime": 17189.7588,
+      "train_tokens_per_second": 9742.257
+    },
+    {
+      "epoch": 1.2814597315436242,
+      "grad_norm": 0.6999166016170123,
+      "learning_rate": 2.8637137250293267e-06,
+      "loss": 0.2936,
+      "num_input_tokens_seen": 167739728,
+      "step": 3055,
+      "train_runtime": 17220.5223,
+      "train_tokens_per_second": 9740.688
+    },
+    {
+      "epoch": 1.2835570469798658,
+      "grad_norm": 0.9286460762802081,
+      "learning_rate": 2.84883226360339e-06,
+      "loss": 0.3042,
+      "num_input_tokens_seen": 168016560,
+      "step": 3060,
+      "train_runtime": 17248.5276,
+      "train_tokens_per_second": 9740.922
+    },
+    {
+      "epoch": 1.2856543624161074,
+      "grad_norm": 0.7380106546975436,
+      "learning_rate": 2.8339741497182776e-06,
+      "loss": 0.3085,
+      "num_input_tokens_seen": 168288640,
+      "step": 3065,
+      "train_runtime": 17281.6269,
+      "train_tokens_per_second": 9738.009
+    },
+    {
+      "epoch": 1.287751677852349,
+      "grad_norm": 0.7487232134814832,
+      "learning_rate": 2.8191395446354332e-06,
+      "loss": 0.3117,
+      "num_input_tokens_seen": 168559912,
+      "step": 3070,
+      "train_runtime": 17306.2831,
+      "train_tokens_per_second": 9739.81
+    },
+    {
+      "epoch": 1.2898489932885906,
+      "grad_norm": 0.7091264960553184,
+      "learning_rate": 2.8043286093611544e-06,
+      "loss": 0.3051,
+      "num_input_tokens_seen": 168831096,
+      "step": 3075,
+      "train_runtime": 17341.6407,
+      "train_tokens_per_second": 9735.59
+    },
+    {
+      "epoch": 1.2919463087248322,
+      "grad_norm": 0.6420792506304508,
+      "learning_rate": 2.7895415046448372e-06,
+      "loss": 0.2779,
+      "num_input_tokens_seen": 169111592,
+      "step": 3080,
+      "train_runtime": 17368.6616,
+      "train_tokens_per_second": 9736.593
+    },
+    {
+      "epoch": 1.2940436241610738,
+      "grad_norm": 0.737250527030706,
+      "learning_rate": 2.774778390977234e-06,
+      "loss": 0.2842,
+      "num_input_tokens_seen": 169397112,
+      "step": 3085,
+      "train_runtime": 17397.9985,
+      "train_tokens_per_second": 9736.586
+    },
+    {
+      "epoch": 1.2961409395973154,
+      "grad_norm": 0.6726602477190619,
+      "learning_rate": 2.7600394285887156e-06,
+      "loss": 0.3051,
+      "num_input_tokens_seen": 169681568,
+      "step": 3090,
+      "train_runtime": 17429.2766,
+      "train_tokens_per_second": 9735.434
+    },
+    {
+      "epoch": 1.298238255033557,
+      "grad_norm": 0.7066776016188387,
+      "learning_rate": 2.7453247774475237e-06,
+      "loss": 0.327,
+      "num_input_tokens_seen": 169974624,
+      "step": 3095,
+      "train_runtime": 17457.5633,
+      "train_tokens_per_second": 9736.446
+    },
+    {
+      "epoch": 1.3003355704697985,
+      "grad_norm": 0.6416874880593003,
+      "learning_rate": 2.730634597258044e-06,
+      "loss": 0.2883,
+      "num_input_tokens_seen": 170257600,
+      "step": 3100,
+      "train_runtime": 17486.3363,
+      "train_tokens_per_second": 9736.608
+    },
+    {
+      "epoch": 1.3024328859060403,
+      "grad_norm": 0.6886924627425067,
+      "learning_rate": 2.715969047459066e-06,
+      "loss": 0.296,
+      "num_input_tokens_seen": 170523024,
+      "step": 3105,
+      "train_runtime": 17511.2042,
+      "train_tokens_per_second": 9737.938
+    },
+    {
+      "epoch": 1.304530201342282,
+      "grad_norm": 0.676411205462604,
+      "learning_rate": 2.701328287222054e-06,
+      "loss": 0.2922,
+      "num_input_tokens_seen": 170803176,
+      "step": 3110,
+      "train_runtime": 17543.2157,
+      "train_tokens_per_second": 9736.138
+    },
+    {
+      "epoch": 1.3066275167785235,
+      "grad_norm": 0.7663789445700779,
+      "learning_rate": 2.686712475449427e-06,
+      "loss": 0.2984,
+      "num_input_tokens_seen": 171076608,
+      "step": 3115,
+      "train_runtime": 17575.7612,
+      "train_tokens_per_second": 9733.667
+    },
+    {
+      "epoch": 1.308724832214765,
+      "grad_norm": 0.8255950124445262,
+      "learning_rate": 2.6721217707728187e-06,
+      "loss": 0.3064,
+      "num_input_tokens_seen": 171330520,
+      "step": 3120,
+      "train_runtime": 17604.2257,
+      "train_tokens_per_second": 9732.352
+    },
+    {
+      "epoch": 1.3108221476510067,
+      "grad_norm": 0.7446751117785054,
+      "learning_rate": 2.657556331551373e-06,
+      "loss": 0.3091,
+      "num_input_tokens_seen": 171605544,
+      "step": 3125,
+      "train_runtime": 17632.8895,
+      "train_tokens_per_second": 9732.128
+    },
+    {
+      "epoch": 1.3129194630872483,
+      "grad_norm": 0.7520475226312245,
+      "learning_rate": 2.6430163158700116e-06,
+      "loss": 0.2984,
+      "num_input_tokens_seen": 171864112,
+      "step": 3130,
+      "train_runtime": 17658.3742,
+      "train_tokens_per_second": 9732.726
+    },
+    {
+      "epoch": 1.3150167785234899,
+      "grad_norm": 0.8425163092696388,
+      "learning_rate": 2.628501881537726e-06,
+      "loss": 0.3,
+      "num_input_tokens_seen": 172117368,
+      "step": 3135,
+      "train_runtime": 17685.8275,
+      "train_tokens_per_second": 9731.937
+    },
+    {
+      "epoch": 1.3171140939597317,
+      "grad_norm": 0.7529439358925656,
+      "learning_rate": 2.614013186085861e-06,
+      "loss": 0.307,
+      "num_input_tokens_seen": 172408920,
+      "step": 3140,
+      "train_runtime": 17718.1182,
+      "train_tokens_per_second": 9730.656
+    },
+    {
+      "epoch": 1.3192114093959733,
+      "grad_norm": 0.7165764937505695,
+      "learning_rate": 2.5995503867664076e-06,
+      "loss": 0.3134,
+      "num_input_tokens_seen": 172667296,
+      "step": 3145,
+      "train_runtime": 17742.825,
+      "train_tokens_per_second": 9731.669
+    },
+    {
+      "epoch": 1.3213087248322148,
+      "grad_norm": 0.7659723057866115,
+      "learning_rate": 2.5851136405502935e-06,
+      "loss": 0.2977,
+      "num_input_tokens_seen": 172934880,
+      "step": 3150,
+      "train_runtime": 17770.1982,
+      "train_tokens_per_second": 9731.736
+    },
+    {
+      "epoch": 1.3234060402684564,
+      "grad_norm": 0.6696466646664914,
+      "learning_rate": 2.5707031041256815e-06,
+      "loss": 0.292,
+      "num_input_tokens_seen": 173209976,
+      "step": 3155,
+      "train_runtime": 17796.6647,
+      "train_tokens_per_second": 9732.721
+    },
+    {
+      "epoch": 1.325503355704698,
+      "grad_norm": 1.2031443582624795,
+      "learning_rate": 2.556318933896272e-06,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 173509272,
+      "step": 3160,
+      "train_runtime": 17816.6142,
+      "train_tokens_per_second": 9738.622
+    },
+    {
+      "epoch": 1.3276006711409396,
+      "grad_norm": 0.7243000338654723,
+      "learning_rate": 2.541961285979597e-06,
+      "loss": 0.3031,
+      "num_input_tokens_seen": 173767688,
+      "step": 3165,
+      "train_runtime": 17848.0485,
+      "train_tokens_per_second": 9735.949
+    },
+    {
+      "epoch": 1.3296979865771812,
+      "grad_norm": 0.7486945268911982,
+      "learning_rate": 2.5276303162053327e-06,
+      "loss": 0.3137,
+      "num_input_tokens_seen": 174016072,
+      "step": 3170,
+      "train_runtime": 17876.8356,
+      "train_tokens_per_second": 9734.165
+    },
+    {
+      "epoch": 1.3317953020134228,
+      "grad_norm": 0.6792331480781683,
+      "learning_rate": 2.5133261801136026e-06,
+      "loss": 0.309,
+      "num_input_tokens_seen": 174281568,
+      "step": 3175,
+      "train_runtime": 17902.8778,
+      "train_tokens_per_second": 9734.835
+    },
+    {
+      "epoch": 1.3338926174496644,
+      "grad_norm": 0.686231062630464,
+      "learning_rate": 2.4990490329533005e-06,
+      "loss": 0.311,
+      "num_input_tokens_seen": 174554032,
+      "step": 3180,
+      "train_runtime": 17931.7992,
+      "train_tokens_per_second": 9734.329
+    },
+    {
+      "epoch": 1.335989932885906,
+      "grad_norm": 0.6534141987756072,
+      "learning_rate": 2.4847990296803907e-06,
+      "loss": 0.3114,
+      "num_input_tokens_seen": 174841368,
+      "step": 3185,
+      "train_runtime": 17959.6861,
+      "train_tokens_per_second": 9735.213
+    },
+    {
+      "epoch": 1.3380872483221475,
+      "grad_norm": 0.6270847724586124,
+      "learning_rate": 2.4705763249562364e-06,
+      "loss": 0.3003,
+      "num_input_tokens_seen": 175136432,
+      "step": 3190,
+      "train_runtime": 17987.5673,
+      "train_tokens_per_second": 9736.527
+    },
+    {
+      "epoch": 1.3401845637583891,
+      "grad_norm": 0.7313908896343564,
+      "learning_rate": 2.4563810731459164e-06,
+      "loss": 0.2866,
+      "num_input_tokens_seen": 175393440,
+      "step": 3195,
+      "train_runtime": 18019.282,
+      "train_tokens_per_second": 9733.653
+    },
+    {
+      "epoch": 1.342281879194631,
+      "grad_norm": 0.6530220884576664,
+      "learning_rate": 2.442213428316551e-06,
+      "loss": 0.2984,
+      "num_input_tokens_seen": 175683688,
+      "step": 3200,
+      "train_runtime": 18045.2582,
+      "train_tokens_per_second": 9735.726
+    },
+    {
+      "epoch": 1.3443791946308725,
+      "grad_norm": 0.7204752881745307,
+      "learning_rate": 2.4280735442356355e-06,
+      "loss": 0.3181,
+      "num_input_tokens_seen": 175945592,
+      "step": 3205,
+      "train_runtime": 18067.7483,
+      "train_tokens_per_second": 9738.103
+    },
+    {
+      "epoch": 1.346476510067114,
+      "grad_norm": 0.911653201816413,
+      "learning_rate": 2.413961574369359e-06,
+      "loss": 0.286,
+      "num_input_tokens_seen": 176181320,
+      "step": 3210,
+      "train_runtime": 18092.4373,
+      "train_tokens_per_second": 9737.843
+    },
+    {
+      "epoch": 1.3485738255033557,
+      "grad_norm": 0.7414681631406747,
+      "learning_rate": 2.3998776718809495e-06,
+      "loss": 0.3262,
+      "num_input_tokens_seen": 176442272,
+      "step": 3215,
+      "train_runtime": 18118.9387,
+      "train_tokens_per_second": 9738.003
+    },
+    {
+      "epoch": 1.3506711409395973,
+      "grad_norm": 0.6785037223306885,
+      "learning_rate": 2.3858219896290064e-06,
+      "loss": 0.3019,
+      "num_input_tokens_seen": 176715760,
+      "step": 3220,
+      "train_runtime": 18151.4272,
+      "train_tokens_per_second": 9735.64
+    },
+    {
+      "epoch": 1.3527684563758389,
+      "grad_norm": 0.7690883016305601,
+      "learning_rate": 2.3717946801658415e-06,
+      "loss": 0.2832,
+      "num_input_tokens_seen": 177003536,
+      "step": 3225,
+      "train_runtime": 18178.2997,
+      "train_tokens_per_second": 9737.079
+    },
+    {
+      "epoch": 1.3548657718120805,
+      "grad_norm": 0.6527650382487596,
+      "learning_rate": 2.3577958957358304e-06,
+      "loss": 0.2933,
+      "num_input_tokens_seen": 177286504,
+      "step": 3230,
+      "train_runtime": 18213.1466,
+      "train_tokens_per_second": 9733.985
+    },
+    {
+      "epoch": 1.3569630872483223,
+      "grad_norm": 0.7648546958130339,
+      "learning_rate": 2.3438257882737486e-06,
+      "loss": 0.3324,
+      "num_input_tokens_seen": 177577840,
+      "step": 3235,
+      "train_runtime": 18245.9926,
+      "train_tokens_per_second": 9732.43
+    },
+    {
+      "epoch": 1.3590604026845639,
+      "grad_norm": 0.7815411915066078,
+      "learning_rate": 2.3298845094031296e-06,
+      "loss": 0.3076,
+      "num_input_tokens_seen": 177849368,
+      "step": 3240,
+      "train_runtime": 18276.1301,
+      "train_tokens_per_second": 9731.238
+    },
+    {
+      "epoch": 1.3611577181208054,
+      "grad_norm": 0.694739087689447,
+      "learning_rate": 2.3159722104346187e-06,
+      "loss": 0.3098,
+      "num_input_tokens_seen": 178119872,
+      "step": 3245,
+      "train_runtime": 18300.1459,
+      "train_tokens_per_second": 9733.249
+    },
+    {
+      "epoch": 1.363255033557047,
+      "grad_norm": 0.6639422506624533,
+      "learning_rate": 2.3020890423643273e-06,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 178415000,
+      "step": 3250,
+      "train_runtime": 18330.7725,
+      "train_tokens_per_second": 9733.087
+    },
+    {
+      "epoch": 1.3653523489932886,
+      "grad_norm": 0.6769890468112392,
+      "learning_rate": 2.288235155872203e-06,
+      "loss": 0.2813,
+      "num_input_tokens_seen": 178717048,
+      "step": 3255,
+      "train_runtime": 18360.4777,
+      "train_tokens_per_second": 9733.791
+    },
+    {
+      "epoch": 1.3674496644295302,
+      "grad_norm": 0.8249275819993918,
+      "learning_rate": 2.2744107013203825e-06,
+      "loss": 0.306,
+      "num_input_tokens_seen": 178998536,
+      "step": 3260,
+      "train_runtime": 18387.5436,
+      "train_tokens_per_second": 9734.772
+    },
+    {
+      "epoch": 1.3695469798657718,
+      "grad_norm": 0.7599266496920702,
+      "learning_rate": 2.2606158287515662e-06,
+      "loss": 0.3293,
+      "num_input_tokens_seen": 179262016,
+      "step": 3265,
+      "train_runtime": 18417.523,
+      "train_tokens_per_second": 9733.232
+    },
+    {
+      "epoch": 1.3716442953020134,
+      "grad_norm": 0.6290767305778602,
+      "learning_rate": 2.246850687887389e-06,
+      "loss": 0.2754,
+      "num_input_tokens_seen": 179532592,
+      "step": 3270,
+      "train_runtime": 18440.5452,
+      "train_tokens_per_second": 9735.753
+    },
+    {
+      "epoch": 1.373741610738255,
+      "grad_norm": 0.8047047387699678,
+      "learning_rate": 2.2331154281267952e-06,
+      "loss": 0.3185,
+      "num_input_tokens_seen": 179763872,
+      "step": 3275,
+      "train_runtime": 18471.831,
+      "train_tokens_per_second": 9731.784
+    },
+    {
+      "epoch": 1.3758389261744965,
+      "grad_norm": 0.6280696431434224,
+      "learning_rate": 2.2194101985444146e-06,
+      "loss": 0.2905,
+      "num_input_tokens_seen": 180046152,
+      "step": 3280,
+      "train_runtime": 18498.0002,
+      "train_tokens_per_second": 9733.277
+    },
+    {
+      "epoch": 1.3779362416107381,
+      "grad_norm": 0.7059851281841262,
+      "learning_rate": 2.205735147888952e-06,
+      "loss": 0.289,
+      "num_input_tokens_seen": 180287088,
+      "step": 3285,
+      "train_runtime": 18523.3123,
+      "train_tokens_per_second": 9732.983
+    },
+    {
+      "epoch": 1.38003355704698,
+      "grad_norm": 0.6108808857838776,
+      "learning_rate": 2.192090424581561e-06,
+      "loss": 0.2889,
+      "num_input_tokens_seen": 180556520,
+      "step": 3290,
+      "train_runtime": 18547.1437,
+      "train_tokens_per_second": 9735.004
+    },
+    {
+      "epoch": 1.3821308724832215,
+      "grad_norm": 0.7361959199574103,
+      "learning_rate": 2.178476176714243e-06,
+      "loss": 0.2887,
+      "num_input_tokens_seen": 180837552,
+      "step": 3295,
+      "train_runtime": 18578.3784,
+      "train_tokens_per_second": 9733.764
+    },
+    {
+      "epoch": 1.3842281879194631,
+      "grad_norm": 0.8043894261381194,
+      "learning_rate": 2.1648925520482384e-06,
+      "loss": 0.303,
+      "num_input_tokens_seen": 181124048,
+      "step": 3300,
+      "train_runtime": 18608.7753,
+      "train_tokens_per_second": 9733.26
+    },
+    {
+      "epoch": 1.3863255033557047,
+      "grad_norm": 0.6596807549096665,
+      "learning_rate": 2.151339698012417e-06,
+      "loss": 0.2994,
+      "num_input_tokens_seen": 181395600,
+      "step": 3305,
+      "train_runtime": 18641.1202,
+      "train_tokens_per_second": 9730.939
+    },
+    {
+      "epoch": 1.3884228187919463,
+      "grad_norm": 0.8091125680542506,
+      "learning_rate": 2.1378177617016834e-06,
+      "loss": 0.3078,
+      "num_input_tokens_seen": 181663816,
+      "step": 3310,
+      "train_runtime": 18670.6447,
+      "train_tokens_per_second": 9729.917
+    },
+    {
+      "epoch": 1.3905201342281879,
+      "grad_norm": 0.7290723161103462,
+      "learning_rate": 2.124326889875378e-06,
+      "loss": 0.3148,
+      "num_input_tokens_seen": 181942136,
+      "step": 3315,
+      "train_runtime": 18697.3238,
+      "train_tokens_per_second": 9730.919
+    },
+    {
+      "epoch": 1.3926174496644295,
+      "grad_norm": 0.7008060128374883,
+      "learning_rate": 2.110867228955687e-06,
+      "loss": 0.3139,
+      "num_input_tokens_seen": 182212976,
+      "step": 3320,
+      "train_runtime": 18726.2676,
+      "train_tokens_per_second": 9730.341
+    },
+    {
+      "epoch": 1.3947147651006713,
+      "grad_norm": 0.7997771399269628,
+      "learning_rate": 2.09743892502605e-06,
+      "loss": 0.2867,
+      "num_input_tokens_seen": 182476616,
+      "step": 3325,
+      "train_runtime": 18760.3987,
+      "train_tokens_per_second": 9726.692
+    },
+    {
+      "epoch": 1.3968120805369129,
+      "grad_norm": 0.7098493855484669,
+      "learning_rate": 2.0840421238295742e-06,
+      "loss": 0.3007,
+      "num_input_tokens_seen": 182773464,
+      "step": 3330,
+      "train_runtime": 18785.8178,
+      "train_tokens_per_second": 9729.332
+    },
+    {
+      "epoch": 1.3989093959731544,
+      "grad_norm": 0.7121174841896368,
+      "learning_rate": 2.07067697076746e-06,
+      "loss": 0.3271,
+      "num_input_tokens_seen": 183043880,
+      "step": 3335,
+      "train_runtime": 18810.8646,
+      "train_tokens_per_second": 9730.753
+    },
+    {
+      "epoch": 1.401006711409396,
+      "grad_norm": 0.6614456221441788,
+      "learning_rate": 2.05734361089741e-06,
+      "loss": 0.3166,
+      "num_input_tokens_seen": 183338424,
+      "step": 3340,
+      "train_runtime": 18840.5297,
+      "train_tokens_per_second": 9731.065
+    },
+    {
+      "epoch": 1.4031040268456376,
+      "grad_norm": 0.6095180557559895,
+      "learning_rate": 2.044042188932068e-06,
+      "loss": 0.2919,
+      "num_input_tokens_seen": 183650952,
+      "step": 3345,
+      "train_runtime": 18868.6556,
+      "train_tokens_per_second": 9733.123
+    },
+    {
+      "epoch": 1.4052013422818792,
+      "grad_norm": 0.9777212491517744,
+      "learning_rate": 2.0307728492374375e-06,
+      "loss": 0.3033,
+      "num_input_tokens_seen": 183935440,
+      "step": 3350,
+      "train_runtime": 18898.6581,
+      "train_tokens_per_second": 9732.725
+    },
+    {
+      "epoch": 1.4072986577181208,
+      "grad_norm": 0.6946077480882675,
+      "learning_rate": 2.0175357358313204e-06,
+      "loss": 0.2989,
+      "num_input_tokens_seen": 184229448,
+      "step": 3355,
+      "train_runtime": 18924.1738,
+      "train_tokens_per_second": 9735.138
+    },
+    {
+      "epoch": 1.4093959731543624,
+      "grad_norm": 0.6376317310013874,
+      "learning_rate": 2.0043309923817537e-06,
+      "loss": 0.3129,
+      "num_input_tokens_seen": 184528016,
+      "step": 3360,
+      "train_runtime": 18959.2045,
+      "train_tokens_per_second": 9732.899
+    },
+    {
+      "epoch": 1.411493288590604,
+      "grad_norm": 0.8211281939585227,
+      "learning_rate": 1.9911587622054506e-06,
+      "loss": 0.3049,
+      "num_input_tokens_seen": 184799328,
+      "step": 3365,
+      "train_runtime": 18989.5665,
+      "train_tokens_per_second": 9731.624
+    },
+    {
+      "epoch": 1.4135906040268456,
+      "grad_norm": 0.6783002299382792,
+      "learning_rate": 1.978019188266241e-06,
+      "loss": 0.3079,
+      "num_input_tokens_seen": 185074856,
+      "step": 3370,
+      "train_runtime": 19014.8385,
+      "train_tokens_per_second": 9733.181
+    },
+    {
+      "epoch": 1.4156879194630871,
+      "grad_norm": 0.748008714535767,
+      "learning_rate": 1.9649124131735253e-06,
+      "loss": 0.2799,
+      "num_input_tokens_seen": 185344416,
+      "step": 3375,
+      "train_runtime": 19046.6633,
+      "train_tokens_per_second": 9731.07
+    },
+    {
+      "epoch": 1.4177852348993287,
+      "grad_norm": 0.7520669722605393,
+      "learning_rate": 1.9518385791807225e-06,
+      "loss": 0.2949,
+      "num_input_tokens_seen": 185611192,
+      "step": 3380,
+      "train_runtime": 19073.6229,
+      "train_tokens_per_second": 9731.302
+    },
+    {
+      "epoch": 1.4198825503355705,
+      "grad_norm": 0.6447899194681384,
+      "learning_rate": 1.938797828183727e-06,
+      "loss": 0.3018,
+      "num_input_tokens_seen": 185897432,
+      "step": 3385,
+      "train_runtime": 19100.256,
+      "train_tokens_per_second": 9732.719
+    },
+    {
+      "epoch": 1.4219798657718121,
+      "grad_norm": 0.6969539262200865,
+      "learning_rate": 1.925790301719374e-06,
+      "loss": 0.2901,
+      "num_input_tokens_seen": 186194200,
+      "step": 3390,
+      "train_runtime": 19135.5964,
+      "train_tokens_per_second": 9730.253
+    },
+    {
+      "epoch": 1.4240771812080537,
+      "grad_norm": 0.7493897403484177,
+      "learning_rate": 1.912816140963892e-06,
+      "loss": 0.2959,
+      "num_input_tokens_seen": 186479840,
+      "step": 3395,
+      "train_runtime": 19166.2843,
+      "train_tokens_per_second": 9729.577
+    },
+    {
+      "epoch": 1.4261744966442953,
+      "grad_norm": 0.6731273283564022,
+      "learning_rate": 1.899875486731384e-06,
+      "loss": 0.2843,
+      "num_input_tokens_seen": 186756088,
+      "step": 3400,
+      "train_runtime": 19199.3767,
+      "train_tokens_per_second": 9727.195
+    },
+    {
+      "epoch": 1.4282718120805369,
+      "grad_norm": 0.7562938315574533,
+      "learning_rate": 1.886968479472287e-06,
+      "loss": 0.3195,
+      "num_input_tokens_seen": 187049568,
+      "step": 3405,
+      "train_runtime": 19228.3094,
+      "train_tokens_per_second": 9727.822
+    },
+    {
+      "epoch": 1.4303691275167785,
+      "grad_norm": 0.7588695122660573,
+      "learning_rate": 1.8740952592718553e-06,
+      "loss": 0.2836,
+      "num_input_tokens_seen": 187328240,
+      "step": 3410,
+      "train_runtime": 19255.3974,
+      "train_tokens_per_second": 9728.609
+    },
+    {
+      "epoch": 1.43246644295302,
+      "grad_norm": 0.7538513275700046,
+      "learning_rate": 1.8612559658486373e-06,
+      "loss": 0.3021,
+      "num_input_tokens_seen": 187603168,
+      "step": 3415,
+      "train_runtime": 19290.6739,
+      "train_tokens_per_second": 9725.071
+    },
+    {
+      "epoch": 1.4345637583892619,
+      "grad_norm": 0.7290088734949237,
+      "learning_rate": 1.8484507385529605e-06,
+      "loss": 0.3025,
+      "num_input_tokens_seen": 187874864,
+      "step": 3420,
+      "train_runtime": 19318.8921,
+      "train_tokens_per_second": 9724.93
+    },
+    {
+      "epoch": 1.4366610738255035,
+      "grad_norm": 0.8058379115595437,
+      "learning_rate": 1.8356797163654172e-06,
+      "loss": 0.3143,
+      "num_input_tokens_seen": 188147240,
+      "step": 3425,
+      "train_runtime": 19344.7529,
+      "train_tokens_per_second": 9726.009
+    },
+    {
+      "epoch": 1.438758389261745,
+      "grad_norm": 0.7730717991421587,
+      "learning_rate": 1.8229430378953578e-06,
+      "loss": 0.2753,
+      "num_input_tokens_seen": 188410520,
+      "step": 3430,
+      "train_runtime": 19371.5956,
+      "train_tokens_per_second": 9726.123
+    },
+    {
+      "epoch": 1.4408557046979866,
+      "grad_norm": 0.8060154630187426,
+      "learning_rate": 1.8102408413793838e-06,
+      "loss": 0.2992,
+      "num_input_tokens_seen": 188665784,
+      "step": 3435,
+      "train_runtime": 19399.2084,
+      "train_tokens_per_second": 9725.437
+    },
+    {
+      "epoch": 1.4429530201342282,
+      "grad_norm": 0.7716395759866018,
+      "learning_rate": 1.7975732646798534e-06,
+      "loss": 0.2938,
+      "num_input_tokens_seen": 188935928,
+      "step": 3440,
+      "train_runtime": 19435.8003,
+      "train_tokens_per_second": 9721.026
+    },
+    {
+      "epoch": 1.4450503355704698,
+      "grad_norm": 0.7570869475508158,
+      "learning_rate": 1.7849404452833763e-06,
+      "loss": 0.3024,
+      "num_input_tokens_seen": 189193448,
+      "step": 3445,
+      "train_runtime": 19462.3982,
+      "train_tokens_per_second": 9720.973
+    },
+    {
+      "epoch": 1.4471476510067114,
+      "grad_norm": 0.8151248513142179,
+      "learning_rate": 1.7723425202993315e-06,
+      "loss": 0.2955,
+      "num_input_tokens_seen": 189459904,
+      "step": 3450,
+      "train_runtime": 19486.9468,
+      "train_tokens_per_second": 9722.401
+    },
+    {
+      "epoch": 1.449244966442953,
+      "grad_norm": 0.7280016161064552,
+      "learning_rate": 1.75977962645837e-06,
+      "loss": 0.3007,
+      "num_input_tokens_seen": 189732856,
+      "step": 3455,
+      "train_runtime": 19514.2643,
+      "train_tokens_per_second": 9722.778
+    },
+    {
+      "epoch": 1.4513422818791946,
+      "grad_norm": 0.7197112973347557,
+      "learning_rate": 1.747251900110936e-06,
+      "loss": 0.2919,
+      "num_input_tokens_seen": 189986368,
+      "step": 3460,
+      "train_runtime": 19543.3697,
+      "train_tokens_per_second": 9721.27
+    },
+    {
+      "epoch": 1.4534395973154361,
+      "grad_norm": 0.7689303309117073,
+      "learning_rate": 1.7347594772257863e-06,
+      "loss": 0.297,
+      "num_input_tokens_seen": 190250816,
+      "step": 3465,
+      "train_runtime": 19576.1288,
+      "train_tokens_per_second": 9718.511
+    },
+    {
+      "epoch": 1.4555369127516777,
+      "grad_norm": 0.7429420085810566,
+      "learning_rate": 1.7223024933885142e-06,
+      "loss": 0.2895,
+      "num_input_tokens_seen": 190527928,
+      "step": 3470,
+      "train_runtime": 19602.1827,
+      "train_tokens_per_second": 9719.73
+    },
+    {
+      "epoch": 1.4576342281879195,
+      "grad_norm": 0.7301159507052207,
+      "learning_rate": 1.7098810838000768e-06,
+      "loss": 0.2978,
+      "num_input_tokens_seen": 190798232,
+      "step": 3475,
+      "train_runtime": 19629.0015,
+      "train_tokens_per_second": 9720.221
+    },
+    {
+      "epoch": 1.4597315436241611,
+      "grad_norm": 0.8714396175167076,
+      "learning_rate": 1.6974953832753282e-06,
+      "loss": 0.3088,
+      "num_input_tokens_seen": 191083456,
+      "step": 3480,
+      "train_runtime": 19652.2522,
+      "train_tokens_per_second": 9723.234
+    },
+    {
+      "epoch": 1.4618288590604027,
+      "grad_norm": 0.7129431801742107,
+      "learning_rate": 1.6851455262415623e-06,
+      "loss": 0.3105,
+      "num_input_tokens_seen": 191363136,
+      "step": 3485,
+      "train_runtime": 19681.9451,
+      "train_tokens_per_second": 9722.776
+    },
+    {
+      "epoch": 1.4639261744966443,
+      "grad_norm": 0.720061570570095,
+      "learning_rate": 1.6728316467370414e-06,
+      "loss": 0.283,
+      "num_input_tokens_seen": 191627640,
+      "step": 3490,
+      "train_runtime": 19711.9529,
+      "train_tokens_per_second": 9721.393
+    },
+    {
+      "epoch": 1.466023489932886,
+      "grad_norm": 1.570378568621867,
+      "learning_rate": 1.6605538784095515e-06,
+      "loss": 0.3086,
+      "num_input_tokens_seen": 191896264,
+      "step": 3495,
+      "train_runtime": 19741.0319,
+      "train_tokens_per_second": 9720.68
+    },
+    {
+      "epoch": 1.4681208053691275,
+      "grad_norm": 0.677099207211974,
+      "learning_rate": 1.6483123545149466e-06,
+      "loss": 0.2916,
+      "num_input_tokens_seen": 192170944,
+      "step": 3500,
+      "train_runtime": 19769.0836,
+      "train_tokens_per_second": 9720.782
+    },
+    {
+      "epoch": 1.470218120805369,
+      "grad_norm": 0.6616950640497966,
+      "learning_rate": 1.6361072079157092e-06,
+      "loss": 0.2849,
+      "num_input_tokens_seen": 192455464,
+      "step": 3505,
+      "train_runtime": 19797.9274,
+      "train_tokens_per_second": 9720.99
+    },
+    {
+      "epoch": 1.4723154362416109,
+      "grad_norm": 0.6892232378382448,
+      "learning_rate": 1.6239385710794986e-06,
+      "loss": 0.3121,
+      "num_input_tokens_seen": 192750728,
+      "step": 3510,
+      "train_runtime": 19832.7885,
+      "train_tokens_per_second": 9718.791
+    },
+    {
+      "epoch": 1.4744127516778525,
+      "grad_norm": 0.6344362867846446,
+      "learning_rate": 1.6118065760777202e-06,
+      "loss": 0.2824,
+      "num_input_tokens_seen": 193035544,
+      "step": 3515,
+      "train_runtime": 19859.9502,
+      "train_tokens_per_second": 9719.84
+    },
+    {
+      "epoch": 1.476510067114094,
+      "grad_norm": 0.7163536770373257,
+      "learning_rate": 1.5997113545840892e-06,
+      "loss": 0.2911,
+      "num_input_tokens_seen": 193284160,
+      "step": 3520,
+      "train_runtime": 19888.5529,
+      "train_tokens_per_second": 9718.362
+    },
+    {
+      "epoch": 1.4786073825503356,
+      "grad_norm": 0.6849025809495934,
+      "learning_rate": 1.587653037873203e-06,
+      "loss": 0.2746,
+      "num_input_tokens_seen": 193537744,
+      "step": 3525,
+      "train_runtime": 19916.3934,
+      "train_tokens_per_second": 9717.51
+    },
+    {
+      "epoch": 1.4807046979865772,
+      "grad_norm": 0.6817248996490238,
+      "learning_rate": 1.5756317568191126e-06,
+      "loss": 0.2898,
+      "num_input_tokens_seen": 193799296,
+      "step": 3530,
+      "train_runtime": 19935.8686,
+      "train_tokens_per_second": 9721.136
+    },
+    {
+      "epoch": 1.4828020134228188,
+      "grad_norm": 0.7429267822941088,
+      "learning_rate": 1.5636476418939123e-06,
+      "loss": 0.2969,
+      "num_input_tokens_seen": 194068016,
+      "step": 3535,
+      "train_runtime": 19962.0446,
+      "train_tokens_per_second": 9721.851
+    },
+    {
+      "epoch": 1.4848993288590604,
+      "grad_norm": 0.7288103787800937,
+      "learning_rate": 1.5517008231663094e-06,
+      "loss": 0.2977,
+      "num_input_tokens_seen": 194350432,
+      "step": 3540,
+      "train_runtime": 19994.6766,
+      "train_tokens_per_second": 9720.109
+    },
+    {
+      "epoch": 1.486996644295302,
+      "grad_norm": 0.6995885842035907,
+      "learning_rate": 1.5397914303002227e-06,
+      "loss": 0.2919,
+      "num_input_tokens_seen": 194613944,
+      "step": 3545,
+      "train_runtime": 20021.4375,
+      "train_tokens_per_second": 9720.278
+    },
+    {
+      "epoch": 1.4890939597315436,
+      "grad_norm": 0.6776871497725921,
+      "learning_rate": 1.5279195925533708e-06,
+      "loss": 0.2955,
+      "num_input_tokens_seen": 194895960,
+      "step": 3550,
+      "train_runtime": 20048.9244,
+      "train_tokens_per_second": 9721.018
+    },
+    {
+      "epoch": 1.4911912751677852,
+      "grad_norm": 0.7365589438857084,
+      "learning_rate": 1.5160854387758738e-06,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 195178792,
+      "step": 3555,
+      "train_runtime": 20074.4352,
+      "train_tokens_per_second": 9722.754
+    },
+    {
+      "epoch": 1.4932885906040267,
+      "grad_norm": 0.6995216027988713,
+      "learning_rate": 1.5042890974088487e-06,
+      "loss": 0.2765,
+      "num_input_tokens_seen": 195453800,
+      "step": 3560,
+      "train_runtime": 20101.2827,
+      "train_tokens_per_second": 9723.449
+    },
+    {
+      "epoch": 1.4953859060402683,
+      "grad_norm": 0.8923869115714458,
+      "learning_rate": 1.4925306964830183e-06,
+      "loss": 0.3313,
+      "num_input_tokens_seen": 195725768,
+      "step": 3565,
+      "train_runtime": 20131.246,
+      "train_tokens_per_second": 9722.487
+    },
+    {
+      "epoch": 1.4974832214765101,
+      "grad_norm": 0.6797035307896964,
+      "learning_rate": 1.4808103636173226e-06,
+      "loss": 0.3069,
+      "num_input_tokens_seen": 195991904,
+      "step": 3570,
+      "train_runtime": 20161.0666,
+      "train_tokens_per_second": 9721.306
+    },
+    {
+      "epoch": 1.4995805369127517,
+      "grad_norm": 0.6878095951564138,
+      "learning_rate": 1.469128226017531e-06,
+      "loss": 0.2634,
+      "num_input_tokens_seen": 196278040,
+      "step": 3575,
+      "train_runtime": 20182.852,
+      "train_tokens_per_second": 9724.99
+    },
+    {
+      "epoch": 1.5016778523489933,
+      "grad_norm": 0.7692911777944031,
+      "learning_rate": 1.4574844104748676e-06,
+      "loss": 0.3085,
+      "num_input_tokens_seen": 196540896,
+      "step": 3580,
+      "train_runtime": 20209.7112,
+      "train_tokens_per_second": 9725.072
+    },
+    {
+      "epoch": 1.503775167785235,
+      "grad_norm": 0.7760683798223811,
+      "learning_rate": 1.4458790433646264e-06,
+      "loss": 0.3004,
+      "num_input_tokens_seen": 196832760,
+      "step": 3585,
+      "train_runtime": 20239.9874,
+      "train_tokens_per_second": 9724.945
+    },
+    {
+      "epoch": 1.5058724832214765,
+      "grad_norm": 0.6630390420426049,
+      "learning_rate": 1.4343122506448065e-06,
+      "loss": 0.2748,
+      "num_input_tokens_seen": 197093168,
+      "step": 3590,
+      "train_runtime": 20261.4289,
+      "train_tokens_per_second": 9727.506
+    },
+    {
+      "epoch": 1.507969798657718,
+      "grad_norm": 0.6786229937427486,
+      "learning_rate": 1.4227841578547413e-06,
+      "loss": 0.304,
+      "num_input_tokens_seen": 197396800,
+      "step": 3595,
+      "train_runtime": 20286.9423,
+      "train_tokens_per_second": 9730.239
+    },
+    {
+      "epoch": 1.5100671140939599,
+      "grad_norm": 0.706807273898724,
+      "learning_rate": 1.4112948901137392e-06,
+      "loss": 0.2839,
+      "num_input_tokens_seen": 197688248,
+      "step": 3600,
+      "train_runtime": 20317.2906,
+      "train_tokens_per_second": 9730.05
+    },
+    {
+      "epoch": 1.5121644295302015,
+      "grad_norm": 0.7077755953706824,
+      "learning_rate": 1.3998445721197197e-06,
+      "loss": 0.3272,
+      "num_input_tokens_seen": 197956712,
+      "step": 3605,
+      "train_runtime": 20346.056,
+      "train_tokens_per_second": 9729.488
+    },
+    {
+      "epoch": 1.514261744966443,
+      "grad_norm": 0.7335697736763683,
+      "learning_rate": 1.3884333281478707e-06,
+      "loss": 0.2725,
+      "num_input_tokens_seen": 198227880,
+      "step": 3610,
+      "train_runtime": 20367.7307,
+      "train_tokens_per_second": 9732.448
+    },
+    {
+      "epoch": 1.5163590604026846,
+      "grad_norm": 0.6919020664837792,
+      "learning_rate": 1.3770612820492879e-06,
+      "loss": 0.2952,
+      "num_input_tokens_seen": 198521216,
+      "step": 3615,
+      "train_runtime": 20396.5484,
+      "train_tokens_per_second": 9733.079
+    },
+    {
+      "epoch": 1.5184563758389262,
+      "grad_norm": 0.6525929103846213,
+      "learning_rate": 1.365728557249636e-06,
+      "loss": 0.2972,
+      "num_input_tokens_seen": 198814072,
+      "step": 3620,
+      "train_runtime": 20426.2325,
+      "train_tokens_per_second": 9733.272
+    },
+    {
+      "epoch": 1.5205536912751678,
+      "grad_norm": 0.734371212329942,
+      "learning_rate": 1.354435276747813e-06,
+      "loss": 0.3125,
+      "num_input_tokens_seen": 199085608,
+      "step": 3625,
+      "train_runtime": 20453.1575,
+      "train_tokens_per_second": 9733.735
+    },
+    {
+      "epoch": 1.5226510067114094,
+      "grad_norm": 0.7316869095203868,
+      "learning_rate": 1.3431815631146068e-06,
+      "loss": 0.2869,
+      "num_input_tokens_seen": 199363880,
+      "step": 3630,
+      "train_runtime": 20480.8304,
+      "train_tokens_per_second": 9734.17
+    },
+    {
+      "epoch": 1.524748322147651,
+      "grad_norm": 0.6895799970964066,
+      "learning_rate": 1.3319675384913705e-06,
+      "loss": 0.2999,
+      "num_input_tokens_seen": 199640608,
+      "step": 3635,
+      "train_runtime": 20509.8875,
+      "train_tokens_per_second": 9733.871
+    },
+    {
+      "epoch": 1.5268456375838926,
+      "grad_norm": 0.7055953662640705,
+      "learning_rate": 1.3207933245886956e-06,
+      "loss": 0.3043,
+      "num_input_tokens_seen": 199903400,
+      "step": 3640,
+      "train_runtime": 20534.4715,
+      "train_tokens_per_second": 9735.016
+    },
+    {
+      "epoch": 1.5289429530201342,
+      "grad_norm": 0.6776718357161892,
+      "learning_rate": 1.3096590426850913e-06,
+      "loss": 0.2839,
+      "num_input_tokens_seen": 200189480,
+      "step": 3645,
+      "train_runtime": 20560.3756,
+      "train_tokens_per_second": 9736.665
+    },
+    {
+      "epoch": 1.5310402684563758,
+      "grad_norm": 0.6700915874976565,
+      "learning_rate": 1.2985648136256673e-06,
+      "loss": 0.3015,
+      "num_input_tokens_seen": 200459560,
+      "step": 3650,
+      "train_runtime": 20584.4789,
+      "train_tokens_per_second": 9738.384
+    },
+    {
+      "epoch": 1.5331375838926173,
+      "grad_norm": 0.7261993949044758,
+      "learning_rate": 1.287510757820823e-06,
+      "loss": 0.2878,
+      "num_input_tokens_seen": 200728264,
+      "step": 3655,
+      "train_runtime": 20612.4493,
+      "train_tokens_per_second": 9738.205
+    },
+    {
+      "epoch": 1.535234899328859,
+      "grad_norm": 0.6644891266786439,
+      "learning_rate": 1.276496995244938e-06,
+      "loss": 0.3082,
+      "num_input_tokens_seen": 201001792,
+      "step": 3660,
+      "train_runtime": 20639.2257,
+      "train_tokens_per_second": 9738.824
+    },
+    {
+      "epoch": 1.5373322147651005,
+      "grad_norm": 0.6421204872452299,
+      "learning_rate": 1.2655236454350777e-06,
+      "loss": 0.3121,
+      "num_input_tokens_seen": 201292912,
+      "step": 3665,
+      "train_runtime": 20668.9565,
+      "train_tokens_per_second": 9738.901
+    },
+    {
+      "epoch": 1.5394295302013423,
+      "grad_norm": 0.853463329352908,
+      "learning_rate": 1.254590827489684e-06,
+      "loss": 0.3101,
+      "num_input_tokens_seen": 201583168,
+      "step": 3670,
+      "train_runtime": 20689.5794,
+      "train_tokens_per_second": 9743.222
+    },
+    {
+      "epoch": 1.541526845637584,
+      "grad_norm": 0.7140020355753031,
+      "learning_rate": 1.2436986600672962e-06,
+      "loss": 0.2847,
+      "num_input_tokens_seen": 201866960,
+      "step": 3675,
+      "train_runtime": 20715.8813,
+      "train_tokens_per_second": 9744.551
+    },
+    {
+      "epoch": 1.5436241610738255,
+      "grad_norm": 0.6574456158021968,
+      "learning_rate": 1.23284726138525e-06,
+      "loss": 0.2893,
+      "num_input_tokens_seen": 202150640,
+      "step": 3680,
+      "train_runtime": 20748.8038,
+      "train_tokens_per_second": 9742.761
+    },
+    {
+      "epoch": 1.545721476510067,
+      "grad_norm": 0.7862177984104778,
+      "learning_rate": 1.2220367492184038e-06,
+      "loss": 0.3039,
+      "num_input_tokens_seen": 202422928,
+      "step": 3685,
+      "train_runtime": 20777.6595,
+      "train_tokens_per_second": 9742.335
+    },
+    {
+      "epoch": 1.5478187919463087,
+      "grad_norm": 0.7141933304409352,
+      "learning_rate": 1.2112672408978555e-06,
+      "loss": 0.291,
+      "num_input_tokens_seen": 202706216,
+      "step": 3690,
+      "train_runtime": 20798.3682,
+      "train_tokens_per_second": 9746.256
+    },
+    {
+      "epoch": 1.5499161073825505,
+      "grad_norm": 0.9340902454941875,
+      "learning_rate": 1.2005388533096719e-06,
+      "loss": 0.2919,
+      "num_input_tokens_seen": 203001376,
+      "step": 3695,
+      "train_runtime": 20827.2908,
+      "train_tokens_per_second": 9746.893
+    },
+    {
+      "epoch": 1.552013422818792,
+      "grad_norm": 0.8729016389049153,
+      "learning_rate": 1.189851702893619e-06,
+      "loss": 0.3205,
+      "num_input_tokens_seen": 203269776,
+      "step": 3700,
+      "train_runtime": 20854.1385,
+      "train_tokens_per_second": 9747.215
+    },
+    {
+      "epoch": 1.5541107382550337,
+      "grad_norm": 0.8613485145445854,
+      "learning_rate": 1.179205905641898e-06,
+      "loss": 0.2992,
+      "num_input_tokens_seen": 203532696,
+      "step": 3705,
+      "train_runtime": 20884.8222,
+      "train_tokens_per_second": 9745.484
+    },
+    {
+      "epoch": 1.5562080536912752,
+      "grad_norm": 0.6726114169701116,
+      "learning_rate": 1.1686015770978842e-06,
+      "loss": 0.2883,
+      "num_input_tokens_seen": 203824936,
+      "step": 3710,
+      "train_runtime": 20916.5004,
+      "train_tokens_per_second": 9744.696
+    },
+    {
+      "epoch": 1.5583053691275168,
+      "grad_norm": 0.7803189647409569,
+      "learning_rate": 1.1580388323548814e-06,
+      "loss": 0.2989,
+      "num_input_tokens_seen": 204084440,
+      "step": 3715,
+      "train_runtime": 20950.6482,
+      "train_tokens_per_second": 9741.199
+    },
+    {
+      "epoch": 1.5604026845637584,
+      "grad_norm": 0.7891485688402584,
+      "learning_rate": 1.147517786054859e-06,
+      "loss": 0.3013,
+      "num_input_tokens_seen": 204338816,
+      "step": 3720,
+      "train_runtime": 20977.8572,
+      "train_tokens_per_second": 9740.691
+    },
+    {
+      "epoch": 1.5625,
+      "grad_norm": 0.6695080949075083,
+      "learning_rate": 1.137038552387223e-06,
+      "loss": 0.3029,
+      "num_input_tokens_seen": 204623000,
+      "step": 3725,
+      "train_runtime": 21004.8644,
+      "train_tokens_per_second": 9741.696
+    },
+    {
+      "epoch": 1.5645973154362416,
+      "grad_norm": 0.7913476009870868,
+      "learning_rate": 1.1266012450875636e-06,
+      "loss": 0.3104,
+      "num_input_tokens_seen": 204877480,
+      "step": 3730,
+      "train_runtime": 21031.5859,
+      "train_tokens_per_second": 9741.419
+    },
+    {
+      "epoch": 1.5666946308724832,
+      "grad_norm": 0.8795126369533507,
+      "learning_rate": 1.1162059774364276e-06,
+      "loss": 0.2953,
+      "num_input_tokens_seen": 205156864,
+      "step": 3735,
+      "train_runtime": 21062.2533,
+      "train_tokens_per_second": 9740.499
+    },
+    {
+      "epoch": 1.5687919463087248,
+      "grad_norm": 0.7163618856405364,
+      "learning_rate": 1.1058528622580884e-06,
+      "loss": 0.2908,
+      "num_input_tokens_seen": 205442032,
+      "step": 3740,
+      "train_runtime": 21088.1021,
+      "train_tokens_per_second": 9742.083
+    },
+    {
+      "epoch": 1.5708892617449663,
+      "grad_norm": 0.7406126215572685,
+      "learning_rate": 1.09554201191932e-06,
+      "loss": 0.3125,
+      "num_input_tokens_seen": 205708304,
+      "step": 3745,
+      "train_runtime": 21115.9558,
+      "train_tokens_per_second": 9741.842
+    },
+    {
+      "epoch": 1.572986577181208,
+      "grad_norm": 0.6889497990498555,
+      "learning_rate": 1.0852735383281786e-06,
+      "loss": 0.2996,
+      "num_input_tokens_seen": 206001152,
+      "step": 3750,
+      "train_runtime": 21146.1114,
+      "train_tokens_per_second": 9741.798
+    },
+    {
+      "epoch": 1.5750838926174495,
+      "grad_norm": 0.7745914600158187,
+      "learning_rate": 1.0750475529327863e-06,
+      "loss": 0.3061,
+      "num_input_tokens_seen": 206270616,
+      "step": 3755,
+      "train_runtime": 21171.9992,
+      "train_tokens_per_second": 9742.614
+    },
+    {
+      "epoch": 1.5771812080536913,
+      "grad_norm": 0.7533218479502534,
+      "learning_rate": 1.0648641667201237e-06,
+      "loss": 0.3057,
+      "num_input_tokens_seen": 206541888,
+      "step": 3760,
+      "train_runtime": 21210.2665,
+      "train_tokens_per_second": 9737.826
+    },
+    {
+      "epoch": 1.579278523489933,
+      "grad_norm": 0.669825481945234,
+      "learning_rate": 1.054723490214825e-06,
+      "loss": 0.2855,
+      "num_input_tokens_seen": 206839896,
+      "step": 3765,
+      "train_runtime": 21236.6745,
+      "train_tokens_per_second": 9739.75
+    },
+    {
+      "epoch": 1.5813758389261745,
+      "grad_norm": 0.6641674278404036,
+      "learning_rate": 1.0446256334779752e-06,
+      "loss": 0.2972,
+      "num_input_tokens_seen": 207122592,
+      "step": 3770,
+      "train_runtime": 21267.5987,
+      "train_tokens_per_second": 9738.88
+    },
+    {
+      "epoch": 1.583473154362416,
+      "grad_norm": 0.6578082107005069,
+      "learning_rate": 1.034570706105923e-06,
+      "loss": 0.2938,
+      "num_input_tokens_seen": 207402456,
+      "step": 3775,
+      "train_runtime": 21301.4799,
+      "train_tokens_per_second": 9736.528
+    },
+    {
+      "epoch": 1.5855704697986577,
+      "grad_norm": 0.7197616091576533,
+      "learning_rate": 1.0245588172290804e-06,
+      "loss": 0.2729,
+      "num_input_tokens_seen": 207676216,
+      "step": 3780,
+      "train_runtime": 21327.8586,
+      "train_tokens_per_second": 9737.322
+    },
+    {
+      "epoch": 1.5876677852348995,
+      "grad_norm": 0.7287693943336037,
+      "learning_rate": 1.014590075510748e-06,
+      "loss": 0.3111,
+      "num_input_tokens_seen": 207943936,
+      "step": 3785,
+      "train_runtime": 21351.001,
+      "train_tokens_per_second": 9739.306
+    },
+    {
+      "epoch": 1.589765100671141,
+      "grad_norm": 0.7814037670136439,
+      "learning_rate": 1.0046645891459305e-06,
+      "loss": 0.2893,
+      "num_input_tokens_seen": 208229168,
+      "step": 3790,
+      "train_runtime": 21376.2842,
+      "train_tokens_per_second": 9741.13
+    },
+    {
+      "epoch": 1.5918624161073827,
+      "grad_norm": 0.7659158564530405,
+      "learning_rate": 9.947824658601645e-07,
+      "loss": 0.286,
+      "num_input_tokens_seen": 208497736,
+      "step": 3795,
+      "train_runtime": 21410.4256,
+      "train_tokens_per_second": 9738.141
+    },
+    {
+      "epoch": 1.5939597315436242,
+      "grad_norm": 0.8369725233507431,
+      "learning_rate": 9.849438129083484e-07,
+      "loss": 0.3098,
+      "num_input_tokens_seen": 208738696,
+      "step": 3800,
+      "train_runtime": 21433.1689,
+      "train_tokens_per_second": 9739.05
+    },
+    {
+      "epoch": 1.5960570469798658,
+      "grad_norm": 0.7929290938235205,
+      "learning_rate": 9.75148737073578e-07,
+      "loss": 0.3036,
+      "num_input_tokens_seen": 209012152,
+      "step": 3805,
+      "train_runtime": 21460.1301,
+      "train_tokens_per_second": 9739.557
+    },
+    {
+      "epoch": 1.5981543624161074,
+      "grad_norm": 0.8689979516236798,
+      "learning_rate": 9.653973446659875e-07,
+      "loss": 0.2835,
+      "num_input_tokens_seen": 209310328,
+      "step": 3810,
+      "train_runtime": 21487.83,
+      "train_tokens_per_second": 9740.878
+    },
+    {
+      "epoch": 1.600251677852349,
+      "grad_norm": 0.6189971304534108,
+      "learning_rate": 9.556897415215999e-07,
+      "loss": 0.3026,
+      "num_input_tokens_seen": 209598432,
+      "step": 3815,
+      "train_runtime": 21518.5176,
+      "train_tokens_per_second": 9740.375
+    },
+    {
+      "epoch": 1.6023489932885906,
+      "grad_norm": 0.6208866968463785,
+      "learning_rate": 9.460260330011706e-07,
+      "loss": 0.2924,
+      "num_input_tokens_seen": 209869144,
+      "step": 3820,
+      "train_runtime": 21549.1756,
+      "train_tokens_per_second": 9739.08
+    },
+    {
+      "epoch": 1.6044463087248322,
+      "grad_norm": 0.6673455557993474,
+      "learning_rate": 9.364063239890476e-07,
+      "loss": 0.2861,
+      "num_input_tokens_seen": 210140504,
+      "step": 3825,
+      "train_runtime": 21580.4796,
+      "train_tokens_per_second": 9737.527
+    },
+    {
+      "epoch": 1.6065436241610738,
+      "grad_norm": 0.7772502716442683,
+      "learning_rate": 9.268307188920384e-07,
+      "loss": 0.3088,
+      "num_input_tokens_seen": 210424160,
+      "step": 3830,
+      "train_runtime": 21610.7804,
+      "train_tokens_per_second": 9737.0
+    },
+    {
+      "epoch": 1.6086409395973154,
+      "grad_norm": 0.8367325246137529,
+      "learning_rate": 9.172993216382669e-07,
+      "loss": 0.2981,
+      "num_input_tokens_seen": 210685864,
+      "step": 3835,
+      "train_runtime": 21631.7769,
+      "train_tokens_per_second": 9739.647
+    },
+    {
+      "epoch": 1.610738255033557,
+      "grad_norm": 0.6464477338145745,
+      "learning_rate": 9.078122356760532e-07,
+      "loss": 0.2888,
+      "num_input_tokens_seen": 210960936,
+      "step": 3840,
+      "train_runtime": 21661.9398,
+      "train_tokens_per_second": 9738.783
+    },
+    {
+      "epoch": 1.6128355704697985,
+      "grad_norm": 0.7257694749278484,
+      "learning_rate": 8.983695639727858e-07,
+      "loss": 0.2954,
+      "num_input_tokens_seen": 211238896,
+      "step": 3845,
+      "train_runtime": 21688.6464,
+      "train_tokens_per_second": 9739.607
+    },
+    {
+      "epoch": 1.6149328859060401,
+      "grad_norm": 0.8837856987473485,
+      "learning_rate": 8.889714090138085e-07,
+      "loss": 0.3006,
+      "num_input_tokens_seen": 211504112,
+      "step": 3850,
+      "train_runtime": 21722.3164,
+      "train_tokens_per_second": 9736.72
+    },
+    {
+      "epoch": 1.617030201342282,
+      "grad_norm": 0.7274467987239595,
+      "learning_rate": 8.796178728013033e-07,
+      "loss": 0.3209,
+      "num_input_tokens_seen": 211787368,
+      "step": 3855,
+      "train_runtime": 21754.5579,
+      "train_tokens_per_second": 9735.31
+    },
+    {
+      "epoch": 1.6191275167785235,
+      "grad_norm": 0.6516086694185496,
+      "learning_rate": 8.703090568531902e-07,
+      "loss": 0.2914,
+      "num_input_tokens_seen": 212051184,
+      "step": 3860,
+      "train_runtime": 21782.7693,
+      "train_tokens_per_second": 9734.813
+    },
+    {
+      "epoch": 1.621224832214765,
+      "grad_norm": 0.7098367251569175,
+      "learning_rate": 8.610450622020178e-07,
+      "loss": 0.3153,
+      "num_input_tokens_seen": 212315840,
+      "step": 3865,
+      "train_runtime": 21804.3889,
+      "train_tokens_per_second": 9737.298
+    },
+    {
+      "epoch": 1.6233221476510067,
+      "grad_norm": 0.7920841714726178,
+      "learning_rate": 8.518259893938697e-07,
+      "loss": 0.3077,
+      "num_input_tokens_seen": 212589256,
+      "step": 3870,
+      "train_runtime": 21837.0681,
+      "train_tokens_per_second": 9735.247
+    },
+    {
+      "epoch": 1.6254194630872483,
+      "grad_norm": 0.6649687986299274,
+      "learning_rate": 8.426519384872733e-07,
+      "loss": 0.3027,
+      "num_input_tokens_seen": 212869312,
+      "step": 3875,
+      "train_runtime": 21865.6125,
+      "train_tokens_per_second": 9735.346
+    },
+    {
+      "epoch": 1.62751677852349,
+      "grad_norm": 0.9063947313979255,
+      "learning_rate": 8.335230090521191e-07,
+      "loss": 0.2866,
+      "num_input_tokens_seen": 213142584,
+      "step": 3880,
+      "train_runtime": 21890.4307,
+      "train_tokens_per_second": 9736.793
+    },
+    {
+      "epoch": 1.6296140939597317,
+      "grad_norm": 0.7942246337186094,
+      "learning_rate": 8.244393001685697e-07,
+      "loss": 0.2829,
+      "num_input_tokens_seen": 213402088,
+      "step": 3885,
+      "train_runtime": 21916.9133,
+      "train_tokens_per_second": 9736.868
+    },
+    {
+      "epoch": 1.6317114093959733,
+      "grad_norm": 0.7566316119245126,
+      "learning_rate": 8.154009104259907e-07,
+      "loss": 0.289,
+      "num_input_tokens_seen": 213668520,
+      "step": 3890,
+      "train_runtime": 21945.6002,
+      "train_tokens_per_second": 9736.281
+    },
+    {
+      "epoch": 1.6338087248322148,
+      "grad_norm": 0.7107153849510338,
+      "learning_rate": 8.064079379218815e-07,
+      "loss": 0.3046,
+      "num_input_tokens_seen": 213955440,
+      "step": 3895,
+      "train_runtime": 21978.9655,
+      "train_tokens_per_second": 9734.555
+    },
+    {
+      "epoch": 1.6359060402684564,
+      "grad_norm": 0.898154551862841,
+      "learning_rate": 7.974604802608055e-07,
+      "loss": 0.2979,
+      "num_input_tokens_seen": 214220256,
+      "step": 3900,
+      "train_runtime": 22012.7903,
+      "train_tokens_per_second": 9731.627
+    },
+    {
+      "epoch": 1.638003355704698,
+      "grad_norm": 0.6698549226951278,
+      "learning_rate": 7.885586345533397e-07,
+      "loss": 0.2971,
+      "num_input_tokens_seen": 214483880,
+      "step": 3905,
+      "train_runtime": 22038.4728,
+      "train_tokens_per_second": 9732.248
+    },
+    {
+      "epoch": 1.6401006711409396,
+      "grad_norm": 0.7375639828871103,
+      "learning_rate": 7.797024974150102e-07,
+      "loss": 0.3178,
+      "num_input_tokens_seen": 214762960,
+      "step": 3910,
+      "train_runtime": 22068.2628,
+      "train_tokens_per_second": 9731.756
+    },
+    {
+      "epoch": 1.6421979865771812,
+      "grad_norm": 0.6843460107974498,
+      "learning_rate": 7.708921649652501e-07,
+      "loss": 0.3121,
+      "num_input_tokens_seen": 215049064,
+      "step": 3915,
+      "train_runtime": 22100.0631,
+      "train_tokens_per_second": 9730.699
+    },
+    {
+      "epoch": 1.6442953020134228,
+      "grad_norm": 0.6217081306932554,
+      "learning_rate": 7.621277328263548e-07,
+      "loss": 0.2919,
+      "num_input_tokens_seen": 215331416,
+      "step": 3920,
+      "train_runtime": 22131.1919,
+      "train_tokens_per_second": 9729.77
+    },
+    {
+      "epoch": 1.6463926174496644,
+      "grad_norm": 0.6772036999928911,
+      "learning_rate": 7.534092961224443e-07,
+      "loss": 0.2848,
+      "num_input_tokens_seen": 215618952,
+      "step": 3925,
+      "train_runtime": 22162.0139,
+      "train_tokens_per_second": 9729.213
+    },
+    {
+      "epoch": 1.648489932885906,
+      "grad_norm": 0.9342397488026769,
+      "learning_rate": 7.447369494784285e-07,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 215900768,
+      "step": 3930,
+      "train_runtime": 22183.8622,
+      "train_tokens_per_second": 9732.335
+    },
+    {
+      "epoch": 1.6505872483221475,
+      "grad_norm": 0.680506878218271,
+      "learning_rate": 7.36110787018986e-07,
+      "loss": 0.3071,
+      "num_input_tokens_seen": 216193600,
+      "step": 3935,
+      "train_runtime": 22214.3035,
+      "train_tokens_per_second": 9732.18
+    },
+    {
+      "epoch": 1.6526845637583891,
+      "grad_norm": 0.6873020712947707,
+      "learning_rate": 7.275309023675364e-07,
+      "loss": 0.2871,
+      "num_input_tokens_seen": 216480520,
+      "step": 3940,
+      "train_runtime": 22244.9078,
+      "train_tokens_per_second": 9731.689
+    },
+    {
+      "epoch": 1.6547818791946307,
+      "grad_norm": 0.7246986556878467,
+      "learning_rate": 7.189973886452251e-07,
+      "loss": 0.2935,
+      "num_input_tokens_seen": 216750304,
+      "step": 3945,
+      "train_runtime": 22271.7759,
+      "train_tokens_per_second": 9732.062
+    },
+    {
+      "epoch": 1.6568791946308725,
+      "grad_norm": 0.6430611799242064,
+      "learning_rate": 7.105103384699175e-07,
+      "loss": 0.3045,
+      "num_input_tokens_seen": 217027832,
+      "step": 3950,
+      "train_runtime": 22298.7567,
+      "train_tokens_per_second": 9732.732
+    },
+    {
+      "epoch": 1.658976510067114,
+      "grad_norm": 0.7037470857102867,
+      "learning_rate": 7.020698439551881e-07,
+      "loss": 0.2859,
+      "num_input_tokens_seen": 217273920,
+      "step": 3955,
+      "train_runtime": 22321.9968,
+      "train_tokens_per_second": 9733.624
+    },
+    {
+      "epoch": 1.6610738255033557,
+      "grad_norm": 0.7317146711392565,
+      "learning_rate": 6.936759967093237e-07,
+      "loss": 0.2799,
+      "num_input_tokens_seen": 217544344,
+      "step": 3960,
+      "train_runtime": 22348.6862,
+      "train_tokens_per_second": 9734.1
+    },
+    {
+      "epoch": 1.6631711409395973,
+      "grad_norm": 0.6510345582311675,
+      "learning_rate": 6.853288878343272e-07,
+      "loss": 0.2825,
+      "num_input_tokens_seen": 217830456,
+      "step": 3965,
+      "train_runtime": 22378.0972,
+      "train_tokens_per_second": 9734.092
+    },
+    {
+      "epoch": 1.665268456375839,
+      "grad_norm": 0.7121256230596269,
+      "learning_rate": 6.770286079249311e-07,
+      "loss": 0.2926,
+      "num_input_tokens_seen": 218118376,
+      "step": 3970,
+      "train_runtime": 22410.2916,
+      "train_tokens_per_second": 9732.956
+    },
+    {
+      "epoch": 1.6673657718120807,
+      "grad_norm": 0.6870302910439189,
+      "learning_rate": 6.687752470676129e-07,
+      "loss": 0.3223,
+      "num_input_tokens_seen": 218374488,
+      "step": 3975,
+      "train_runtime": 22434.3638,
+      "train_tokens_per_second": 9733.928
+    },
+    {
+      "epoch": 1.6694630872483223,
+      "grad_norm": 0.6922178912532768,
+      "learning_rate": 6.605688948396182e-07,
+      "loss": 0.2836,
+      "num_input_tokens_seen": 218643344,
+      "step": 3980,
+      "train_runtime": 22459.0097,
+      "train_tokens_per_second": 9735.217
+    },
+    {
+      "epoch": 1.6715604026845639,
+      "grad_norm": 0.8761793031057835,
+      "learning_rate": 6.524096403079861e-07,
+      "loss": 0.2947,
+      "num_input_tokens_seen": 218919152,
+      "step": 3985,
+      "train_runtime": 22484.076,
+      "train_tokens_per_second": 9736.631
+    },
+    {
+      "epoch": 1.6736577181208054,
+      "grad_norm": 0.7590734628561895,
+      "learning_rate": 6.442975720285871e-07,
+      "loss": 0.3025,
+      "num_input_tokens_seen": 219188776,
+      "step": 3990,
+      "train_runtime": 22516.395,
+      "train_tokens_per_second": 9734.63
+    },
+    {
+      "epoch": 1.675755033557047,
+      "grad_norm": 0.730584669464086,
+      "learning_rate": 6.362327780451566e-07,
+      "loss": 0.2785,
+      "num_input_tokens_seen": 219472024,
+      "step": 3995,
+      "train_runtime": 22549.8052,
+      "train_tokens_per_second": 9732.768
+    },
+    {
+      "epoch": 1.6778523489932886,
+      "grad_norm": 0.6670941549373599,
+      "learning_rate": 6.282153458883444e-07,
+      "loss": 0.3049,
+      "num_input_tokens_seen": 219748896,
+      "step": 4000,
+      "train_runtime": 22579.0956,
+      "train_tokens_per_second": 9732.405
+    },
+    {
+      "epoch": 1.6799496644295302,
+      "grad_norm": 0.7699909656314539,
+      "learning_rate": 6.2024536257476e-07,
+      "loss": 0.3112,
+      "num_input_tokens_seen": 220025688,
+      "step": 4005,
+      "train_runtime": 22607.1962,
+      "train_tokens_per_second": 9732.551
+    },
+    {
+      "epoch": 1.6820469798657718,
+      "grad_norm": 0.7035714093030574,
+      "learning_rate": 6.123229146060311e-07,
+      "loss": 0.2858,
+      "num_input_tokens_seen": 220306704,
+      "step": 4010,
+      "train_runtime": 22635.1183,
+      "train_tokens_per_second": 9732.96
+    },
+    {
+      "epoch": 1.6841442953020134,
+      "grad_norm": 0.7130061807882924,
+      "learning_rate": 6.044480879678644e-07,
+      "loss": 0.2978,
+      "num_input_tokens_seen": 220582216,
+      "step": 4015,
+      "train_runtime": 22664.1986,
+      "train_tokens_per_second": 9732.628
+    },
+    {
+      "epoch": 1.686241610738255,
+      "grad_norm": 0.7810765354783282,
+      "learning_rate": 5.966209681291108e-07,
+      "loss": 0.3,
+      "num_input_tokens_seen": 220885376,
+      "step": 4020,
+      "train_runtime": 22693.6348,
+      "train_tokens_per_second": 9733.363
+    },
+    {
+      "epoch": 1.6883389261744965,
+      "grad_norm": 0.6454794320430959,
+      "learning_rate": 5.888416400408404e-07,
+      "loss": 0.2861,
+      "num_input_tokens_seen": 221164784,
+      "step": 4025,
+      "train_runtime": 22724.4758,
+      "train_tokens_per_second": 9732.448
+    },
+    {
+      "epoch": 1.6904362416107381,
+      "grad_norm": 0.6285242354696032,
+      "learning_rate": 5.811101881354187e-07,
+      "loss": 0.2934,
+      "num_input_tokens_seen": 221439832,
+      "step": 4030,
+      "train_runtime": 22753.7352,
+      "train_tokens_per_second": 9732.021
+    },
+    {
+      "epoch": 1.6925335570469797,
+      "grad_norm": 0.6675608471230159,
+      "learning_rate": 5.734266963255891e-07,
+      "loss": 0.3013,
+      "num_input_tokens_seen": 221714544,
+      "step": 4035,
+      "train_runtime": 22786.0663,
+      "train_tokens_per_second": 9730.269
+    },
+    {
+      "epoch": 1.6946308724832215,
+      "grad_norm": 0.7075195841984412,
+      "learning_rate": 5.657912480035665e-07,
+      "loss": 0.2911,
+      "num_input_tokens_seen": 221972744,
+      "step": 4040,
+      "train_runtime": 22814.3068,
+      "train_tokens_per_second": 9729.541
+    },
+    {
+      "epoch": 1.6967281879194631,
+      "grad_norm": 0.7183840746675496,
+      "learning_rate": 5.582039260401268e-07,
+      "loss": 0.2973,
+      "num_input_tokens_seen": 222245928,
+      "step": 4045,
+      "train_runtime": 22843.4352,
+      "train_tokens_per_second": 9729.094
+    },
+    {
+      "epoch": 1.6988255033557047,
+      "grad_norm": 0.8730920321772149,
+      "learning_rate": 5.506648127837116e-07,
+      "loss": 0.3024,
+      "num_input_tokens_seen": 222544328,
+      "step": 4050,
+      "train_runtime": 22870.0,
+      "train_tokens_per_second": 9730.841
+    },
+    {
+      "epoch": 1.7009228187919463,
+      "grad_norm": 0.6037709964407112,
+      "learning_rate": 5.431739900595323e-07,
+      "loss": 0.2984,
+      "num_input_tokens_seen": 222834896,
+      "step": 4055,
+      "train_runtime": 22897.9943,
+      "train_tokens_per_second": 9731.634
+    },
+    {
+      "epoch": 1.7030201342281879,
+      "grad_norm": 0.7179979957366062,
+      "learning_rate": 5.357315391686818e-07,
+      "loss": 0.3149,
+      "num_input_tokens_seen": 223115968,
+      "step": 4060,
+      "train_runtime": 22930.647,
+      "train_tokens_per_second": 9730.034
+    },
+    {
+      "epoch": 1.7051174496644297,
+      "grad_norm": 0.6848260563241362,
+      "learning_rate": 5.283375408872538e-07,
+      "loss": 0.2983,
+      "num_input_tokens_seen": 223387552,
+      "step": 4065,
+      "train_runtime": 22952.8793,
+      "train_tokens_per_second": 9732.441
+    },
+    {
+      "epoch": 1.7072147651006713,
+      "grad_norm": 0.7274227594141781,
+      "learning_rate": 5.209920754654646e-07,
+      "loss": 0.3039,
+      "num_input_tokens_seen": 223644352,
+      "step": 4070,
+      "train_runtime": 22980.1197,
+      "train_tokens_per_second": 9732.08
+    },
+    {
+      "epoch": 1.7093120805369129,
+      "grad_norm": 0.6746751396250549,
+      "learning_rate": 5.136952226267833e-07,
+      "loss": 0.2898,
+      "num_input_tokens_seen": 223919792,
+      "step": 4075,
+      "train_runtime": 23004.3142,
+      "train_tokens_per_second": 9733.817
+    },
+    {
+      "epoch": 1.7114093959731544,
+      "grad_norm": 0.8248176954419442,
+      "learning_rate": 5.064470615670652e-07,
+      "loss": 0.3158,
+      "num_input_tokens_seen": 224199688,
+      "step": 4080,
+      "train_runtime": 23035.0175,
+      "train_tokens_per_second": 9732.994
+    },
+    {
+      "epoch": 1.713506711409396,
+      "grad_norm": 0.7719785594681554,
+      "learning_rate": 4.992476709536925e-07,
+      "loss": 0.302,
+      "num_input_tokens_seen": 224476120,
+      "step": 4085,
+      "train_runtime": 23066.5056,
+      "train_tokens_per_second": 9731.692
+    },
+    {
+      "epoch": 1.7156040268456376,
+      "grad_norm": 0.7714057670569905,
+      "learning_rate": 4.92097128924724e-07,
+      "loss": 0.2842,
+      "num_input_tokens_seen": 224744352,
+      "step": 4090,
+      "train_runtime": 23092.5368,
+      "train_tokens_per_second": 9732.337
+    },
+    {
+      "epoch": 1.7177013422818792,
+      "grad_norm": 0.6525044649976216,
+      "learning_rate": 4.849955130880391e-07,
+      "loss": 0.3091,
+      "num_input_tokens_seen": 225006760,
+      "step": 4095,
+      "train_runtime": 23121.0252,
+      "train_tokens_per_second": 9731.695
+    },
+    {
+      "epoch": 1.7197986577181208,
+      "grad_norm": 0.6484642320570236,
+      "learning_rate": 4.779429005205055e-07,
+      "loss": 0.2971,
+      "num_input_tokens_seen": 225290736,
+      "step": 4100,
+      "train_runtime": 23151.6296,
+      "train_tokens_per_second": 9731.096
+    },
+    {
+      "epoch": 1.7218959731543624,
+      "grad_norm": 0.7147191038109459,
+      "learning_rate": 4.709393677671337e-07,
+      "loss": 0.3045,
+      "num_input_tokens_seen": 225590904,
+      "step": 4105,
+      "train_runtime": 23177.7912,
+      "train_tokens_per_second": 9733.063
+    },
+    {
+      "epoch": 1.723993288590604,
+      "grad_norm": 0.7716793766328921,
+      "learning_rate": 4.639849908402516e-07,
+      "loss": 0.2752,
+      "num_input_tokens_seen": 225844576,
+      "step": 4110,
+      "train_runtime": 23206.4955,
+      "train_tokens_per_second": 9731.955
+    },
+    {
+      "epoch": 1.7260906040268456,
+      "grad_norm": 0.6370652490916938,
+      "learning_rate": 4.570798452186775e-07,
+      "loss": 0.3133,
+      "num_input_tokens_seen": 226130472,
+      "step": 4115,
+      "train_runtime": 23239.7502,
+      "train_tokens_per_second": 9730.331
+    },
+    {
+      "epoch": 1.7281879194630871,
+      "grad_norm": 0.7462472852937805,
+      "learning_rate": 4.5022400584690207e-07,
+      "loss": 0.3005,
+      "num_input_tokens_seen": 226394656,
+      "step": 4120,
+      "train_runtime": 23263.3063,
+      "train_tokens_per_second": 9731.835
+    },
+    {
+      "epoch": 1.7302852348993287,
+      "grad_norm": 0.794249978627542,
+      "learning_rate": 4.4341754713427334e-07,
+      "loss": 0.3038,
+      "num_input_tokens_seen": 226672784,
+      "step": 4125,
+      "train_runtime": 23291.508,
+      "train_tokens_per_second": 9731.993
+    },
+    {
+      "epoch": 1.7323825503355703,
+      "grad_norm": 0.724303754347132,
+      "learning_rate": 4.3666054295419035e-07,
+      "loss": 0.3141,
+      "num_input_tokens_seen": 226959960,
+      "step": 4130,
+      "train_runtime": 23319.6896,
+      "train_tokens_per_second": 9732.546
+    },
+    {
+      "epoch": 1.7344798657718121,
+      "grad_norm": 0.8193340898447615,
+      "learning_rate": 4.299530666433005e-07,
+      "loss": 0.3181,
+      "num_input_tokens_seen": 227214248,
+      "step": 4135,
+      "train_runtime": 23345.803,
+      "train_tokens_per_second": 9732.552
+    },
+    {
+      "epoch": 1.7365771812080537,
+      "grad_norm": 0.6869797179901593,
+      "learning_rate": 4.2329519100070607e-07,
+      "loss": 0.2925,
+      "num_input_tokens_seen": 227488192,
+      "step": 4140,
+      "train_runtime": 23370.9245,
+      "train_tokens_per_second": 9733.812
+    },
+    {
+      "epoch": 1.7386744966442953,
+      "grad_norm": 0.6476836091125182,
+      "learning_rate": 4.1668698828716994e-07,
+      "loss": 0.2745,
+      "num_input_tokens_seen": 227791312,
+      "step": 4145,
+      "train_runtime": 23398.4763,
+      "train_tokens_per_second": 9735.305
+    },
+    {
+      "epoch": 1.7407718120805369,
+      "grad_norm": 0.6703806823169557,
+      "learning_rate": 4.101285302243335e-07,
+      "loss": 0.2889,
+      "num_input_tokens_seen": 228064408,
+      "step": 4150,
+      "train_runtime": 23429.0979,
+      "train_tokens_per_second": 9734.238
+    },
+    {
+      "epoch": 1.7428691275167785,
+      "grad_norm": 0.677457266374521,
+      "learning_rate": 4.0361988799394014e-07,
+      "loss": 0.2979,
+      "num_input_tokens_seen": 228322552,
+      "step": 4155,
+      "train_runtime": 23454.6414,
+      "train_tokens_per_second": 9734.643
+    },
+    {
+      "epoch": 1.7449664429530203,
+      "grad_norm": 0.7694794654436745,
+      "learning_rate": 3.9716113223705945e-07,
+      "loss": 0.2963,
+      "num_input_tokens_seen": 228613776,
+      "step": 4160,
+      "train_runtime": 23490.1589,
+      "train_tokens_per_second": 9732.321
+    },
+    {
+      "epoch": 1.7470637583892619,
+      "grad_norm": 0.6310539917082916,
+      "learning_rate": 3.9075233305332196e-07,
+      "loss": 0.2945,
+      "num_input_tokens_seen": 228919384,
+      "step": 4165,
+      "train_runtime": 23513.3886,
+      "train_tokens_per_second": 9735.704
+    },
+    {
+      "epoch": 1.7491610738255035,
+      "grad_norm": 0.6993092741117894,
+      "learning_rate": 3.843935600001586e-07,
+      "loss": 0.286,
+      "num_input_tokens_seen": 229195344,
+      "step": 4170,
+      "train_runtime": 23535.6335,
+      "train_tokens_per_second": 9738.227
+    },
+    {
+      "epoch": 1.751258389261745,
+      "grad_norm": 0.7354840423080397,
+      "learning_rate": 3.780848820920441e-07,
+      "loss": 0.3106,
+      "num_input_tokens_seen": 229475688,
+      "step": 4175,
+      "train_runtime": 23564.6976,
+      "train_tokens_per_second": 9738.113
+    },
+    {
+      "epoch": 1.7533557046979866,
+      "grad_norm": 0.6750786212530938,
+      "learning_rate": 3.718263677997497e-07,
+      "loss": 0.3034,
+      "num_input_tokens_seen": 229768096,
+      "step": 4180,
+      "train_runtime": 23597.4223,
+      "train_tokens_per_second": 9737.0
+    },
+    {
+      "epoch": 1.7554530201342282,
+      "grad_norm": 0.7187886123648156,
+      "learning_rate": 3.6561808504960215e-07,
+      "loss": 0.2749,
+      "num_input_tokens_seen": 230028336,
+      "step": 4185,
+      "train_runtime": 23621.7266,
+      "train_tokens_per_second": 9737.998
+    },
+    {
+      "epoch": 1.7575503355704698,
+      "grad_norm": 0.7334723049748033,
+      "learning_rate": 3.5946010122274056e-07,
+      "loss": 0.2965,
+      "num_input_tokens_seen": 230295984,
+      "step": 4190,
+      "train_runtime": 23647.7616,
+      "train_tokens_per_second": 9738.595
+    },
+    {
+      "epoch": 1.7596476510067114,
+      "grad_norm": 0.8081901994313802,
+      "learning_rate": 3.533524831543894e-07,
+      "loss": 0.2941,
+      "num_input_tokens_seen": 230574400,
+      "step": 4195,
+      "train_runtime": 23675.5564,
+      "train_tokens_per_second": 9738.922
+    },
+    {
+      "epoch": 1.761744966442953,
+      "grad_norm": 0.8203570193097668,
+      "learning_rate": 3.472952971331323e-07,
+      "loss": 0.2878,
+      "num_input_tokens_seen": 230850480,
+      "step": 4200,
+      "train_runtime": 23708.9109,
+      "train_tokens_per_second": 9736.866
+    },
+    {
+      "epoch": 1.7638422818791946,
+      "grad_norm": 0.6430226374339747,
+      "learning_rate": 3.4128860890019156e-07,
+      "loss": 0.285,
+      "num_input_tokens_seen": 231157088,
+      "step": 4205,
+      "train_runtime": 23737.95,
+      "train_tokens_per_second": 9737.871
+    },
+    {
+      "epoch": 1.7659395973154361,
+      "grad_norm": 0.7325840386625063,
+      "learning_rate": 3.3533248364871763e-07,
+      "loss": 0.2927,
+      "num_input_tokens_seen": 231458880,
+      "step": 4210,
+      "train_runtime": 23764.9322,
+      "train_tokens_per_second": 9739.514
+    },
+    {
+      "epoch": 1.7680369127516777,
+      "grad_norm": 0.8018112735332531,
+      "learning_rate": 3.2942698602307755e-07,
+      "loss": 0.3109,
+      "num_input_tokens_seen": 231723544,
+      "step": 4215,
+      "train_runtime": 23788.3728,
+      "train_tokens_per_second": 9741.042
+    },
+    {
+      "epoch": 1.7701342281879193,
+      "grad_norm": 0.7336061156730936,
+      "learning_rate": 3.2357218011815464e-07,
+      "loss": 0.2997,
+      "num_input_tokens_seen": 231981712,
+      "step": 4220,
+      "train_runtime": 23813.4144,
+      "train_tokens_per_second": 9741.64
+    },
+    {
+      "epoch": 1.772231543624161,
+      "grad_norm": 0.8204165085174919,
+      "learning_rate": 3.177681294786539e-07,
+      "loss": 0.3031,
+      "num_input_tokens_seen": 232230376,
+      "step": 4225,
+      "train_runtime": 23846.2712,
+      "train_tokens_per_second": 9738.645
+    },
+    {
+      "epoch": 1.7743288590604027,
+      "grad_norm": 0.7324248095283954,
+      "learning_rate": 3.120148970984138e-07,
+      "loss": 0.3021,
+      "num_input_tokens_seen": 232521784,
+      "step": 4230,
+      "train_runtime": 23866.3957,
+      "train_tokens_per_second": 9742.643
+    },
+    {
+      "epoch": 1.7764261744966443,
+      "grad_norm": 0.7623423302545881,
+      "learning_rate": 3.06312545419718e-07,
+      "loss": 0.3058,
+      "num_input_tokens_seen": 232776224,
+      "step": 4235,
+      "train_runtime": 23894.42,
+      "train_tokens_per_second": 9741.865
+    },
+    {
+      "epoch": 1.778523489932886,
+      "grad_norm": 0.7160648386229338,
+      "learning_rate": 3.006611363326206e-07,
+      "loss": 0.3,
+      "num_input_tokens_seen": 233024192,
+      "step": 4240,
+      "train_runtime": 23924.6152,
+      "train_tokens_per_second": 9739.935
+    },
+    {
+      "epoch": 1.7806208053691275,
+      "grad_norm": 0.7310225525955845,
+      "learning_rate": 2.9506073117427534e-07,
+      "loss": 0.2907,
+      "num_input_tokens_seen": 233274136,
+      "step": 4245,
+      "train_runtime": 23950.6083,
+      "train_tokens_per_second": 9739.8
+    },
+    {
+      "epoch": 1.7827181208053693,
+      "grad_norm": 0.6573026156971243,
+      "learning_rate": 2.895113907282671e-07,
+      "loss": 0.3,
+      "num_input_tokens_seen": 233544960,
+      "step": 4250,
+      "train_runtime": 23985.4011,
+      "train_tokens_per_second": 9736.963
+    },
+    {
+      "epoch": 1.7848154362416109,
+      "grad_norm": 0.6849191806577527,
+      "learning_rate": 2.8401317522395333e-07,
+      "loss": 0.2876,
+      "num_input_tokens_seen": 233840288,
+      "step": 4255,
+      "train_runtime": 24010.3348,
+      "train_tokens_per_second": 9739.151
+    },
+    {
+      "epoch": 1.7869127516778525,
+      "grad_norm": 0.9502231271113692,
+      "learning_rate": 2.7856614433581397e-07,
+      "loss": 0.2873,
+      "num_input_tokens_seen": 234083912,
+      "step": 4260,
+      "train_runtime": 24039.5554,
+      "train_tokens_per_second": 9737.448
+    },
+    {
+      "epoch": 1.789010067114094,
+      "grad_norm": 0.6835352509350766,
+      "learning_rate": 2.7317035718279704e-07,
+      "loss": 0.3028,
+      "num_input_tokens_seen": 234332560,
+      "step": 4265,
+      "train_runtime": 24070.372,
+      "train_tokens_per_second": 9735.311
+    },
+    {
+      "epoch": 1.7911073825503356,
+      "grad_norm": 0.8008466508812631,
+      "learning_rate": 2.678258723276816e-07,
+      "loss": 0.3039,
+      "num_input_tokens_seen": 234624136,
+      "step": 4270,
+      "train_runtime": 24099.3879,
+      "train_tokens_per_second": 9735.689
+    },
+    {
+      "epoch": 1.7932046979865772,
+      "grad_norm": 0.6590648058284801,
+      "learning_rate": 2.6253274777644e-07,
+      "loss": 0.2957,
+      "num_input_tokens_seen": 234892712,
+      "step": 4275,
+      "train_runtime": 24129.6502,
+      "train_tokens_per_second": 9734.609
+    },
+    {
+      "epoch": 1.7953020134228188,
+      "grad_norm": 0.643863260212462,
+      "learning_rate": 2.5729104097761193e-07,
+      "loss": 0.3011,
+      "num_input_tokens_seen": 235183192,
+      "step": 4280,
+      "train_runtime": 24153.0378,
+      "train_tokens_per_second": 9737.21
+    },
+    {
+      "epoch": 1.7973993288590604,
+      "grad_norm": 0.6684813470180424,
+      "learning_rate": 2.5210080882167644e-07,
+      "loss": 0.2704,
+      "num_input_tokens_seen": 235470776,
+      "step": 4285,
+      "train_runtime": 24176.9486,
+      "train_tokens_per_second": 9739.475
+    },
+    {
+      "epoch": 1.799496644295302,
+      "grad_norm": 0.7933923689049249,
+      "learning_rate": 2.4696210764043547e-07,
+      "loss": 0.3232,
+      "num_input_tokens_seen": 235738944,
+      "step": 4290,
+      "train_runtime": 24209.3655,
+      "train_tokens_per_second": 9737.51
+    },
+    {
+      "epoch": 1.8015939597315436,
+      "grad_norm": 0.6153733501695048,
+      "learning_rate": 2.418749932064052e-07,
+      "loss": 0.2844,
+      "num_input_tokens_seen": 236019944,
+      "step": 4295,
+      "train_runtime": 24232.8612,
+      "train_tokens_per_second": 9739.665
+    },
+    {
+      "epoch": 1.8036912751677852,
+      "grad_norm": 0.6994051505180399,
+      "learning_rate": 2.3683952073220718e-07,
+      "loss": 0.2938,
+      "num_input_tokens_seen": 236282304,
+      "step": 4300,
+      "train_runtime": 24258.7805,
+      "train_tokens_per_second": 9740.073
+    },
+    {
+      "epoch": 1.8057885906040267,
+      "grad_norm": 0.824475976125504,
+      "learning_rate": 2.3185574486997264e-07,
+      "loss": 0.3108,
+      "num_input_tokens_seen": 236542272,
+      "step": 4305,
+      "train_runtime": 24286.4589,
+      "train_tokens_per_second": 9739.677
+    },
+    {
+      "epoch": 1.8078859060402683,
+      "grad_norm": 0.6933708263485241,
+      "learning_rate": 2.2692371971074488e-07,
+      "loss": 0.2946,
+      "num_input_tokens_seen": 236805848,
+      "step": 4310,
+      "train_runtime": 24320.242,
+      "train_tokens_per_second": 9736.986
+    },
+    {
+      "epoch": 1.80998322147651,
+      "grad_norm": 0.8348603313485197,
+      "learning_rate": 2.220434987838982e-07,
+      "loss": 0.3122,
+      "num_input_tokens_seen": 237073920,
+      "step": 4315,
+      "train_runtime": 24352.3031,
+      "train_tokens_per_second": 9735.174
+    },
+    {
+      "epoch": 1.8120805369127517,
+      "grad_norm": 0.7333496434913449,
+      "learning_rate": 2.172151350565499e-07,
+      "loss": 0.2899,
+      "num_input_tokens_seen": 237337800,
+      "step": 4320,
+      "train_runtime": 24380.9262,
+      "train_tokens_per_second": 9734.569
+    },
+    {
+      "epoch": 1.8141778523489933,
+      "grad_norm": 0.703145594261103,
+      "learning_rate": 2.1243868093299148e-07,
+      "loss": 0.2896,
+      "num_input_tokens_seen": 237592656,
+      "step": 4325,
+      "train_runtime": 24403.5247,
+      "train_tokens_per_second": 9735.998
+    },
+    {
+      "epoch": 1.816275167785235,
+      "grad_norm": 0.7388899452121637,
+      "learning_rate": 2.077141882541167e-07,
+      "loss": 0.2883,
+      "num_input_tokens_seen": 237868792,
+      "step": 4330,
+      "train_runtime": 24430.3326,
+      "train_tokens_per_second": 9736.617
+    },
+    {
+      "epoch": 1.8183724832214765,
+      "grad_norm": 0.7770038806555781,
+      "learning_rate": 2.0304170829685887e-07,
+      "loss": 0.2958,
+      "num_input_tokens_seen": 238118648,
+      "step": 4335,
+      "train_runtime": 24456.4653,
+      "train_tokens_per_second": 9736.429
+    },
+    {
+      "epoch": 1.820469798657718,
+      "grad_norm": 0.7595460379816341,
+      "learning_rate": 1.9842129177363555e-07,
+      "loss": 0.2999,
+      "num_input_tokens_seen": 238350464,
+      "step": 4340,
+      "train_runtime": 24485.7502,
+      "train_tokens_per_second": 9734.252
+    },
+    {
+      "epoch": 1.8225671140939599,
+      "grad_norm": 0.6871242844733958,
+      "learning_rate": 1.938529888317975e-07,
+      "loss": 0.3028,
+      "num_input_tokens_seen": 238637976,
+      "step": 4345,
+      "train_runtime": 24516.6544,
+      "train_tokens_per_second": 9733.709
+    },
+    {
+      "epoch": 1.8246644295302015,
+      "grad_norm": 0.7305394549285885,
+      "learning_rate": 1.8933684905308557e-07,
+      "loss": 0.2916,
+      "num_input_tokens_seen": 238910120,
+      "step": 4350,
+      "train_runtime": 24545.157,
+      "train_tokens_per_second": 9733.493
+    },
+    {
+      "epoch": 1.826761744966443,
+      "grad_norm": 0.6725969775888032,
+      "learning_rate": 1.8487292145308978e-07,
+      "loss": 0.2948,
+      "num_input_tokens_seen": 239189688,
+      "step": 4355,
+      "train_runtime": 24582.8526,
+      "train_tokens_per_second": 9729.94
+    },
+    {
+      "epoch": 1.8288590604026846,
+      "grad_norm": 0.8070077258250887,
+      "learning_rate": 1.8046125448071993e-07,
+      "loss": 0.3109,
+      "num_input_tokens_seen": 239463856,
+      "step": 4360,
+      "train_runtime": 24608.3365,
+      "train_tokens_per_second": 9731.005
+    },
+    {
+      "epoch": 1.8309563758389262,
+      "grad_norm": 0.6375621029584307,
+      "learning_rate": 1.7610189601768024e-07,
+      "loss": 0.3041,
+      "num_input_tokens_seen": 239760648,
+      "step": 4365,
+      "train_runtime": 24640.7339,
+      "train_tokens_per_second": 9730.256
+    },
+    {
+      "epoch": 1.8330536912751678,
+      "grad_norm": 0.7575722930327489,
+      "learning_rate": 1.717948933779462e-07,
+      "loss": 0.2951,
+      "num_input_tokens_seen": 240041120,
+      "step": 4370,
+      "train_runtime": 24669.8202,
+      "train_tokens_per_second": 9730.153
+    },
+    {
+      "epoch": 1.8351510067114094,
+      "grad_norm": 0.6474093272065156,
+      "learning_rate": 1.675402933072545e-07,
+      "loss": 0.2953,
+      "num_input_tokens_seen": 240298864,
+      "step": 4375,
+      "train_runtime": 24694.8598,
+      "train_tokens_per_second": 9730.724
+    },
+    {
+      "epoch": 1.837248322147651,
+      "grad_norm": 0.6378641069883187,
+      "learning_rate": 1.6333814198259468e-07,
+      "loss": 0.2913,
+      "num_input_tokens_seen": 240593360,
+      "step": 4380,
+      "train_runtime": 24721.3565,
+      "train_tokens_per_second": 9732.207
+    },
+    {
+      "epoch": 1.8393456375838926,
+      "grad_norm": 0.6750565214933042,
+      "learning_rate": 1.5918848501170647e-07,
+      "loss": 0.2944,
+      "num_input_tokens_seen": 240893920,
+      "step": 4385,
+      "train_runtime": 24745.2698,
+      "train_tokens_per_second": 9734.948
+    },
+    {
+      "epoch": 1.8414429530201342,
+      "grad_norm": 0.9287620627369902,
+      "learning_rate": 1.5509136743258656e-07,
+      "loss": 0.3117,
+      "num_input_tokens_seen": 241152096,
+      "step": 4390,
+      "train_runtime": 24770.1729,
+      "train_tokens_per_second": 9735.584
+    },
+    {
+      "epoch": 1.8435402684563758,
+      "grad_norm": 0.6786501407137806,
+      "learning_rate": 1.5104683371299945e-07,
+      "loss": 0.3135,
+      "num_input_tokens_seen": 241421688,
+      "step": 4395,
+      "train_runtime": 24796.9817,
+      "train_tokens_per_second": 9735.93
+    },
+    {
+      "epoch": 1.8456375838926173,
+      "grad_norm": 0.7631632326238778,
+      "learning_rate": 1.4705492774999385e-07,
+      "loss": 0.2976,
+      "num_input_tokens_seen": 241694320,
+      "step": 4400,
+      "train_runtime": 24825.2237,
+      "train_tokens_per_second": 9735.837
+    },
+    {
+      "epoch": 1.847734899328859,
+      "grad_norm": 0.8157229610458067,
+      "learning_rate": 1.4311569286942883e-07,
+      "loss": 0.2901,
+      "num_input_tokens_seen": 241942608,
+      "step": 4405,
+      "train_runtime": 24850.1354,
+      "train_tokens_per_second": 9736.068
+    },
+    {
+      "epoch": 1.8498322147651005,
+      "grad_norm": 0.7510661688930036,
+      "learning_rate": 1.3922917182549956e-07,
+      "loss": 0.3047,
+      "num_input_tokens_seen": 242205264,
+      "step": 4410,
+      "train_runtime": 24881.7462,
+      "train_tokens_per_second": 9734.255
+    },
+    {
+      "epoch": 1.8519295302013423,
+      "grad_norm": 0.6666573981894589,
+      "learning_rate": 1.3539540680027553e-07,
+      "loss": 0.2966,
+      "num_input_tokens_seen": 242496520,
+      "step": 4415,
+      "train_runtime": 24911.2796,
+      "train_tokens_per_second": 9734.406
+    },
+    {
+      "epoch": 1.854026845637584,
+      "grad_norm": 0.6669031885872662,
+      "learning_rate": 1.3161443940324537e-07,
+      "loss": 0.3058,
+      "num_input_tokens_seen": 242749680,
+      "step": 4420,
+      "train_runtime": 24937.9752,
+      "train_tokens_per_second": 9734.138
+    },
+    {
+      "epoch": 1.8561241610738255,
+      "grad_norm": 0.7824787897720596,
+      "learning_rate": 1.2788631067085944e-07,
+      "loss": 0.3131,
+      "num_input_tokens_seen": 243015024,
+      "step": 4425,
+      "train_runtime": 24966.5088,
+      "train_tokens_per_second": 9733.641
+    },
+    {
+      "epoch": 1.858221476510067,
+      "grad_norm": 0.7578904290837603,
+      "learning_rate": 1.242110610660896e-07,
+      "loss": 0.2874,
+      "num_input_tokens_seen": 243273296,
+      "step": 4430,
+      "train_runtime": 24996.0102,
+      "train_tokens_per_second": 9732.485
+    },
+    {
+      "epoch": 1.8603187919463087,
+      "grad_norm": 0.8508909366296068,
+      "learning_rate": 1.2058873047798625e-07,
+      "loss": 0.2984,
+      "num_input_tokens_seen": 243541800,
+      "step": 4435,
+      "train_runtime": 25025.5284,
+      "train_tokens_per_second": 9731.735
+    },
+    {
+      "epoch": 1.8624161073825505,
+      "grad_norm": 0.6716174720833409,
+      "learning_rate": 1.1701935822124866e-07,
+      "loss": 0.3132,
+      "num_input_tokens_seen": 243837624,
+      "step": 4440,
+      "train_runtime": 25051.9037,
+      "train_tokens_per_second": 9733.297
+    },
+    {
+      "epoch": 1.864513422818792,
+      "grad_norm": 0.6802526996549326,
+      "learning_rate": 1.1350298303579588e-07,
+      "loss": 0.2889,
+      "num_input_tokens_seen": 244129760,
+      "step": 4445,
+      "train_runtime": 25077.4753,
+      "train_tokens_per_second": 9735.021
+    },
+    {
+      "epoch": 1.8666107382550337,
+      "grad_norm": 0.7745970059163191,
+      "learning_rate": 1.1003964308634707e-07,
+      "loss": 0.2857,
+      "num_input_tokens_seen": 244385272,
+      "step": 4450,
+      "train_runtime": 25096.4827,
+      "train_tokens_per_second": 9737.83
+    },
+    {
+      "epoch": 1.8687080536912752,
+      "grad_norm": 0.7547962132295963,
+      "learning_rate": 1.066293759620085e-07,
+      "loss": 0.3229,
+      "num_input_tokens_seen": 244658624,
+      "step": 4455,
+      "train_runtime": 25120.3987,
+      "train_tokens_per_second": 9739.44
+    },
+    {
+      "epoch": 1.8708053691275168,
+      "grad_norm": 0.8246310032030325,
+      "learning_rate": 1.0327221867586224e-07,
+      "loss": 0.2849,
+      "num_input_tokens_seen": 244935528,
+      "step": 4460,
+      "train_runtime": 25146.0527,
+      "train_tokens_per_second": 9740.516
+    },
+    {
+      "epoch": 1.8729026845637584,
+      "grad_norm": 0.6848928889022334,
+      "learning_rate": 9.996820766456916e-08,
+      "loss": 0.2875,
+      "num_input_tokens_seen": 245207328,
+      "step": 4465,
+      "train_runtime": 25175.5363,
+      "train_tokens_per_second": 9739.905
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 0.640731998618335,
+      "learning_rate": 9.671737878796994e-08,
+      "loss": 0.29,
+      "num_input_tokens_seen": 245483776,
+      "step": 4470,
+      "train_runtime": 25206.6645,
+      "train_tokens_per_second": 9738.844
+    },
+    {
+      "epoch": 1.8770973154362416,
+      "grad_norm": 0.6853307668455891,
+      "learning_rate": 9.351976732869527e-08,
+      "loss": 0.2815,
+      "num_input_tokens_seen": 245748192,
+      "step": 4475,
+      "train_runtime": 25232.9484,
+      "train_tokens_per_second": 9739.179
+    },
+    {
+      "epoch": 1.8791946308724832,
+      "grad_norm": 0.8531718921815428,
+      "learning_rate": 9.03754079917879e-08,
+      "loss": 0.2836,
+      "num_input_tokens_seen": 246030184,
+      "step": 4480,
+      "train_runtime": 25256.7428,
+      "train_tokens_per_second": 9741.168
+    },
+    {
+      "epoch": 1.8812919463087248,
+      "grad_norm": 0.6659523835245774,
+      "learning_rate": 8.728433490431897e-08,
+      "loss": 0.304,
+      "num_input_tokens_seen": 246321568,
+      "step": 4485,
+      "train_runtime": 25283.3502,
+      "train_tokens_per_second": 9742.442
+    },
+    {
+      "epoch": 1.8833892617449663,
+      "grad_norm": 0.6442406027414248,
+      "learning_rate": 8.424658161502397e-08,
+      "loss": 0.293,
+      "num_input_tokens_seen": 246618720,
+      "step": 4490,
+      "train_runtime": 25310.402,
+      "train_tokens_per_second": 9743.769
+    },
+    {
+      "epoch": 1.885486577181208,
+      "grad_norm": 0.9630934831955126,
+      "learning_rate": 8.126218109393458e-08,
+      "loss": 0.2857,
+      "num_input_tokens_seen": 246893080,
+      "step": 4495,
+      "train_runtime": 25333.7104,
+      "train_tokens_per_second": 9745.634
+    },
+    {
+      "epoch": 1.8875838926174495,
+      "grad_norm": 0.717208902103637,
+      "learning_rate": 7.833116573202237e-08,
+      "loss": 0.2928,
+      "num_input_tokens_seen": 247150424,
+      "step": 4500,
+      "train_runtime": 25364.6871,
+      "train_tokens_per_second": 9743.878
+    },
+    {
+      "epoch": 1.8896812080536913,
+      "grad_norm": 0.6157240421629911,
+      "learning_rate": 7.54535673408463e-08,
+      "loss": 0.2804,
+      "num_input_tokens_seen": 247432680,
+      "step": 4505,
+      "train_runtime": 25389.5408,
+      "train_tokens_per_second": 9745.457
+    },
+    {
+      "epoch": 1.891778523489933,
+      "grad_norm": 0.7878669606919032,
+      "learning_rate": 7.262941715220961e-08,
+      "loss": 0.2931,
+      "num_input_tokens_seen": 247708592,
+      "step": 4510,
+      "train_runtime": 25411.6195,
+      "train_tokens_per_second": 9747.848
+    },
+    {
+      "epoch": 1.8938758389261745,
+      "grad_norm": 0.7324586700091172,
+      "learning_rate": 6.985874581781848e-08,
+      "loss": 0.3057,
+      "num_input_tokens_seen": 248005616,
+      "step": 4515,
+      "train_runtime": 25439.9571,
+      "train_tokens_per_second": 9748.665
+    },
+    {
+      "epoch": 1.895973154362416,
+      "grad_norm": 0.6939322991027119,
+      "learning_rate": 6.714158340895116e-08,
+      "loss": 0.3061,
+      "num_input_tokens_seen": 248266432,
+      "step": 4520,
+      "train_runtime": 25475.5718,
+      "train_tokens_per_second": 9745.274
+    },
+    {
+      "epoch": 1.8980704697986577,
+      "grad_norm": 0.697066962035294,
+      "learning_rate": 6.447795941612877e-08,
+      "loss": 0.3287,
+      "num_input_tokens_seen": 248547312,
+      "step": 4525,
+      "train_runtime": 25506.9076,
+      "train_tokens_per_second": 9744.314
+    },
+    {
+      "epoch": 1.9001677852348995,
+      "grad_norm": 0.8283364948600845,
+      "learning_rate": 6.186790274879895e-08,
+      "loss": 0.301,
+      "num_input_tokens_seen": 248810984,
+      "step": 4530,
+      "train_runtime": 25540.806,
+      "train_tokens_per_second": 9741.704
+    },
+    {
+      "epoch": 1.902265100671141,
+      "grad_norm": 0.6075188818236129,
+      "learning_rate": 5.9311441735020456e-08,
+      "loss": 0.3033,
+      "num_input_tokens_seen": 249099368,
+      "step": 4535,
+      "train_runtime": 25574.5571,
+      "train_tokens_per_second": 9740.124
+    },
+    {
+      "epoch": 1.9043624161073827,
+      "grad_norm": 0.6611142372964638,
+      "learning_rate": 5.680860412115352e-08,
+      "loss": 0.2812,
+      "num_input_tokens_seen": 249385912,
+      "step": 4540,
+      "train_runtime": 25598.6846,
+      "train_tokens_per_second": 9742.138
+    },
+    {
+      "epoch": 1.9064597315436242,
+      "grad_norm": 0.6959216511933211,
+      "learning_rate": 5.435941707156389e-08,
+      "loss": 0.3058,
+      "num_input_tokens_seen": 249652752,
+      "step": 4545,
+      "train_runtime": 25627.7087,
+      "train_tokens_per_second": 9741.517
+    },
+    {
+      "epoch": 1.9085570469798658,
+      "grad_norm": 0.6506440877701528,
+      "learning_rate": 5.196390716832145e-08,
+      "loss": 0.2889,
+      "num_input_tokens_seen": 249922680,
+      "step": 4550,
+      "train_runtime": 25659.4709,
+      "train_tokens_per_second": 9739.978
+    },
+    {
+      "epoch": 1.9106543624161074,
+      "grad_norm": 0.6846129629743077,
+      "learning_rate": 4.962210041091708e-08,
+      "loss": 0.3085,
+      "num_input_tokens_seen": 250201704,
+      "step": 4555,
+      "train_runtime": 25692.0479,
+      "train_tokens_per_second": 9738.488
+    },
+    {
+      "epoch": 1.912751677852349,
+      "grad_norm": 0.7443524407392804,
+      "learning_rate": 4.733402221597905e-08,
+      "loss": 0.2876,
+      "num_input_tokens_seen": 250450792,
+      "step": 4560,
+      "train_runtime": 25716.9684,
+      "train_tokens_per_second": 9738.737
+    },
+    {
+      "epoch": 1.9148489932885906,
+      "grad_norm": 0.6747352360886211,
+      "learning_rate": 4.509969741699538e-08,
+      "loss": 0.3209,
+      "num_input_tokens_seen": 250741424,
+      "step": 4565,
+      "train_runtime": 25747.0009,
+      "train_tokens_per_second": 9738.665
+    },
+    {
+      "epoch": 1.9169463087248322,
+      "grad_norm": 0.7030835054795233,
+      "learning_rate": 4.29191502640458e-08,
+      "loss": 0.3038,
+      "num_input_tokens_seen": 251008296,
+      "step": 4570,
+      "train_runtime": 25774.5488,
+      "train_tokens_per_second": 9738.611
+    },
+    {
+      "epoch": 1.9190436241610738,
+      "grad_norm": 0.8148696216275789,
+      "learning_rate": 4.07924044235386e-08,
+      "loss": 0.2942,
+      "num_input_tokens_seen": 251284184,
+      "step": 4575,
+      "train_runtime": 25807.6433,
+      "train_tokens_per_second": 9736.813
+    },
+    {
+      "epoch": 1.9211409395973154,
+      "grad_norm": 0.6749873790169337,
+      "learning_rate": 3.8719482977953025e-08,
+      "loss": 0.2732,
+      "num_input_tokens_seen": 251600456,
+      "step": 4580,
+      "train_runtime": 25838.9331,
+      "train_tokens_per_second": 9737.262
+    },
+    {
+      "epoch": 1.923238255033557,
+      "grad_norm": 0.6689379138836348,
+      "learning_rate": 3.670040842558953e-08,
+      "loss": 0.2889,
+      "num_input_tokens_seen": 251888600,
+      "step": 4585,
+      "train_runtime": 25868.7793,
+      "train_tokens_per_second": 9737.166
+    },
+    {
+      "epoch": 1.9253355704697985,
+      "grad_norm": 0.837819812175137,
+      "learning_rate": 3.473520268032604e-08,
+      "loss": 0.2946,
+      "num_input_tokens_seen": 252163880,
+      "step": 4590,
+      "train_runtime": 25894.1118,
+      "train_tokens_per_second": 9738.271
+    },
+    {
+      "epoch": 1.9274328859060401,
+      "grad_norm": 0.7216801460953775,
+      "learning_rate": 3.2823887071377624e-08,
+      "loss": 0.3105,
+      "num_input_tokens_seen": 252449080,
+      "step": 4595,
+      "train_runtime": 25923.8832,
+      "train_tokens_per_second": 9738.089
+    },
+    {
+      "epoch": 1.929530201342282,
+      "grad_norm": 0.7599254665536599,
+      "learning_rate": 3.096648234306832e-08,
+      "loss": 0.2894,
+      "num_input_tokens_seen": 252732240,
+      "step": 4600,
+      "train_runtime": 25947.4723,
+      "train_tokens_per_second": 9740.149
+    },
+    {
+      "epoch": 1.9316275167785235,
+      "grad_norm": 0.7902061094631286,
+      "learning_rate": 2.916300865460353e-08,
+      "loss": 0.2954,
+      "num_input_tokens_seen": 253013368,
+      "step": 4605,
+      "train_runtime": 25978.2177,
+      "train_tokens_per_second": 9739.443
+    },
+    {
+      "epoch": 1.933724832214765,
+      "grad_norm": 0.7480745932070173,
+      "learning_rate": 2.7413485579852462e-08,
+      "loss": 0.2981,
+      "num_input_tokens_seen": 253288968,
+      "step": 4610,
+      "train_runtime": 26009.404,
+      "train_tokens_per_second": 9738.361
+    },
+    {
+      "epoch": 1.9358221476510067,
+      "grad_norm": 0.736207709761847,
+      "learning_rate": 2.571793210713491e-08,
+      "loss": 0.2992,
+      "num_input_tokens_seen": 253540712,
+      "step": 4615,
+      "train_runtime": 26042.5875,
+      "train_tokens_per_second": 9735.619
+    },
+    {
+      "epoch": 1.9379194630872483,
+      "grad_norm": 0.7540654443771758,
+      "learning_rate": 2.4076366639015914e-08,
+      "loss": 0.2988,
+      "num_input_tokens_seen": 253791576,
+      "step": 4620,
+      "train_runtime": 26071.4585,
+      "train_tokens_per_second": 9734.46
+    },
+    {
+      "epoch": 1.94001677852349,
+      "grad_norm": 0.6830441658432953,
+      "learning_rate": 2.2488806992105317e-08,
+      "loss": 0.3034,
+      "num_input_tokens_seen": 254079832,
+      "step": 4625,
+      "train_runtime": 26093.706,
+      "train_tokens_per_second": 9737.208
+    },
+    {
+      "epoch": 1.9421140939597317,
+      "grad_norm": 0.7010981513327568,
+      "learning_rate": 2.095527039686518e-08,
+      "loss": 0.3136,
+      "num_input_tokens_seen": 254365112,
+      "step": 4630,
+      "train_runtime": 26123.8747,
+      "train_tokens_per_second": 9736.883
+    },
+    {
+      "epoch": 1.9442114093959733,
+      "grad_norm": 0.8156562781142562,
+      "learning_rate": 1.9475773497422135e-08,
+      "loss": 0.2784,
+      "num_input_tokens_seen": 254637944,
+      "step": 4635,
+      "train_runtime": 26155.0964,
+      "train_tokens_per_second": 9735.691
+    },
+    {
+      "epoch": 1.9463087248322148,
+      "grad_norm": 0.6394694249931103,
+      "learning_rate": 1.8050332351387535e-08,
+      "loss": 0.2976,
+      "num_input_tokens_seen": 254908256,
+      "step": 4640,
+      "train_runtime": 26180.5072,
+      "train_tokens_per_second": 9736.567
+    },
+    {
+      "epoch": 1.9484060402684564,
+      "grad_norm": 0.7550027281114031,
+      "learning_rate": 1.6678962429680922e-08,
+      "loss": 0.2855,
+      "num_input_tokens_seen": 255170784,
+      "step": 4645,
+      "train_runtime": 26209.7021,
+      "train_tokens_per_second": 9735.738
+    },
+    {
+      "epoch": 1.950503355704698,
+      "grad_norm": 0.7749435865400793,
+      "learning_rate": 1.5361678616365728e-08,
+      "loss": 0.3127,
+      "num_input_tokens_seen": 255436592,
+      "step": 4650,
+      "train_runtime": 26239.8071,
+      "train_tokens_per_second": 9734.698
+    },
+    {
+      "epoch": 1.9526006711409396,
+      "grad_norm": 0.7488112165247042,
+      "learning_rate": 1.4098495208484386e-08,
+      "loss": 0.2863,
+      "num_input_tokens_seen": 255713952,
+      "step": 4655,
+      "train_runtime": 26259.6951,
+      "train_tokens_per_second": 9737.887
+    },
+    {
+      "epoch": 1.9546979865771812,
+      "grad_norm": 0.8017283476362448,
+      "learning_rate": 1.2889425915905696e-08,
+      "loss": 0.2996,
+      "num_input_tokens_seen": 255992128,
+      "step": 4660,
+      "train_runtime": 26288.3771,
+      "train_tokens_per_second": 9737.845
+    },
+    {
+      "epoch": 1.9567953020134228,
+      "grad_norm": 0.6836024010601744,
+      "learning_rate": 1.1734483861173818e-08,
+      "loss": 0.3023,
+      "num_input_tokens_seen": 256272976,
+      "step": 4665,
+      "train_runtime": 26315.122,
+      "train_tokens_per_second": 9738.62
+    },
+    {
+      "epoch": 1.9588926174496644,
+      "grad_norm": 0.762687027798425,
+      "learning_rate": 1.0633681579367838e-08,
+      "loss": 0.305,
+      "num_input_tokens_seen": 256522472,
+      "step": 4670,
+      "train_runtime": 26339.4458,
+      "train_tokens_per_second": 9739.099
+    },
+    {
+      "epoch": 1.960989932885906,
+      "grad_norm": 1.0842944997684845,
+      "learning_rate": 9.5870310179641e-09,
+      "loss": 0.3003,
+      "num_input_tokens_seen": 256832104,
+      "step": 4675,
+      "train_runtime": 26368.4046,
+      "train_tokens_per_second": 9740.146
+    },
+    {
+      "epoch": 1.9630872483221475,
+      "grad_norm": 0.8228607754660361,
+      "learning_rate": 8.594543536707966e-09,
+      "loss": 0.3,
+      "num_input_tokens_seen": 257100736,
+      "step": 4680,
+      "train_runtime": 26389.7892,
+      "train_tokens_per_second": 9742.432
+    },
+    {
+      "epoch": 1.9651845637583891,
+      "grad_norm": 0.6838773722169406,
+      "learning_rate": 7.656229907487822e-09,
+      "loss": 0.2964,
+      "num_input_tokens_seen": 257384400,
+      "step": 4685,
+      "train_runtime": 26416.0665,
+      "train_tokens_per_second": 9743.479
+    },
+    {
+      "epoch": 1.9672818791946307,
+      "grad_norm": 0.8507602931544558,
+      "learning_rate": 6.772100314223484e-09,
+      "loss": 0.2966,
+      "num_input_tokens_seen": 257645120,
+      "step": 4690,
+      "train_runtime": 26447.7396,
+      "train_tokens_per_second": 9741.669
+    },
+    {
+      "epoch": 1.9693791946308725,
+      "grad_norm": 0.7353984602654964,
+      "learning_rate": 5.942164352749635e-09,
+      "loss": 0.3066,
+      "num_input_tokens_seen": 257898192,
+      "step": 4695,
+      "train_runtime": 26473.0395,
+      "train_tokens_per_second": 9741.918
+    },
+    {
+      "epoch": 1.971476510067114,
+      "grad_norm": 0.6948253531405713,
+      "learning_rate": 5.166431030714792e-09,
+      "loss": 0.2923,
+      "num_input_tokens_seen": 258165992,
+      "step": 4700,
+      "train_runtime": 26499.4288,
+      "train_tokens_per_second": 9742.323
+    },
+    {
+      "epoch": 1.9735738255033557,
+      "grad_norm": 0.6772336722091568,
+      "learning_rate": 4.444908767484712e-09,
+      "loss": 0.2852,
+      "num_input_tokens_seen": 258440664,
+      "step": 4705,
+      "train_runtime": 26523.5296,
+      "train_tokens_per_second": 9743.826
+    },
+    {
+      "epoch": 1.9756711409395973,
+      "grad_norm": 0.6496459102335541,
+      "learning_rate": 3.777605394046923e-09,
+      "loss": 0.2968,
+      "num_input_tokens_seen": 258715000,
+      "step": 4710,
+      "train_runtime": 26550.4564,
+      "train_tokens_per_second": 9744.277
+    },
+    {
+      "epoch": 1.977768456375839,
+      "grad_norm": 0.6777245046322888,
+      "learning_rate": 3.164528152930224e-09,
+      "loss": 0.3018,
+      "num_input_tokens_seen": 258986904,
+      "step": 4715,
+      "train_runtime": 26579.6316,
+      "train_tokens_per_second": 9743.811
+    },
+    {
+      "epoch": 1.9798657718120807,
+      "grad_norm": 0.8564604834231629,
+      "learning_rate": 2.605683698123085e-09,
+      "loss": 0.3193,
+      "num_input_tokens_seen": 259250568,
+      "step": 4720,
+      "train_runtime": 26613.3219,
+      "train_tokens_per_second": 9741.383
+    },
+    {
+      "epoch": 1.9819630872483223,
+      "grad_norm": 0.8169185325874124,
+      "learning_rate": 2.1010780950031506e-09,
+      "loss": 0.2988,
+      "num_input_tokens_seen": 259491944,
+      "step": 4725,
+      "train_runtime": 26639.4851,
+      "train_tokens_per_second": 9740.877
+    },
+    {
+      "epoch": 1.9840604026845639,
+      "grad_norm": 1.0230987910772302,
+      "learning_rate": 1.6507168202695157e-09,
+      "loss": 0.3295,
+      "num_input_tokens_seen": 259787424,
+      "step": 4730,
+      "train_runtime": 26666.4777,
+      "train_tokens_per_second": 9742.097
+    },
+    {
+      "epoch": 1.9861577181208054,
+      "grad_norm": 0.7241153679392265,
+      "learning_rate": 1.2546047618861024e-09,
+      "loss": 0.2951,
+      "num_input_tokens_seen": 260078032,
+      "step": 4735,
+      "train_runtime": 26689.3799,
+      "train_tokens_per_second": 9744.626
+    },
+    {
+      "epoch": 1.988255033557047,
+      "grad_norm": 0.681459643528608,
+      "learning_rate": 9.127462190244851e-10,
+      "loss": 0.3121,
+      "num_input_tokens_seen": 260366400,
+      "step": 4740,
+      "train_runtime": 26725.2744,
+      "train_tokens_per_second": 9742.328
+    },
+    {
+      "epoch": 1.9903523489932886,
+      "grad_norm": 0.8197232795188837,
+      "learning_rate": 6.251449020228118e-10,
+      "loss": 0.3056,
+      "num_input_tokens_seen": 260624856,
+      "step": 4745,
+      "train_runtime": 26757.6592,
+      "train_tokens_per_second": 9740.196
+    },
+    {
+      "epoch": 1.9924496644295302,
+      "grad_norm": 0.6582439050789943,
+      "learning_rate": 3.918039323402845e-10,
+      "loss": 0.2822,
+      "num_input_tokens_seen": 260875472,
+      "step": 4750,
+      "train_runtime": 26789.3956,
+      "train_tokens_per_second": 9738.013
+    },
+    {
+      "epoch": 1.9945469798657718,
+      "grad_norm": 0.830226468222832,
+      "learning_rate": 2.1272584252551853e-10,
+      "loss": 0.2859,
+      "num_input_tokens_seen": 261149224,
+      "step": 4755,
+      "train_runtime": 26810.6386,
+      "train_tokens_per_second": 9740.507
+    },
+    {
+      "epoch": 1.9966442953020134,
+      "grad_norm": 0.7133975073378902,
+      "learning_rate": 8.79125761898969e-11,
+      "loss": 0.2997,
+      "num_input_tokens_seen": 261431696,
+      "step": 4760,
+      "train_runtime": 26840.2205,
+      "train_tokens_per_second": 9740.296
+    },
+    {
+      "epoch": 1.998741610738255,
+      "grad_norm": 0.6989194201629791,
+      "learning_rate": 1.7365487984810638e-11,
+      "loss": 0.3174,
+      "num_input_tokens_seen": 261703896,
+      "step": 4765,
+      "train_runtime": 26870.5911,
+      "train_tokens_per_second": 9739.417
+    },
+    {
+      "epoch": 2.0,
+      "num_input_tokens_seen": 261863352,
+      "step": 4768,
+      "total_flos": 484275418791936.0,
+      "train_loss": 0.345425343490987,
+      "train_runtime": 26998.2642,
+      "train_samples_per_second": 0.706,
+      "train_steps_per_second": 0.177
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4768,
+  "num_input_tokens_seen": 261863352,
+  "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": 484275418791936.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}