diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17424 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 8692,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001150483202945237,
+      "grad_norm": 1.766325007533239,
+      "learning_rate": 9.999994774591762e-06,
+      "loss": 0.909,
+      "num_input_tokens_seen": 284352,
+      "step": 5,
+      "train_runtime": 31.8621,
+      "train_tokens_per_second": 8924.467
+    },
+    {
+      "epoch": 0.002300966405890474,
+      "grad_norm": 1.4977177815945257,
+      "learning_rate": 9.999973546389506e-06,
+      "loss": 0.7805,
+      "num_input_tokens_seen": 581176,
+      "step": 10,
+      "train_runtime": 59.5781,
+      "train_tokens_per_second": 9754.864
+    },
+    {
+      "epoch": 0.003451449608835711,
+      "grad_norm": 1.0829632671990435,
+      "learning_rate": 9.999935988874494e-06,
+      "loss": 0.7063,
+      "num_input_tokens_seen": 866224,
+      "step": 15,
+      "train_runtime": 95.5027,
+      "train_tokens_per_second": 9070.149
+    },
+    {
+      "epoch": 0.004601932811780948,
+      "grad_norm": 0.9748199320947489,
+      "learning_rate": 9.999882102169385e-06,
+      "loss": 0.6595,
+      "num_input_tokens_seen": 1158400,
+      "step": 20,
+      "train_runtime": 129.0961,
+      "train_tokens_per_second": 8973.162
+    },
+    {
+      "epoch": 0.005752416014726185,
+      "grad_norm": 1.2093320932662874,
+      "learning_rate": 9.999811886450166e-06,
+      "loss": 0.6575,
+      "num_input_tokens_seen": 1463096,
+      "step": 25,
+      "train_runtime": 150.0408,
+      "train_tokens_per_second": 9751.323
+    },
+    {
+      "epoch": 0.006902899217671422,
+      "grad_norm": 0.9828100334290493,
+      "learning_rate": 9.999725341946154e-06,
+      "loss": 0.658,
+      "num_input_tokens_seen": 1758224,
+      "step": 30,
+      "train_runtime": 176.7733,
+      "train_tokens_per_second": 9946.208
+    },
+    {
+      "epoch": 0.008053382420616659,
+      "grad_norm": 0.9382389824038813,
+      "learning_rate": 9.999622468939991e-06,
+      "loss": 0.6512,
+      "num_input_tokens_seen": 2042552,
+      "step": 35,
+      "train_runtime": 198.0886,
+      "train_tokens_per_second": 10311.306
+    },
+    {
+      "epoch": 0.009203865623561895,
+      "grad_norm": 1.0289259709553886,
+      "learning_rate": 9.999503267767651e-06,
+      "loss": 0.6327,
+      "num_input_tokens_seen": 2347032,
+      "step": 40,
+      "train_runtime": 223.1623,
+      "train_tokens_per_second": 10517.155
+    },
+    {
+      "epoch": 0.010354348826507133,
+      "grad_norm": 1.1816159841971012,
+      "learning_rate": 9.999367738818428e-06,
+      "loss": 0.6656,
+      "num_input_tokens_seen": 2594752,
+      "step": 45,
+      "train_runtime": 251.4412,
+      "train_tokens_per_second": 10319.516
+    },
+    {
+      "epoch": 0.01150483202945237,
+      "grad_norm": 1.0940534462802884,
+      "learning_rate": 9.999215882534944e-06,
+      "loss": 0.6404,
+      "num_input_tokens_seen": 2893800,
+      "step": 50,
+      "train_runtime": 287.4078,
+      "train_tokens_per_second": 10068.619
+    },
+    {
+      "epoch": 0.012655315232397607,
+      "grad_norm": 1.0086989065703342,
+      "learning_rate": 9.999047699413145e-06,
+      "loss": 0.6297,
+      "num_input_tokens_seen": 3210000,
+      "step": 55,
+      "train_runtime": 313.5712,
+      "train_tokens_per_second": 10236.911
+    },
+    {
+      "epoch": 0.013805798435342844,
+      "grad_norm": 1.0251171653500366,
+      "learning_rate": 9.998863190002293e-06,
+      "loss": 0.6326,
+      "num_input_tokens_seen": 3500424,
+      "step": 60,
+      "train_runtime": 347.0606,
+      "train_tokens_per_second": 10085.916
+    },
+    {
+      "epoch": 0.01495628163828808,
+      "grad_norm": 0.9196778163732294,
+      "learning_rate": 9.998662354904978e-06,
+      "loss": 0.6206,
+      "num_input_tokens_seen": 3781008,
+      "step": 65,
+      "train_runtime": 377.0088,
+      "train_tokens_per_second": 10028.965
+    },
+    {
+      "epoch": 0.016106764841233318,
+      "grad_norm": 1.104080574342752,
+      "learning_rate": 9.9984451947771e-06,
+      "loss": 0.5788,
+      "num_input_tokens_seen": 4055520,
+      "step": 70,
+      "train_runtime": 403.5576,
+      "train_tokens_per_second": 10049.421
+    },
+    {
+      "epoch": 0.017257248044178556,
+      "grad_norm": 1.0385273575353133,
+      "learning_rate": 9.99821171032788e-06,
+      "loss": 0.5984,
+      "num_input_tokens_seen": 4364144,
+      "step": 75,
+      "train_runtime": 424.762,
+      "train_tokens_per_second": 10274.328
+    },
+    {
+      "epoch": 0.01840773124712379,
+      "grad_norm": 1.066202873643926,
+      "learning_rate": 9.997961902319848e-06,
+      "loss": 0.6195,
+      "num_input_tokens_seen": 4676760,
+      "step": 80,
+      "train_runtime": 450.8342,
+      "train_tokens_per_second": 10373.569
+    },
+    {
+      "epoch": 0.019558214450069028,
+      "grad_norm": 1.012668713401408,
+      "learning_rate": 9.997695771568848e-06,
+      "loss": 0.595,
+      "num_input_tokens_seen": 4966944,
+      "step": 85,
+      "train_runtime": 475.4493,
+      "train_tokens_per_second": 10446.842
+    },
+    {
+      "epoch": 0.020708697653014266,
+      "grad_norm": 1.193830281082122,
+      "learning_rate": 9.997413318944033e-06,
+      "loss": 0.6066,
+      "num_input_tokens_seen": 5240672,
+      "step": 90,
+      "train_runtime": 502.9857,
+      "train_tokens_per_second": 10419.127
+    },
+    {
+      "epoch": 0.021859180855959504,
+      "grad_norm": 2.0956209535412063,
+      "learning_rate": 9.997114545367856e-06,
+      "loss": 0.5713,
+      "num_input_tokens_seen": 5529784,
+      "step": 95,
+      "train_runtime": 535.9077,
+      "train_tokens_per_second": 10318.538
+    },
+    {
+      "epoch": 0.02300966405890474,
+      "grad_norm": 0.9018337526368458,
+      "learning_rate": 9.996799451816078e-06,
+      "loss": 0.5942,
+      "num_input_tokens_seen": 5817360,
+      "step": 100,
+      "train_runtime": 563.1315,
+      "train_tokens_per_second": 10330.375
+    },
+    {
+      "epoch": 0.024160147261849976,
+      "grad_norm": 1.1880000462108928,
+      "learning_rate": 9.996468039317756e-06,
+      "loss": 0.5891,
+      "num_input_tokens_seen": 6100040,
+      "step": 105,
+      "train_runtime": 589.9606,
+      "train_tokens_per_second": 10339.742
+    },
+    {
+      "epoch": 0.025310630464795213,
+      "grad_norm": 1.0080850299533164,
+      "learning_rate": 9.996120308955243e-06,
+      "loss": 0.5686,
+      "num_input_tokens_seen": 6372616,
+      "step": 110,
+      "train_runtime": 617.0443,
+      "train_tokens_per_second": 10327.648
+    },
+    {
+      "epoch": 0.02646111366774045,
+      "grad_norm": 0.9918805875206164,
+      "learning_rate": 9.995756261864184e-06,
+      "loss": 0.5667,
+      "num_input_tokens_seen": 6662664,
+      "step": 115,
+      "train_runtime": 648.8213,
+      "train_tokens_per_second": 10268.874
+    },
+    {
+      "epoch": 0.02761159687068569,
+      "grad_norm": 1.0576316511920598,
+      "learning_rate": 9.995375899233515e-06,
+      "loss": 0.5739,
+      "num_input_tokens_seen": 6956296,
+      "step": 120,
+      "train_runtime": 667.2231,
+      "train_tokens_per_second": 10425.743
+    },
+    {
+      "epoch": 0.028762080073630927,
+      "grad_norm": 0.9691694911320798,
+      "learning_rate": 9.994979222305453e-06,
+      "loss": 0.5808,
+      "num_input_tokens_seen": 7253584,
+      "step": 125,
+      "train_runtime": 693.7493,
+      "train_tokens_per_second": 10455.627
+    },
+    {
+      "epoch": 0.02991256327657616,
+      "grad_norm": 1.3106007861799989,
+      "learning_rate": 9.994566232375498e-06,
+      "loss": 0.5744,
+      "num_input_tokens_seen": 7525768,
+      "step": 130,
+      "train_runtime": 721.004,
+      "train_tokens_per_second": 10437.901
+    },
+    {
+      "epoch": 0.0310630464795214,
+      "grad_norm": 0.9771512086576776,
+      "learning_rate": 9.994136930792425e-06,
+      "loss": 0.6382,
+      "num_input_tokens_seen": 7805496,
+      "step": 135,
+      "train_runtime": 747.2073,
+      "train_tokens_per_second": 10446.226
+    },
+    {
+      "epoch": 0.032213529682466636,
+      "grad_norm": 1.1108336391930276,
+      "learning_rate": 9.993691318958282e-06,
+      "loss": 0.5925,
+      "num_input_tokens_seen": 8088080,
+      "step": 140,
+      "train_runtime": 769.5601,
+      "train_tokens_per_second": 10510.005
+    },
+    {
+      "epoch": 0.033364012885411874,
+      "grad_norm": 1.3412472988209507,
+      "learning_rate": 9.993229398328382e-06,
+      "loss": 0.5801,
+      "num_input_tokens_seen": 8379744,
+      "step": 145,
+      "train_runtime": 797.736,
+      "train_tokens_per_second": 10504.407
+    },
+    {
+      "epoch": 0.03451449608835711,
+      "grad_norm": 1.1964882303243665,
+      "learning_rate": 9.992751170411306e-06,
+      "loss": 0.5853,
+      "num_input_tokens_seen": 8678464,
+      "step": 150,
+      "train_runtime": 826.7935,
+      "train_tokens_per_second": 10496.531
+    },
+    {
+      "epoch": 0.03566497929130235,
+      "grad_norm": 1.0278764880484794,
+      "learning_rate": 9.992256636768884e-06,
+      "loss": 0.5896,
+      "num_input_tokens_seen": 8982064,
+      "step": 155,
+      "train_runtime": 856.6555,
+      "train_tokens_per_second": 10485.037
+    },
+    {
+      "epoch": 0.03681546249424758,
+      "grad_norm": 1.071254429302389,
+      "learning_rate": 9.991745799016206e-06,
+      "loss": 0.5829,
+      "num_input_tokens_seen": 9262432,
+      "step": 160,
+      "train_runtime": 891.8273,
+      "train_tokens_per_second": 10385.903
+    },
+    {
+      "epoch": 0.03796594569719282,
+      "grad_norm": 1.0691648184325047,
+      "learning_rate": 9.991218658821609e-06,
+      "loss": 0.6314,
+      "num_input_tokens_seen": 9517952,
+      "step": 165,
+      "train_runtime": 917.441,
+      "train_tokens_per_second": 10374.457
+    },
+    {
+      "epoch": 0.039116428900138056,
+      "grad_norm": 0.9273755042376516,
+      "learning_rate": 9.990675217906665e-06,
+      "loss": 0.5316,
+      "num_input_tokens_seen": 9818952,
+      "step": 170,
+      "train_runtime": 948.3611,
+      "train_tokens_per_second": 10353.601
+    },
+    {
+      "epoch": 0.040266912103083294,
+      "grad_norm": 0.9904068681419756,
+      "learning_rate": 9.990115478046189e-06,
+      "loss": 0.5686,
+      "num_input_tokens_seen": 10103144,
+      "step": 175,
+      "train_runtime": 974.2285,
+      "train_tokens_per_second": 10370.405
+    },
+    {
+      "epoch": 0.04141739530602853,
+      "grad_norm": 0.9288190784163819,
+      "learning_rate": 9.989539441068223e-06,
+      "loss": 0.5753,
+      "num_input_tokens_seen": 10404968,
+      "step": 180,
+      "train_runtime": 998.2122,
+      "train_tokens_per_second": 10423.603
+    },
+    {
+      "epoch": 0.04256787850897377,
+      "grad_norm": 1.1200237777337045,
+      "learning_rate": 9.988947108854037e-06,
+      "loss": 0.6008,
+      "num_input_tokens_seen": 10699984,
+      "step": 185,
+      "train_runtime": 1028.4739,
+      "train_tokens_per_second": 10403.749
+    },
+    {
+      "epoch": 0.04371836171191901,
+      "grad_norm": 1.118565286047391,
+      "learning_rate": 9.988338483338116e-06,
+      "loss": 0.5398,
+      "num_input_tokens_seen": 11011088,
+      "step": 190,
+      "train_runtime": 1050.4787,
+      "train_tokens_per_second": 10481.971
+    },
+    {
+      "epoch": 0.044868844914864245,
+      "grad_norm": 0.8817271642109321,
+      "learning_rate": 9.987713566508156e-06,
+      "loss": 0.5305,
+      "num_input_tokens_seen": 11312176,
+      "step": 195,
+      "train_runtime": 1080.8153,
+      "train_tokens_per_second": 10466.336
+    },
+    {
+      "epoch": 0.04601932811780948,
+      "grad_norm": 1.255494172726284,
+      "learning_rate": 9.987072360405061e-06,
+      "loss": 0.5585,
+      "num_input_tokens_seen": 11565944,
+      "step": 200,
+      "train_runtime": 1108.9369,
+      "train_tokens_per_second": 10429.758
+    },
+    {
+      "epoch": 0.04716981132075472,
+      "grad_norm": 0.9772429802936903,
+      "learning_rate": 9.986414867122935e-06,
+      "loss": 0.5766,
+      "num_input_tokens_seen": 11853880,
+      "step": 205,
+      "train_runtime": 1134.9374,
+      "train_tokens_per_second": 10444.523
+    },
+    {
+      "epoch": 0.04832029452369995,
+      "grad_norm": 1.0695339316876415,
+      "learning_rate": 9.98574108880907e-06,
+      "loss": 0.5334,
+      "num_input_tokens_seen": 12116936,
+      "step": 210,
+      "train_runtime": 1161.7404,
+      "train_tokens_per_second": 10429.986
+    },
+    {
+      "epoch": 0.04947077772664519,
+      "grad_norm": 0.9195037498512013,
+      "learning_rate": 9.985051027663945e-06,
+      "loss": 0.5518,
+      "num_input_tokens_seen": 12418528,
+      "step": 215,
+      "train_runtime": 1192.4275,
+      "train_tokens_per_second": 10414.494
+    },
+    {
+      "epoch": 0.05062126092959043,
+      "grad_norm": 1.0662779613209175,
+      "learning_rate": 9.984344685941217e-06,
+      "loss": 0.5567,
+      "num_input_tokens_seen": 12697288,
+      "step": 220,
+      "train_runtime": 1222.9874,
+      "train_tokens_per_second": 10382.191
+    },
+    {
+      "epoch": 0.051771744132535664,
+      "grad_norm": 0.9734149773108488,
+      "learning_rate": 9.983622065947714e-06,
+      "loss": 0.5364,
+      "num_input_tokens_seen": 12984792,
+      "step": 225,
+      "train_runtime": 1251.4028,
+      "train_tokens_per_second": 10376.189
+    },
+    {
+      "epoch": 0.0529222273354809,
+      "grad_norm": 0.8982545540145891,
+      "learning_rate": 9.982883170043427e-06,
+      "loss": 0.5636,
+      "num_input_tokens_seen": 13245176,
+      "step": 230,
+      "train_runtime": 1278.6588,
+      "train_tokens_per_second": 10358.648
+    },
+    {
+      "epoch": 0.05407271053842614,
+      "grad_norm": 0.8897518831605272,
+      "learning_rate": 9.982128000641497e-06,
+      "loss": 0.5721,
+      "num_input_tokens_seen": 13530616,
+      "step": 235,
+      "train_runtime": 1304.7999,
+      "train_tokens_per_second": 10369.878
+    },
+    {
+      "epoch": 0.05522319374137138,
+      "grad_norm": 0.9607187232316375,
+      "learning_rate": 9.981356560208222e-06,
+      "loss": 0.5558,
+      "num_input_tokens_seen": 13825816,
+      "step": 240,
+      "train_runtime": 1330.206,
+      "train_tokens_per_second": 10393.74
+    },
+    {
+      "epoch": 0.056373676944316616,
+      "grad_norm": 0.8575267448085204,
+      "learning_rate": 9.98056885126303e-06,
+      "loss": 0.5254,
+      "num_input_tokens_seen": 14122792,
+      "step": 245,
+      "train_runtime": 1356.6815,
+      "train_tokens_per_second": 10409.807
+    },
+    {
+      "epoch": 0.05752416014726185,
+      "grad_norm": 0.9608955416662621,
+      "learning_rate": 9.979764876378484e-06,
+      "loss": 0.5769,
+      "num_input_tokens_seen": 14423112,
+      "step": 250,
+      "train_runtime": 1390.6529,
+      "train_tokens_per_second": 10371.468
+    },
+    {
+      "epoch": 0.058674643350207084,
+      "grad_norm": 0.9768421398121152,
+      "learning_rate": 9.978944638180271e-06,
+      "loss": 0.5385,
+      "num_input_tokens_seen": 14734024,
+      "step": 255,
+      "train_runtime": 1422.7513,
+      "train_tokens_per_second": 10356.008
+    },
+    {
+      "epoch": 0.05982512655315232,
+      "grad_norm": 1.169476358539382,
+      "learning_rate": 9.97810813934719e-06,
+      "loss": 0.5779,
+      "num_input_tokens_seen": 15016184,
+      "step": 260,
+      "train_runtime": 1452.7422,
+      "train_tokens_per_second": 10336.441
+    },
+    {
+      "epoch": 0.06097560975609756,
+      "grad_norm": 1.196714112519288,
+      "learning_rate": 9.977255382611144e-06,
+      "loss": 0.5521,
+      "num_input_tokens_seen": 15320184,
+      "step": 265,
+      "train_runtime": 1478.8566,
+      "train_tokens_per_second": 10359.479
+    },
+    {
+      "epoch": 0.0621260929590428,
+      "grad_norm": 1.0003269070465257,
+      "learning_rate": 9.976386370757136e-06,
+      "loss": 0.5625,
+      "num_input_tokens_seen": 15626896,
+      "step": 270,
+      "train_runtime": 1506.8821,
+      "train_tokens_per_second": 10370.351
+    },
+    {
+      "epoch": 0.06327657616198804,
+      "grad_norm": 0.7872853414934483,
+      "learning_rate": 9.975501106623253e-06,
+      "loss": 0.5501,
+      "num_input_tokens_seen": 15947392,
+      "step": 275,
+      "train_runtime": 1545.8634,
+      "train_tokens_per_second": 10316.171
+    },
+    {
+      "epoch": 0.06442705936493327,
+      "grad_norm": 1.0130639433922382,
+      "learning_rate": 9.974599593100664e-06,
+      "loss": 0.562,
+      "num_input_tokens_seen": 16241840,
+      "step": 280,
+      "train_runtime": 1577.6089,
+      "train_tokens_per_second": 10295.226
+    },
+    {
+      "epoch": 0.06557754256787851,
+      "grad_norm": 0.9982978729720613,
+      "learning_rate": 9.9736818331336e-06,
+      "loss": 0.5631,
+      "num_input_tokens_seen": 16541464,
+      "step": 285,
+      "train_runtime": 1601.9517,
+      "train_tokens_per_second": 10325.82
+    },
+    {
+      "epoch": 0.06672802577082375,
+      "grad_norm": 1.016015970892931,
+      "learning_rate": 9.972747829719355e-06,
+      "loss": 0.548,
+      "num_input_tokens_seen": 16821552,
+      "step": 290,
+      "train_runtime": 1626.6357,
+      "train_tokens_per_second": 10341.315
+    },
+    {
+      "epoch": 0.06787850897376899,
+      "grad_norm": 0.8613258143359411,
+      "learning_rate": 9.971797585908277e-06,
+      "loss": 0.5364,
+      "num_input_tokens_seen": 17104560,
+      "step": 295,
+      "train_runtime": 1657.4111,
+      "train_tokens_per_second": 10320.047
+    },
+    {
+      "epoch": 0.06902899217671422,
+      "grad_norm": 1.1507039450592036,
+      "learning_rate": 9.970831104803742e-06,
+      "loss": 0.6015,
+      "num_input_tokens_seen": 17368896,
+      "step": 300,
+      "train_runtime": 1686.2216,
+      "train_tokens_per_second": 10300.482
+    },
+    {
+      "epoch": 0.07017947537965946,
+      "grad_norm": 1.086809604655306,
+      "learning_rate": 9.969848389562162e-06,
+      "loss": 0.5456,
+      "num_input_tokens_seen": 17630184,
+      "step": 305,
+      "train_runtime": 1716.0138,
+      "train_tokens_per_second": 10273.917
+    },
+    {
+      "epoch": 0.0713299585826047,
+      "grad_norm": 1.1941199818069477,
+      "learning_rate": 9.968849443392972e-06,
+      "loss": 0.55,
+      "num_input_tokens_seen": 17920776,
+      "step": 310,
+      "train_runtime": 1738.7959,
+      "train_tokens_per_second": 10306.429
+    },
+    {
+      "epoch": 0.07248044178554994,
+      "grad_norm": 0.8336846595195279,
+      "learning_rate": 9.967834269558604e-06,
+      "loss": 0.5234,
+      "num_input_tokens_seen": 18202152,
+      "step": 315,
+      "train_runtime": 1768.6705,
+      "train_tokens_per_second": 10291.432
+    },
+    {
+      "epoch": 0.07363092498849516,
+      "grad_norm": 0.9281786875387816,
+      "learning_rate": 9.966802871374497e-06,
+      "loss": 0.5662,
+      "num_input_tokens_seen": 18465944,
+      "step": 320,
+      "train_runtime": 1790.2205,
+      "train_tokens_per_second": 10314.899
+    },
+    {
+      "epoch": 0.0747814081914404,
+      "grad_norm": 1.0455199298076596,
+      "learning_rate": 9.965755252209073e-06,
+      "loss": 0.5516,
+      "num_input_tokens_seen": 18760320,
+      "step": 325,
+      "train_runtime": 1822.0113,
+      "train_tokens_per_second": 10296.49
+    },
+    {
+      "epoch": 0.07593189139438564,
+      "grad_norm": 0.9289305777270558,
+      "learning_rate": 9.96469141548373e-06,
+      "loss": 0.5709,
+      "num_input_tokens_seen": 19050304,
+      "step": 330,
+      "train_runtime": 1849.3434,
+      "train_tokens_per_second": 10301.118
+    },
+    {
+      "epoch": 0.07708237459733087,
+      "grad_norm": 0.9784627057003694,
+      "learning_rate": 9.963611364672831e-06,
+      "loss": 0.544,
+      "num_input_tokens_seen": 19317816,
+      "step": 335,
+      "train_runtime": 1875.7962,
+      "train_tokens_per_second": 10298.462
+    },
+    {
+      "epoch": 0.07823285780027611,
+      "grad_norm": 6.088806899922095,
+      "learning_rate": 9.962515103303694e-06,
+      "loss": 0.5318,
+      "num_input_tokens_seen": 19588960,
+      "step": 340,
+      "train_runtime": 1904.7419,
+      "train_tokens_per_second": 10284.312
+    },
+    {
+      "epoch": 0.07938334100322135,
+      "grad_norm": 1.1571082102262917,
+      "learning_rate": 9.961402634956575e-06,
+      "loss": 0.518,
+      "num_input_tokens_seen": 19882032,
+      "step": 345,
+      "train_runtime": 1929.4507,
+      "train_tokens_per_second": 10304.504
+    },
+    {
+      "epoch": 0.08053382420616659,
+      "grad_norm": 0.8761677171112716,
+      "learning_rate": 9.96027396326466e-06,
+      "loss": 0.5416,
+      "num_input_tokens_seen": 20166528,
+      "step": 350,
+      "train_runtime": 1960.0909,
+      "train_tokens_per_second": 10288.568
+    },
+    {
+      "epoch": 0.08168430740911183,
+      "grad_norm": 1.0659126887122947,
+      "learning_rate": 9.959129091914058e-06,
+      "loss": 0.5246,
+      "num_input_tokens_seen": 20442152,
+      "step": 355,
+      "train_runtime": 1988.8978,
+      "train_tokens_per_second": 10278.131
+    },
+    {
+      "epoch": 0.08283479061205706,
+      "grad_norm": 1.064296509118394,
+      "learning_rate": 9.957968024643781e-06,
+      "loss": 0.5448,
+      "num_input_tokens_seen": 20752112,
+      "step": 360,
+      "train_runtime": 2014.342,
+      "train_tokens_per_second": 10302.179
+    },
+    {
+      "epoch": 0.0839852738150023,
+      "grad_norm": 0.8506445344617584,
+      "learning_rate": 9.956790765245733e-06,
+      "loss": 0.5319,
+      "num_input_tokens_seen": 21029760,
+      "step": 365,
+      "train_runtime": 2041.1344,
+      "train_tokens_per_second": 10302.977
+    },
+    {
+      "epoch": 0.08513575701794754,
+      "grad_norm": 1.032001373361533,
+      "learning_rate": 9.955597317564705e-06,
+      "loss": 0.5422,
+      "num_input_tokens_seen": 21309112,
+      "step": 370,
+      "train_runtime": 2065.9088,
+      "train_tokens_per_second": 10314.643
+    },
+    {
+      "epoch": 0.08628624022089278,
+      "grad_norm": 0.9840747877199333,
+      "learning_rate": 9.95438768549835e-06,
+      "loss": 0.5619,
+      "num_input_tokens_seen": 21578216,
+      "step": 375,
+      "train_runtime": 2096.2264,
+      "train_tokens_per_second": 10293.839
+    },
+    {
+      "epoch": 0.08743672342383801,
+      "grad_norm": 0.8861730249408858,
+      "learning_rate": 9.953161872997185e-06,
+      "loss": 0.5396,
+      "num_input_tokens_seen": 21839432,
+      "step": 380,
+      "train_runtime": 2123.9678,
+      "train_tokens_per_second": 10282.374
+    },
+    {
+      "epoch": 0.08858720662678325,
+      "grad_norm": 1.1236998152097732,
+      "learning_rate": 9.951919884064564e-06,
+      "loss": 0.5263,
+      "num_input_tokens_seen": 22120760,
+      "step": 385,
+      "train_runtime": 2155.8674,
+      "train_tokens_per_second": 10260.724
+    },
+    {
+      "epoch": 0.08973768982972849,
+      "grad_norm": 0.8211935509162231,
+      "learning_rate": 9.950661722756674e-06,
+      "loss": 0.5293,
+      "num_input_tokens_seen": 22395664,
+      "step": 390,
+      "train_runtime": 2194.2169,
+      "train_tokens_per_second": 10206.677
+    },
+    {
+      "epoch": 0.09088817303267373,
+      "grad_norm": 0.9524622981909562,
+      "learning_rate": 9.949387393182519e-06,
+      "loss": 0.5251,
+      "num_input_tokens_seen": 22667224,
+      "step": 395,
+      "train_runtime": 2227.1816,
+      "train_tokens_per_second": 10177.537
+    },
+    {
+      "epoch": 0.09203865623561897,
+      "grad_norm": 1.0644210310913742,
+      "learning_rate": 9.948096899503905e-06,
+      "loss": 0.5423,
+      "num_input_tokens_seen": 22964224,
+      "step": 400,
+      "train_runtime": 2255.507,
+      "train_tokens_per_second": 10181.402
+    },
+    {
+      "epoch": 0.0931891394385642,
+      "grad_norm": 0.7871290726495593,
+      "learning_rate": 9.946790245935429e-06,
+      "loss": 0.519,
+      "num_input_tokens_seen": 23273560,
+      "step": 405,
+      "train_runtime": 2283.8081,
+      "train_tokens_per_second": 10190.681
+    },
+    {
+      "epoch": 0.09433962264150944,
+      "grad_norm": 1.044026828487096,
+      "learning_rate": 9.945467436744468e-06,
+      "loss": 0.5468,
+      "num_input_tokens_seen": 23528880,
+      "step": 410,
+      "train_runtime": 2316.2188,
+      "train_tokens_per_second": 10158.315
+    },
+    {
+      "epoch": 0.09549010584445466,
+      "grad_norm": 1.0902623094209076,
+      "learning_rate": 9.944128476251154e-06,
+      "loss": 0.5117,
+      "num_input_tokens_seen": 23799000,
+      "step": 415,
+      "train_runtime": 2343.0969,
+      "train_tokens_per_second": 10157.07
+    },
+    {
+      "epoch": 0.0966405890473999,
+      "grad_norm": 1.018862344685691,
+      "learning_rate": 9.942773368828373e-06,
+      "loss": 0.5145,
+      "num_input_tokens_seen": 24069432,
+      "step": 420,
+      "train_runtime": 2372.5444,
+      "train_tokens_per_second": 10144.987
+    },
+    {
+      "epoch": 0.09779107225034514,
+      "grad_norm": 0.8955069553709278,
+      "learning_rate": 9.941402118901743e-06,
+      "loss": 0.528,
+      "num_input_tokens_seen": 24361648,
+      "step": 425,
+      "train_runtime": 2399.7621,
+      "train_tokens_per_second": 10151.693
+    },
+    {
+      "epoch": 0.09894155545329038,
+      "grad_norm": 0.8955055015179952,
+      "learning_rate": 9.9400147309496e-06,
+      "loss": 0.5352,
+      "num_input_tokens_seen": 24625328,
+      "step": 430,
+      "train_runtime": 2430.0546,
+      "train_tokens_per_second": 10133.652
+    },
+    {
+      "epoch": 0.10009203865623562,
+      "grad_norm": 0.8142807480260092,
+      "learning_rate": 9.938611209502988e-06,
+      "loss": 0.5607,
+      "num_input_tokens_seen": 24922168,
+      "step": 435,
+      "train_runtime": 2452.9371,
+      "train_tokens_per_second": 10160.133
+    },
+    {
+      "epoch": 0.10124252185918085,
+      "grad_norm": 0.9583067011721819,
+      "learning_rate": 9.93719155914564e-06,
+      "loss": 0.517,
+      "num_input_tokens_seen": 25199496,
+      "step": 440,
+      "train_runtime": 2481.6913,
+      "train_tokens_per_second": 10154.162
+    },
+    {
+      "epoch": 0.10239300506212609,
+      "grad_norm": 0.877165179724369,
+      "learning_rate": 9.935755784513961e-06,
+      "loss": 0.5467,
+      "num_input_tokens_seen": 25474376,
+      "step": 445,
+      "train_runtime": 2514.3133,
+      "train_tokens_per_second": 10131.743
+    },
+    {
+      "epoch": 0.10354348826507133,
+      "grad_norm": 1.033730847295833,
+      "learning_rate": 9.93430389029702e-06,
+      "loss": 0.5279,
+      "num_input_tokens_seen": 25756816,
+      "step": 450,
+      "train_runtime": 2546.7348,
+      "train_tokens_per_second": 10113.663
+    },
+    {
+      "epoch": 0.10469397146801657,
+      "grad_norm": 0.8807888279332817,
+      "learning_rate": 9.932835881236529e-06,
+      "loss": 0.5182,
+      "num_input_tokens_seen": 26046896,
+      "step": 455,
+      "train_runtime": 2570.0847,
+      "train_tokens_per_second": 10134.645
+    },
+    {
+      "epoch": 0.1058444546709618,
+      "grad_norm": 0.8709703069126551,
+      "learning_rate": 9.93135176212683e-06,
+      "loss": 0.5625,
+      "num_input_tokens_seen": 26355088,
+      "step": 460,
+      "train_runtime": 2594.2446,
+      "train_tokens_per_second": 10159.06
+    },
+    {
+      "epoch": 0.10699493787390704,
+      "grad_norm": 1.0813952755458804,
+      "learning_rate": 9.929851537814871e-06,
+      "loss": 0.545,
+      "num_input_tokens_seen": 26650264,
+      "step": 465,
+      "train_runtime": 2618.5204,
+      "train_tokens_per_second": 10177.604
+    },
+    {
+      "epoch": 0.10814542107685228,
+      "grad_norm": 0.9977898945576259,
+      "learning_rate": 9.928335213200216e-06,
+      "loss": 0.5324,
+      "num_input_tokens_seen": 26954128,
+      "step": 470,
+      "train_runtime": 2642.8792,
+      "train_tokens_per_second": 10198.774
+    },
+    {
+      "epoch": 0.10929590427979752,
+      "grad_norm": 1.006734536570242,
+      "learning_rate": 9.92680279323499e-06,
+      "loss": 0.5482,
+      "num_input_tokens_seen": 27266312,
+      "step": 475,
+      "train_runtime": 2662.9671,
+      "train_tokens_per_second": 10239.072
+    },
+    {
+      "epoch": 0.11044638748274276,
+      "grad_norm": 0.9419359743879657,
+      "learning_rate": 9.925254282923896e-06,
+      "loss": 0.5716,
+      "num_input_tokens_seen": 27560256,
+      "step": 480,
+      "train_runtime": 2691.8246,
+      "train_tokens_per_second": 10238.503
+    },
+    {
+      "epoch": 0.111596870685688,
+      "grad_norm": 0.7732134565976387,
+      "learning_rate": 9.92368968732418e-06,
+      "loss": 0.5479,
+      "num_input_tokens_seen": 27864632,
+      "step": 485,
+      "train_runtime": 2718.8595,
+      "train_tokens_per_second": 10248.647
+    },
+    {
+      "epoch": 0.11274735388863323,
+      "grad_norm": 0.7907335986065439,
+      "learning_rate": 9.922109011545629e-06,
+      "loss": 0.5408,
+      "num_input_tokens_seen": 28152424,
+      "step": 490,
+      "train_runtime": 2750.0189,
+      "train_tokens_per_second": 10237.175
+    },
+    {
+      "epoch": 0.11389783709157847,
+      "grad_norm": 0.9177923239631248,
+      "learning_rate": 9.920512260750533e-06,
+      "loss": 0.5213,
+      "num_input_tokens_seen": 28424168,
+      "step": 495,
+      "train_runtime": 2777.9806,
+      "train_tokens_per_second": 10231.953
+    },
+    {
+      "epoch": 0.1150483202945237,
+      "grad_norm": 0.810363781899224,
+      "learning_rate": 9.918899440153693e-06,
+      "loss": 0.5327,
+      "num_input_tokens_seen": 28728744,
+      "step": 500,
+      "train_runtime": 2811.5996,
+      "train_tokens_per_second": 10217.936
+    },
+    {
+      "epoch": 0.11619880349746893,
+      "grad_norm": 0.8273259840752242,
+      "learning_rate": 9.917270555022384e-06,
+      "loss": 0.5445,
+      "num_input_tokens_seen": 29020200,
+      "step": 505,
+      "train_runtime": 2842.2302,
+      "train_tokens_per_second": 10210.362
+    },
+    {
+      "epoch": 0.11734928670041417,
+      "grad_norm": 0.8743501554990264,
+      "learning_rate": 9.915625610676352e-06,
+      "loss": 0.5142,
+      "num_input_tokens_seen": 29318576,
+      "step": 510,
+      "train_runtime": 2874.7685,
+      "train_tokens_per_second": 10198.587
+    },
+    {
+      "epoch": 0.1184997699033594,
+      "grad_norm": 1.0362492966915384,
+      "learning_rate": 9.913964612487786e-06,
+      "loss": 0.5378,
+      "num_input_tokens_seen": 29605832,
+      "step": 515,
+      "train_runtime": 2897.1651,
+      "train_tokens_per_second": 10218.897
+    },
+    {
+      "epoch": 0.11965025310630464,
+      "grad_norm": 0.9445286539997724,
+      "learning_rate": 9.912287565881309e-06,
+      "loss": 0.5416,
+      "num_input_tokens_seen": 29897040,
+      "step": 520,
+      "train_runtime": 2923.3717,
+      "train_tokens_per_second": 10226.903
+    },
+    {
+      "epoch": 0.12080073630924988,
+      "grad_norm": 1.2184416084495149,
+      "learning_rate": 9.910594476333948e-06,
+      "loss": 0.5139,
+      "num_input_tokens_seen": 30202456,
+      "step": 525,
+      "train_runtime": 2957.9452,
+      "train_tokens_per_second": 10210.62
+    },
+    {
+      "epoch": 0.12195121951219512,
+      "grad_norm": 0.899487961407593,
+      "learning_rate": 9.908885349375134e-06,
+      "loss": 0.5275,
+      "num_input_tokens_seen": 30480312,
+      "step": 530,
+      "train_runtime": 2989.7734,
+      "train_tokens_per_second": 10194.857
+    },
+    {
+      "epoch": 0.12310170271514036,
+      "grad_norm": 0.884663011836274,
+      "learning_rate": 9.907160190586672e-06,
+      "loss": 0.5484,
+      "num_input_tokens_seen": 30753776,
+      "step": 535,
+      "train_runtime": 3017.3488,
+      "train_tokens_per_second": 10192.317
+    },
+    {
+      "epoch": 0.1242521859180856,
+      "grad_norm": 0.8189189114343147,
+      "learning_rate": 9.905419005602722e-06,
+      "loss": 0.5192,
+      "num_input_tokens_seen": 31059160,
+      "step": 540,
+      "train_runtime": 3041.7403,
+      "train_tokens_per_second": 10210.984
+    },
+    {
+      "epoch": 0.12540266912103085,
+      "grad_norm": 0.9536943662035456,
+      "learning_rate": 9.903661800109781e-06,
+      "loss": 0.5218,
+      "num_input_tokens_seen": 31355496,
+      "step": 545,
+      "train_runtime": 3074.1292,
+      "train_tokens_per_second": 10199.798
+    },
+    {
+      "epoch": 0.12655315232397607,
+      "grad_norm": 0.8772730840776594,
+      "learning_rate": 9.901888579846675e-06,
+      "loss": 0.5546,
+      "num_input_tokens_seen": 31639080,
+      "step": 550,
+      "train_runtime": 3099.4811,
+      "train_tokens_per_second": 10207.864
+    },
+    {
+      "epoch": 0.1277036355269213,
+      "grad_norm": 0.7834258454271067,
+      "learning_rate": 9.900099350604526e-06,
+      "loss": 0.5542,
+      "num_input_tokens_seen": 31917480,
+      "step": 555,
+      "train_runtime": 3125.6019,
+      "train_tokens_per_second": 10211.627
+    },
+    {
+      "epoch": 0.12885411872986655,
+      "grad_norm": 0.8703404368890796,
+      "learning_rate": 9.898294118226745e-06,
+      "loss": 0.5174,
+      "num_input_tokens_seen": 32206480,
+      "step": 560,
+      "train_runtime": 3155.9154,
+      "train_tokens_per_second": 10205.115
+    },
+    {
+      "epoch": 0.13000460193281177,
+      "grad_norm": 0.9505220431522421,
+      "learning_rate": 9.896472888609001e-06,
+      "loss": 0.535,
+      "num_input_tokens_seen": 32485928,
+      "step": 565,
+      "train_runtime": 3179.0395,
+      "train_tokens_per_second": 10218.787
+    },
+    {
+      "epoch": 0.13115508513575702,
+      "grad_norm": 0.9252931343781008,
+      "learning_rate": 9.894635667699212e-06,
+      "loss": 0.5378,
+      "num_input_tokens_seen": 32786264,
+      "step": 570,
+      "train_runtime": 3210.3412,
+      "train_tokens_per_second": 10212.704
+    },
+    {
+      "epoch": 0.13230556833870225,
+      "grad_norm": 0.9774445789504773,
+      "learning_rate": 9.892782461497521e-06,
+      "loss": 0.5294,
+      "num_input_tokens_seen": 33065936,
+      "step": 575,
+      "train_runtime": 3237.4802,
+      "train_tokens_per_second": 10213.479
+    },
+    {
+      "epoch": 0.1334560515416475,
+      "grad_norm": 0.8356052100744731,
+      "learning_rate": 9.890913276056277e-06,
+      "loss": 0.5236,
+      "num_input_tokens_seen": 33338616,
+      "step": 580,
+      "train_runtime": 3265.1369,
+      "train_tokens_per_second": 10210.48
+    },
+    {
+      "epoch": 0.13460653474459272,
+      "grad_norm": 0.9076754443737494,
+      "learning_rate": 9.889028117480013e-06,
+      "loss": 0.5284,
+      "num_input_tokens_seen": 33631240,
+      "step": 585,
+      "train_runtime": 3296.2439,
+      "train_tokens_per_second": 10202.898
+    },
+    {
+      "epoch": 0.13575701794753797,
+      "grad_norm": 0.8333609620536444,
+      "learning_rate": 9.887126991925433e-06,
+      "loss": 0.538,
+      "num_input_tokens_seen": 33911824,
+      "step": 590,
+      "train_runtime": 3324.1737,
+      "train_tokens_per_second": 10201.58
+    },
+    {
+      "epoch": 0.1369075011504832,
+      "grad_norm": 0.8067824652891697,
+      "learning_rate": 9.885209905601382e-06,
+      "loss": 0.5064,
+      "num_input_tokens_seen": 34212416,
+      "step": 595,
+      "train_runtime": 3348.7537,
+      "train_tokens_per_second": 10216.462
+    },
+    {
+      "epoch": 0.13805798435342845,
+      "grad_norm": 0.8974033978190282,
+      "learning_rate": 9.883276864768838e-06,
+      "loss": 0.5415,
+      "num_input_tokens_seen": 34512952,
+      "step": 600,
+      "train_runtime": 3378.1774,
+      "train_tokens_per_second": 10216.441
+    },
+    {
+      "epoch": 0.13920846755637367,
+      "grad_norm": 1.0255654720601914,
+      "learning_rate": 9.881327875740876e-06,
+      "loss": 0.5086,
+      "num_input_tokens_seen": 34807488,
+      "step": 605,
+      "train_runtime": 3409.3885,
+      "train_tokens_per_second": 10209.305
+    },
+    {
+      "epoch": 0.14035895075931892,
+      "grad_norm": 0.7811624968111102,
+      "learning_rate": 9.87936294488266e-06,
+      "loss": 0.5249,
+      "num_input_tokens_seen": 35095984,
+      "step": 610,
+      "train_runtime": 3440.6841,
+      "train_tokens_per_second": 10200.293
+    },
+    {
+      "epoch": 0.14150943396226415,
+      "grad_norm": 0.8740189525238817,
+      "learning_rate": 9.877382078611419e-06,
+      "loss": 0.5644,
+      "num_input_tokens_seen": 35401504,
+      "step": 615,
+      "train_runtime": 3470.2347,
+      "train_tokens_per_second": 10201.473
+    },
+    {
+      "epoch": 0.1426599171652094,
+      "grad_norm": 0.8938122077729749,
+      "learning_rate": 9.875385283396425e-06,
+      "loss": 0.5304,
+      "num_input_tokens_seen": 35704080,
+      "step": 620,
+      "train_runtime": 3495.2141,
+      "train_tokens_per_second": 10215.134
+    },
+    {
+      "epoch": 0.14381040036815462,
+      "grad_norm": 0.8385103105631326,
+      "learning_rate": 9.873372565758972e-06,
+      "loss": 0.528,
+      "num_input_tokens_seen": 36001936,
+      "step": 625,
+      "train_runtime": 3526.9131,
+      "train_tokens_per_second": 10207.775
+    },
+    {
+      "epoch": 0.14496088357109987,
+      "grad_norm": 0.8836898449570656,
+      "learning_rate": 9.871343932272352e-06,
+      "loss": 0.5213,
+      "num_input_tokens_seen": 36299992,
+      "step": 630,
+      "train_runtime": 3556.9955,
+      "train_tokens_per_second": 10205.24
+    },
+    {
+      "epoch": 0.1461113667740451,
+      "grad_norm": 0.9090334087818889,
+      "learning_rate": 9.869299389561838e-06,
+      "loss": 0.5187,
+      "num_input_tokens_seen": 36553720,
+      "step": 635,
+      "train_runtime": 3575.1207,
+      "train_tokens_per_second": 10224.472
+    },
+    {
+      "epoch": 0.14726184997699032,
+      "grad_norm": 0.8861339480913603,
+      "learning_rate": 9.86723894430466e-06,
+      "loss": 0.524,
+      "num_input_tokens_seen": 36844720,
+      "step": 640,
+      "train_runtime": 3603.4921,
+      "train_tokens_per_second": 10224.726
+    },
+    {
+      "epoch": 0.14841233317993557,
+      "grad_norm": 0.8709124179479024,
+      "learning_rate": 9.865162603229988e-06,
+      "loss": 0.5376,
+      "num_input_tokens_seen": 37139288,
+      "step": 645,
+      "train_runtime": 3632.867,
+      "train_tokens_per_second": 10223.135
+    },
+    {
+      "epoch": 0.1495628163828808,
+      "grad_norm": 0.9269869447835128,
+      "learning_rate": 9.863070373118901e-06,
+      "loss": 0.5266,
+      "num_input_tokens_seen": 37441960,
+      "step": 650,
+      "train_runtime": 3663.6292,
+      "train_tokens_per_second": 10219.91
+    },
+    {
+      "epoch": 0.15071329958582605,
+      "grad_norm": 0.8977394986419668,
+      "learning_rate": 9.86096226080437e-06,
+      "loss": 0.4886,
+      "num_input_tokens_seen": 37736400,
+      "step": 655,
+      "train_runtime": 3694.4135,
+      "train_tokens_per_second": 10214.449
+    },
+    {
+      "epoch": 0.15186378278877127,
+      "grad_norm": 0.8953968196053297,
+      "learning_rate": 9.858838273171236e-06,
+      "loss": 0.4997,
+      "num_input_tokens_seen": 38003056,
+      "step": 660,
+      "train_runtime": 3717.475,
+      "train_tokens_per_second": 10222.814
+    },
+    {
+      "epoch": 0.15301426599171652,
+      "grad_norm": 0.9417201938445735,
+      "learning_rate": 9.856698417156189e-06,
+      "loss": 0.5168,
+      "num_input_tokens_seen": 38281680,
+      "step": 665,
+      "train_runtime": 3746.9149,
+      "train_tokens_per_second": 10216.853
+    },
+    {
+      "epoch": 0.15416474919466175,
+      "grad_norm": 0.8558144924726236,
+      "learning_rate": 9.854542699747741e-06,
+      "loss": 0.5477,
+      "num_input_tokens_seen": 38588696,
+      "step": 670,
+      "train_runtime": 3769.887,
+      "train_tokens_per_second": 10236.035
+    },
+    {
+      "epoch": 0.155315232397607,
+      "grad_norm": 0.9055644525868062,
+      "learning_rate": 9.852371127986207e-06,
+      "loss": 0.5422,
+      "num_input_tokens_seen": 38864320,
+      "step": 675,
+      "train_runtime": 3795.0019,
+      "train_tokens_per_second": 10240.922
+    },
+    {
+      "epoch": 0.15646571560055222,
+      "grad_norm": 0.8561186286205198,
+      "learning_rate": 9.85018370896368e-06,
+      "loss": 0.5083,
+      "num_input_tokens_seen": 39157776,
+      "step": 680,
+      "train_runtime": 3827.481,
+      "train_tokens_per_second": 10230.691
+    },
+    {
+      "epoch": 0.15761619880349748,
+      "grad_norm": 1.0003098280332372,
+      "learning_rate": 9.847980449824002e-06,
+      "loss": 0.5025,
+      "num_input_tokens_seen": 39441264,
+      "step": 685,
+      "train_runtime": 3851.6759,
+      "train_tokens_per_second": 10240.027
+    },
+    {
+      "epoch": 0.1587666820064427,
+      "grad_norm": 0.9237993043230763,
+      "learning_rate": 9.84576135776276e-06,
+      "loss": 0.5516,
+      "num_input_tokens_seen": 39730552,
+      "step": 690,
+      "train_runtime": 3875.9025,
+      "train_tokens_per_second": 10250.658
+    },
+    {
+      "epoch": 0.15991716520938795,
+      "grad_norm": 0.8953604700742006,
+      "learning_rate": 9.843526440027237e-06,
+      "loss": 0.5412,
+      "num_input_tokens_seen": 40046792,
+      "step": 695,
+      "train_runtime": 3900.099,
+      "train_tokens_per_second": 10268.148
+    },
+    {
+      "epoch": 0.16106764841233318,
+      "grad_norm": 0.9036791611346572,
+      "learning_rate": 9.841275703916409e-06,
+      "loss": 0.5169,
+      "num_input_tokens_seen": 40316776,
+      "step": 700,
+      "train_runtime": 3926.9742,
+      "train_tokens_per_second": 10266.626
+    },
+    {
+      "epoch": 0.16221813161527843,
+      "grad_norm": 0.813871057777017,
+      "learning_rate": 9.839009156780908e-06,
+      "loss": 0.5291,
+      "num_input_tokens_seen": 40619584,
+      "step": 705,
+      "train_runtime": 3949.8139,
+      "train_tokens_per_second": 10283.924
+    },
+    {
+      "epoch": 0.16336861481822365,
+      "grad_norm": 0.8353073497527778,
+      "learning_rate": 9.836726806023005e-06,
+      "loss": 0.5447,
+      "num_input_tokens_seen": 40920192,
+      "step": 710,
+      "train_runtime": 3982.883,
+      "train_tokens_per_second": 10274.013
+    },
+    {
+      "epoch": 0.1645190980211689,
+      "grad_norm": 0.9454716226567895,
+      "learning_rate": 9.834428659096582e-06,
+      "loss": 0.5302,
+      "num_input_tokens_seen": 41194096,
+      "step": 715,
+      "train_runtime": 4012.661,
+      "train_tokens_per_second": 10266.029
+    },
+    {
+      "epoch": 0.16566958122411413,
+      "grad_norm": 0.8758995307236644,
+      "learning_rate": 9.83211472350711e-06,
+      "loss": 0.4963,
+      "num_input_tokens_seen": 41489328,
+      "step": 720,
+      "train_runtime": 4043.1588,
+      "train_tokens_per_second": 10261.612
+    },
+    {
+      "epoch": 0.16682006442705938,
+      "grad_norm": 0.8501625142020435,
+      "learning_rate": 9.829785006811632e-06,
+      "loss": 0.505,
+      "num_input_tokens_seen": 41759152,
+      "step": 725,
+      "train_runtime": 4070.2486,
+      "train_tokens_per_second": 10259.607
+    },
+    {
+      "epoch": 0.1679705476300046,
+      "grad_norm": 0.9784786803716303,
+      "learning_rate": 9.827439516618712e-06,
+      "loss": 0.5556,
+      "num_input_tokens_seen": 42066672,
+      "step": 730,
+      "train_runtime": 4099.3368,
+      "train_tokens_per_second": 10261.824
+    },
+    {
+      "epoch": 0.16912103083294983,
+      "grad_norm": 1.0352071262106552,
+      "learning_rate": 9.825078260588446e-06,
+      "loss": 0.5183,
+      "num_input_tokens_seen": 42361848,
+      "step": 735,
+      "train_runtime": 4133.6866,
+      "train_tokens_per_second": 10247.958
+    },
+    {
+      "epoch": 0.17027151403589508,
+      "grad_norm": 0.8543146076609304,
+      "learning_rate": 9.82270124643241e-06,
+      "loss": 0.5507,
+      "num_input_tokens_seen": 42655600,
+      "step": 740,
+      "train_runtime": 4156.6393,
+      "train_tokens_per_second": 10262.04
+    },
+    {
+      "epoch": 0.1714219972388403,
+      "grad_norm": 1.0039503520971576,
+      "learning_rate": 9.820308481913649e-06,
+      "loss": 0.5031,
+      "num_input_tokens_seen": 42921552,
+      "step": 745,
+      "train_runtime": 4186.6734,
+      "train_tokens_per_second": 10251.947
+    },
+    {
+      "epoch": 0.17257248044178555,
+      "grad_norm": 0.9384092928282539,
+      "learning_rate": 9.817899974846641e-06,
+      "loss": 0.5094,
+      "num_input_tokens_seen": 43228888,
+      "step": 750,
+      "train_runtime": 4225.2728,
+      "train_tokens_per_second": 10231.029
+    },
+    {
+      "epoch": 0.17372296364473078,
+      "grad_norm": 0.9075789591579785,
+      "learning_rate": 9.815475733097283e-06,
+      "loss": 0.5152,
+      "num_input_tokens_seen": 43535888,
+      "step": 755,
+      "train_runtime": 4248.5439,
+      "train_tokens_per_second": 10247.249
+    },
+    {
+      "epoch": 0.17487344684767603,
+      "grad_norm": 0.8747932146029501,
+      "learning_rate": 9.813035764582857e-06,
+      "loss": 0.5133,
+      "num_input_tokens_seen": 43822408,
+      "step": 760,
+      "train_runtime": 4280.4597,
+      "train_tokens_per_second": 10237.781
+    },
+    {
+      "epoch": 0.17602393005062125,
+      "grad_norm": 0.9765326067182066,
+      "learning_rate": 9.810580077272004e-06,
+      "loss": 0.5152,
+      "num_input_tokens_seen": 44115016,
+      "step": 765,
+      "train_runtime": 4312.364,
+      "train_tokens_per_second": 10229.891
+    },
+    {
+      "epoch": 0.1771744132535665,
+      "grad_norm": 1.0733638505898226,
+      "learning_rate": 9.808108679184709e-06,
+      "loss": 0.5207,
+      "num_input_tokens_seen": 44397032,
+      "step": 770,
+      "train_runtime": 4339.1268,
+      "train_tokens_per_second": 10231.79
+    },
+    {
+      "epoch": 0.17832489645651173,
+      "grad_norm": 0.8083369733187006,
+      "learning_rate": 9.805621578392257e-06,
+      "loss": 0.5415,
+      "num_input_tokens_seen": 44710328,
+      "step": 775,
+      "train_runtime": 4375.7826,
+      "train_tokens_per_second": 10217.676
+    },
+    {
+      "epoch": 0.17947537965945698,
+      "grad_norm": 0.8417225825111836,
+      "learning_rate": 9.803118783017221e-06,
+      "loss": 0.4986,
+      "num_input_tokens_seen": 45006608,
+      "step": 780,
+      "train_runtime": 4407.8077,
+      "train_tokens_per_second": 10210.656
+    },
+    {
+      "epoch": 0.1806258628624022,
+      "grad_norm": 0.8385763196344598,
+      "learning_rate": 9.800600301233431e-06,
+      "loss": 0.495,
+      "num_input_tokens_seen": 45284776,
+      "step": 785,
+      "train_runtime": 4433.8225,
+      "train_tokens_per_second": 10213.484
+    },
+    {
+      "epoch": 0.18177634606534745,
+      "grad_norm": 0.8218599997953213,
+      "learning_rate": 9.798066141265946e-06,
+      "loss": 0.5323,
+      "num_input_tokens_seen": 45584648,
+      "step": 790,
+      "train_runtime": 4465.7982,
+      "train_tokens_per_second": 10207.503
+    },
+    {
+      "epoch": 0.18292682926829268,
+      "grad_norm": 0.8777153932212114,
+      "learning_rate": 9.795516311391026e-06,
+      "loss": 0.5139,
+      "num_input_tokens_seen": 45866792,
+      "step": 795,
+      "train_runtime": 4495.3186,
+      "train_tokens_per_second": 10203.235
+    },
+    {
+      "epoch": 0.18407731247123793,
+      "grad_norm": 0.903983214372834,
+      "learning_rate": 9.792950819936113e-06,
+      "loss": 0.497,
+      "num_input_tokens_seen": 46156400,
+      "step": 800,
+      "train_runtime": 4526.8205,
+      "train_tokens_per_second": 10196.207
+    },
+    {
+      "epoch": 0.18522779567418315,
+      "grad_norm": 0.8386904980259033,
+      "learning_rate": 9.790369675279789e-06,
+      "loss": 0.5475,
+      "num_input_tokens_seen": 46450088,
+      "step": 805,
+      "train_runtime": 4553.6351,
+      "train_tokens_per_second": 10200.661
+    },
+    {
+      "epoch": 0.1863782788771284,
+      "grad_norm": 0.8120906167015628,
+      "learning_rate": 9.787772885851766e-06,
+      "loss": 0.521,
+      "num_input_tokens_seen": 46758768,
+      "step": 810,
+      "train_runtime": 4574.4597,
+      "train_tokens_per_second": 10221.703
+    },
+    {
+      "epoch": 0.18752876208007363,
+      "grad_norm": 0.9495163499862753,
+      "learning_rate": 9.785160460132842e-06,
+      "loss": 0.5179,
+      "num_input_tokens_seen": 47041480,
+      "step": 815,
+      "train_runtime": 4597.0774,
+      "train_tokens_per_second": 10232.91
+    },
+    {
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.0457030536457161,
+      "learning_rate": 9.78253240665489e-06,
+      "loss": 0.4789,
+      "num_input_tokens_seen": 47308152,
+      "step": 820,
+      "train_runtime": 4623.9494,
+      "train_tokens_per_second": 10231.114
+    },
+    {
+      "epoch": 0.1898297284859641,
+      "grad_norm": 0.7777588479232288,
+      "learning_rate": 9.779888734000813e-06,
+      "loss": 0.5414,
+      "num_input_tokens_seen": 47600680,
+      "step": 825,
+      "train_runtime": 4652.9921,
+      "train_tokens_per_second": 10230.123
+    },
+    {
+      "epoch": 0.19098021168890933,
+      "grad_norm": 0.8635294725816689,
+      "learning_rate": 9.777229450804532e-06,
+      "loss": 0.5149,
+      "num_input_tokens_seen": 47872360,
+      "step": 830,
+      "train_runtime": 4679.5356,
+      "train_tokens_per_second": 10230.152
+    },
+    {
+      "epoch": 0.19213069489185458,
+      "grad_norm": 0.8403864394084557,
+      "learning_rate": 9.774554565750943e-06,
+      "loss": 0.5192,
+      "num_input_tokens_seen": 48180056,
+      "step": 835,
+      "train_runtime": 4706.6884,
+      "train_tokens_per_second": 10236.508
+    },
+    {
+      "epoch": 0.1932811780947998,
+      "grad_norm": 0.8842858264436628,
+      "learning_rate": 9.771864087575902e-06,
+      "loss": 0.5203,
+      "num_input_tokens_seen": 48489256,
+      "step": 840,
+      "train_runtime": 4732.4801,
+      "train_tokens_per_second": 10246.056
+    },
+    {
+      "epoch": 0.19443166129774506,
+      "grad_norm": 0.7710554546780247,
+      "learning_rate": 9.769158025066185e-06,
+      "loss": 0.5072,
+      "num_input_tokens_seen": 48785776,
+      "step": 845,
+      "train_runtime": 4758.0273,
+      "train_tokens_per_second": 10253.362
+    },
+    {
+      "epoch": 0.19558214450069028,
+      "grad_norm": 0.8398640977877693,
+      "learning_rate": 9.766436387059468e-06,
+      "loss": 0.5038,
+      "num_input_tokens_seen": 49082080,
+      "step": 850,
+      "train_runtime": 4792.375,
+      "train_tokens_per_second": 10241.703
+    },
+    {
+      "epoch": 0.19673262770363553,
+      "grad_norm": 0.7663785701498119,
+      "learning_rate": 9.763699182444293e-06,
+      "loss": 0.491,
+      "num_input_tokens_seen": 49388152,
+      "step": 855,
+      "train_runtime": 4819.6973,
+      "train_tokens_per_second": 10247.148
+    },
+    {
+      "epoch": 0.19788311090658076,
+      "grad_norm": 0.8242304593160233,
+      "learning_rate": 9.760946420160044e-06,
+      "loss": 0.5276,
+      "num_input_tokens_seen": 49690648,
+      "step": 860,
+      "train_runtime": 4847.8573,
+      "train_tokens_per_second": 10250.023
+    },
+    {
+      "epoch": 0.199033594109526,
+      "grad_norm": 0.8246819086040686,
+      "learning_rate": 9.758178109196908e-06,
+      "loss": 0.5219,
+      "num_input_tokens_seen": 49987984,
+      "step": 865,
+      "train_runtime": 4876.7688,
+      "train_tokens_per_second": 10250.226
+    },
+    {
+      "epoch": 0.20018407731247123,
+      "grad_norm": 0.8905648163621304,
+      "learning_rate": 9.755394258595859e-06,
+      "loss": 0.5222,
+      "num_input_tokens_seen": 50279152,
+      "step": 870,
+      "train_runtime": 4906.6248,
+      "train_tokens_per_second": 10247.197
+    },
+    {
+      "epoch": 0.20133456051541648,
+      "grad_norm": 0.7970124383986557,
+      "learning_rate": 9.752594877448615e-06,
+      "loss": 0.5114,
+      "num_input_tokens_seen": 50565776,
+      "step": 875,
+      "train_runtime": 4934.4469,
+      "train_tokens_per_second": 10247.506
+    },
+    {
+      "epoch": 0.2024850437183617,
+      "grad_norm": 0.876254673535774,
+      "learning_rate": 9.749779974897623e-06,
+      "loss": 0.5007,
+      "num_input_tokens_seen": 50822240,
+      "step": 880,
+      "train_runtime": 4958.5462,
+      "train_tokens_per_second": 10249.423
+    },
+    {
+      "epoch": 0.20363552692130696,
+      "grad_norm": 0.9035405546058869,
+      "learning_rate": 9.746949560136012e-06,
+      "loss": 0.513,
+      "num_input_tokens_seen": 51093984,
+      "step": 885,
+      "train_runtime": 4988.8914,
+      "train_tokens_per_second": 10241.551
+    },
+    {
+      "epoch": 0.20478601012425218,
+      "grad_norm": 0.7906855572780273,
+      "learning_rate": 9.74410364240758e-06,
+      "loss": 0.5092,
+      "num_input_tokens_seen": 51377272,
+      "step": 890,
+      "train_runtime": 5013.0008,
+      "train_tokens_per_second": 10248.806
+    },
+    {
+      "epoch": 0.20593649332719743,
+      "grad_norm": 0.9304390403247936,
+      "learning_rate": 9.741242231006751e-06,
+      "loss": 0.496,
+      "num_input_tokens_seen": 51648512,
+      "step": 895,
+      "train_runtime": 5040.2433,
+      "train_tokens_per_second": 10247.226
+    },
+    {
+      "epoch": 0.20708697653014266,
+      "grad_norm": 0.8243492456888373,
+      "learning_rate": 9.738365335278552e-06,
+      "loss": 0.512,
+      "num_input_tokens_seen": 51952232,
+      "step": 900,
+      "train_runtime": 5067.4277,
+      "train_tokens_per_second": 10252.19
+    },
+    {
+      "epoch": 0.2082374597330879,
+      "grad_norm": 0.8071488866748895,
+      "learning_rate": 9.735472964618575e-06,
+      "loss": 0.5318,
+      "num_input_tokens_seen": 52250840,
+      "step": 905,
+      "train_runtime": 5105.1707,
+      "train_tokens_per_second": 10234.886
+    },
+    {
+      "epoch": 0.20938794293603313,
+      "grad_norm": 0.9306052421542447,
+      "learning_rate": 9.732565128472958e-06,
+      "loss": 0.5096,
+      "num_input_tokens_seen": 52552920,
+      "step": 910,
+      "train_runtime": 5134.8221,
+      "train_tokens_per_second": 10234.614
+    },
+    {
+      "epoch": 0.21053842613897836,
+      "grad_norm": 0.9011956671242125,
+      "learning_rate": 9.729641836338344e-06,
+      "loss": 0.5164,
+      "num_input_tokens_seen": 52841360,
+      "step": 915,
+      "train_runtime": 5167.695,
+      "train_tokens_per_second": 10225.325
+    },
+    {
+      "epoch": 0.2116889093419236,
+      "grad_norm": 0.8765779627477555,
+      "learning_rate": 9.726703097761855e-06,
+      "loss": 0.5134,
+      "num_input_tokens_seen": 53147360,
+      "step": 920,
+      "train_runtime": 5200.6744,
+      "train_tokens_per_second": 10219.321
+    },
+    {
+      "epoch": 0.21283939254486883,
+      "grad_norm": 0.8523933956078252,
+      "learning_rate": 9.723748922341055e-06,
+      "loss": 0.5345,
+      "num_input_tokens_seen": 53450248,
+      "step": 925,
+      "train_runtime": 5238.0127,
+      "train_tokens_per_second": 10204.299
+    },
+    {
+      "epoch": 0.21398987574781408,
+      "grad_norm": 0.9082110321071749,
+      "learning_rate": 9.720779319723929e-06,
+      "loss": 0.4984,
+      "num_input_tokens_seen": 53727456,
+      "step": 930,
+      "train_runtime": 5271.5612,
+      "train_tokens_per_second": 10191.944
+    },
+    {
+      "epoch": 0.2151403589507593,
+      "grad_norm": 0.9023296897062206,
+      "learning_rate": 9.71779429960884e-06,
+      "loss": 0.5265,
+      "num_input_tokens_seen": 54006608,
+      "step": 935,
+      "train_runtime": 5297.277,
+      "train_tokens_per_second": 10195.164
+    },
+    {
+      "epoch": 0.21629084215370456,
+      "grad_norm": 0.8597516124510834,
+      "learning_rate": 9.714793871744506e-06,
+      "loss": 0.555,
+      "num_input_tokens_seen": 54301624,
+      "step": 940,
+      "train_runtime": 5326.1929,
+      "train_tokens_per_second": 10195.204
+    },
+    {
+      "epoch": 0.21744132535664978,
+      "grad_norm": 0.9180772349260234,
+      "learning_rate": 9.711778045929962e-06,
+      "loss": 0.5151,
+      "num_input_tokens_seen": 54584696,
+      "step": 945,
+      "train_runtime": 5360.3293,
+      "train_tokens_per_second": 10183.086
+    },
+    {
+      "epoch": 0.21859180855959504,
+      "grad_norm": 0.9422122470794204,
+      "learning_rate": 9.708746832014538e-06,
+      "loss": 0.5222,
+      "num_input_tokens_seen": 54854944,
+      "step": 950,
+      "train_runtime": 5389.5663,
+      "train_tokens_per_second": 10177.989
+    },
+    {
+      "epoch": 0.21974229176254026,
+      "grad_norm": 0.953303974448042,
+      "learning_rate": 9.705700239897809e-06,
+      "loss": 0.5387,
+      "num_input_tokens_seen": 55133272,
+      "step": 955,
+      "train_runtime": 5408.6377,
+      "train_tokens_per_second": 10193.56
+    },
+    {
+      "epoch": 0.2208927749654855,
+      "grad_norm": 0.8391225900579069,
+      "learning_rate": 9.70263827952958e-06,
+      "loss": 0.5258,
+      "num_input_tokens_seen": 55408480,
+      "step": 960,
+      "train_runtime": 5438.1001,
+      "train_tokens_per_second": 10188.941
+    },
+    {
+      "epoch": 0.22204325816843073,
+      "grad_norm": 0.8378747723999114,
+      "learning_rate": 9.699560960909847e-06,
+      "loss": 0.4842,
+      "num_input_tokens_seen": 55712616,
+      "step": 965,
+      "train_runtime": 5465.8541,
+      "train_tokens_per_second": 10192.847
+    },
+    {
+      "epoch": 0.223193741371376,
+      "grad_norm": 0.7583430784357973,
+      "learning_rate": 9.696468294088762e-06,
+      "loss": 0.5104,
+      "num_input_tokens_seen": 55996032,
+      "step": 970,
+      "train_runtime": 5494.3638,
+      "train_tokens_per_second": 10191.541
+    },
+    {
+      "epoch": 0.2243442245743212,
+      "grad_norm": 0.7638650084594644,
+      "learning_rate": 9.693360289166603e-06,
+      "loss": 0.4974,
+      "num_input_tokens_seen": 56275056,
+      "step": 975,
+      "train_runtime": 5525.1207,
+      "train_tokens_per_second": 10185.308
+    },
+    {
+      "epoch": 0.22549470777726646,
+      "grad_norm": 0.8917225764217294,
+      "learning_rate": 9.690236956293742e-06,
+      "loss": 0.5309,
+      "num_input_tokens_seen": 56575056,
+      "step": 980,
+      "train_runtime": 5548.9698,
+      "train_tokens_per_second": 10195.596
+    },
+    {
+      "epoch": 0.22664519098021169,
+      "grad_norm": 0.9089809475355154,
+      "learning_rate": 9.687098305670606e-06,
+      "loss": 0.5082,
+      "num_input_tokens_seen": 56861184,
+      "step": 985,
+      "train_runtime": 5584.3619,
+      "train_tokens_per_second": 10182.217
+    },
+    {
+      "epoch": 0.22779567418315694,
+      "grad_norm": 1.0294435320201858,
+      "learning_rate": 9.683944347547651e-06,
+      "loss": 0.5125,
+      "num_input_tokens_seen": 57132568,
+      "step": 990,
+      "train_runtime": 5615.9652,
+      "train_tokens_per_second": 10173.241
+    },
+    {
+      "epoch": 0.22894615738610216,
+      "grad_norm": 0.8869220439339139,
+      "learning_rate": 9.680775092225329e-06,
+      "loss": 0.5115,
+      "num_input_tokens_seen": 57442632,
+      "step": 995,
+      "train_runtime": 5636.2721,
+      "train_tokens_per_second": 10191.6
+    },
+    {
+      "epoch": 0.2300966405890474,
+      "grad_norm": 0.8208948626700496,
+      "learning_rate": 9.677590550054044e-06,
+      "loss": 0.4957,
+      "num_input_tokens_seen": 57750120,
+      "step": 1000,
+      "train_runtime": 5671.0713,
+      "train_tokens_per_second": 10183.282
+    },
+    {
+      "epoch": 0.23124712379199264,
+      "grad_norm": 0.8498446706379854,
+      "learning_rate": 9.674390731434129e-06,
+      "loss": 0.5187,
+      "num_input_tokens_seen": 58019960,
+      "step": 1005,
+      "train_runtime": 5706.277,
+      "train_tokens_per_second": 10167.743
+    },
+    {
+      "epoch": 0.23239760699493786,
+      "grad_norm": 0.917329755107614,
+      "learning_rate": 9.671175646815807e-06,
+      "loss": 0.5243,
+      "num_input_tokens_seen": 58310160,
+      "step": 1010,
+      "train_runtime": 5739.3451,
+      "train_tokens_per_second": 10159.724
+    },
+    {
+      "epoch": 0.2335480901978831,
+      "grad_norm": 0.9995361917694549,
+      "learning_rate": 9.667945306699157e-06,
+      "loss": 0.5031,
+      "num_input_tokens_seen": 58594520,
+      "step": 1015,
+      "train_runtime": 5765.0524,
+      "train_tokens_per_second": 10163.745
+    },
+    {
+      "epoch": 0.23469857340082834,
+      "grad_norm": 0.9450838087474566,
+      "learning_rate": 9.664699721634081e-06,
+      "loss": 0.5373,
+      "num_input_tokens_seen": 58889256,
+      "step": 1020,
+      "train_runtime": 5790.5916,
+      "train_tokens_per_second": 10169.817
+    },
+    {
+      "epoch": 0.2358490566037736,
+      "grad_norm": 0.8696253974594186,
+      "learning_rate": 9.661438902220274e-06,
+      "loss": 0.5109,
+      "num_input_tokens_seen": 59184904,
+      "step": 1025,
+      "train_runtime": 5816.1682,
+      "train_tokens_per_second": 10175.927
+    },
+    {
+      "epoch": 0.2369995398067188,
+      "grad_norm": 0.9410392122438129,
+      "learning_rate": 9.658162859107176e-06,
+      "loss": 0.5083,
+      "num_input_tokens_seen": 59473864,
+      "step": 1030,
+      "train_runtime": 5845.8966,
+      "train_tokens_per_second": 10173.609
+    },
+    {
+      "epoch": 0.23815002300966406,
+      "grad_norm": 0.81176165739399,
+      "learning_rate": 9.654871602993951e-06,
+      "loss": 0.4989,
+      "num_input_tokens_seen": 59775904,
+      "step": 1035,
+      "train_runtime": 5871.0907,
+      "train_tokens_per_second": 10181.397
+    },
+    {
+      "epoch": 0.2393005062126093,
+      "grad_norm": 0.8876009041790817,
+      "learning_rate": 9.65156514462945e-06,
+      "loss": 0.4969,
+      "num_input_tokens_seen": 60035608,
+      "step": 1040,
+      "train_runtime": 5894.856,
+      "train_tokens_per_second": 10184.406
+    },
+    {
+      "epoch": 0.24045098941555454,
+      "grad_norm": 0.7415157878739326,
+      "learning_rate": 9.648243494812161e-06,
+      "loss": 0.4879,
+      "num_input_tokens_seen": 60313408,
+      "step": 1045,
+      "train_runtime": 5921.5607,
+      "train_tokens_per_second": 10185.39
+    },
+    {
+      "epoch": 0.24160147261849976,
+      "grad_norm": 1.1007584310827623,
+      "learning_rate": 9.6449066643902e-06,
+      "loss": 0.5325,
+      "num_input_tokens_seen": 60614072,
+      "step": 1050,
+      "train_runtime": 5951.3462,
+      "train_tokens_per_second": 10184.935
+    },
+    {
+      "epoch": 0.24275195582144501,
+      "grad_norm": 0.8520567443527186,
+      "learning_rate": 9.64155466426125e-06,
+      "loss": 0.5033,
+      "num_input_tokens_seen": 60897296,
+      "step": 1055,
+      "train_runtime": 5978.761,
+      "train_tokens_per_second": 10185.605
+    },
+    {
+      "epoch": 0.24390243902439024,
+      "grad_norm": 0.9652913304383662,
+      "learning_rate": 9.638187505372545e-06,
+      "loss": 0.5299,
+      "num_input_tokens_seen": 61184168,
+      "step": 1060,
+      "train_runtime": 6011.3929,
+      "train_tokens_per_second": 10178.035
+    },
+    {
+      "epoch": 0.2450529222273355,
+      "grad_norm": 0.9504976025902643,
+      "learning_rate": 9.634805198720816e-06,
+      "loss": 0.5171,
+      "num_input_tokens_seen": 61471080,
+      "step": 1065,
+      "train_runtime": 6038.3975,
+      "train_tokens_per_second": 10180.032
+    },
+    {
+      "epoch": 0.24620340543028071,
+      "grad_norm": 1.0100160975323746,
+      "learning_rate": 9.631407755352273e-06,
+      "loss": 0.548,
+      "num_input_tokens_seen": 61741456,
+      "step": 1070,
+      "train_runtime": 6069.3385,
+      "train_tokens_per_second": 10172.683
+    },
+    {
+      "epoch": 0.24735388863322597,
+      "grad_norm": 0.9694809980612422,
+      "learning_rate": 9.627995186362559e-06,
+      "loss": 0.49,
+      "num_input_tokens_seen": 62039456,
+      "step": 1075,
+      "train_runtime": 6100.1233,
+      "train_tokens_per_second": 10170.197
+    },
+    {
+      "epoch": 0.2485043718361712,
+      "grad_norm": 0.7002793802236155,
+      "learning_rate": 9.62456750289671e-06,
+      "loss": 0.4924,
+      "num_input_tokens_seen": 62340752,
+      "step": 1080,
+      "train_runtime": 6126.3863,
+      "train_tokens_per_second": 10175.779
+    },
+    {
+      "epoch": 0.24965485503911644,
+      "grad_norm": 0.7953195919940435,
+      "learning_rate": 9.621124716149132e-06,
+      "loss": 0.5014,
+      "num_input_tokens_seen": 62622400,
+      "step": 1085,
+      "train_runtime": 6151.2342,
+      "train_tokens_per_second": 10180.461
+    },
+    {
+      "epoch": 0.2508053382420617,
+      "grad_norm": 0.797352016746775,
+      "learning_rate": 9.617666837363553e-06,
+      "loss": 0.5158,
+      "num_input_tokens_seen": 62914896,
+      "step": 1090,
+      "train_runtime": 6172.9843,
+      "train_tokens_per_second": 10191.974
+    },
+    {
+      "epoch": 0.2519558214450069,
+      "grad_norm": 0.859378670646729,
+      "learning_rate": 9.614193877832987e-06,
+      "loss": 0.5109,
+      "num_input_tokens_seen": 63186104,
+      "step": 1095,
+      "train_runtime": 6201.9069,
+      "train_tokens_per_second": 10188.174
+    },
+    {
+      "epoch": 0.25310630464795214,
+      "grad_norm": 0.8708550472429465,
+      "learning_rate": 9.610705848899704e-06,
+      "loss": 0.4905,
+      "num_input_tokens_seen": 63493256,
+      "step": 1100,
+      "train_runtime": 6233.9836,
+      "train_tokens_per_second": 10185.021
+    },
+    {
+      "epoch": 0.2542567878508974,
+      "grad_norm": 0.9005327992965338,
+      "learning_rate": 9.607202761955188e-06,
+      "loss": 0.5108,
+      "num_input_tokens_seen": 63773008,
+      "step": 1105,
+      "train_runtime": 6260.4416,
+      "train_tokens_per_second": 10186.663
+    },
+    {
+      "epoch": 0.2554072710538426,
+      "grad_norm": 0.9265367512656177,
+      "learning_rate": 9.603684628440097e-06,
+      "loss": 0.5185,
+      "num_input_tokens_seen": 64064024,
+      "step": 1110,
+      "train_runtime": 6295.5265,
+      "train_tokens_per_second": 10176.119
+    },
+    {
+      "epoch": 0.25655775425678784,
+      "grad_norm": 0.8603320522241217,
+      "learning_rate": 9.600151459844235e-06,
+      "loss": 0.5275,
+      "num_input_tokens_seen": 64375360,
+      "step": 1115,
+      "train_runtime": 6333.7129,
+      "train_tokens_per_second": 10163.921
+    },
+    {
+      "epoch": 0.2577082374597331,
+      "grad_norm": 0.9250841461316469,
+      "learning_rate": 9.596603267706508e-06,
+      "loss": 0.5247,
+      "num_input_tokens_seen": 64645504,
+      "step": 1120,
+      "train_runtime": 6360.3033,
+      "train_tokens_per_second": 10163.903
+    },
+    {
+      "epoch": 0.25885872066267834,
+      "grad_norm": 0.8766209463748473,
+      "learning_rate": 9.59304006361488e-06,
+      "loss": 0.5161,
+      "num_input_tokens_seen": 64943072,
+      "step": 1125,
+      "train_runtime": 6390.7489,
+      "train_tokens_per_second": 10162.044
+    },
+    {
+      "epoch": 0.26000920386562354,
+      "grad_norm": 0.88538135833435,
+      "learning_rate": 9.589461859206351e-06,
+      "loss": 0.5196,
+      "num_input_tokens_seen": 65238136,
+      "step": 1130,
+      "train_runtime": 6415.0496,
+      "train_tokens_per_second": 10169.545
+    },
+    {
+      "epoch": 0.2611596870685688,
+      "grad_norm": 0.7461850016304836,
+      "learning_rate": 9.585868666166905e-06,
+      "loss": 0.5286,
+      "num_input_tokens_seen": 65548232,
+      "step": 1135,
+      "train_runtime": 6449.7107,
+      "train_tokens_per_second": 10162.972
+    },
+    {
+      "epoch": 0.26231017027151404,
+      "grad_norm": 0.931508003966584,
+      "learning_rate": 9.58226049623148e-06,
+      "loss": 0.4941,
+      "num_input_tokens_seen": 65826696,
+      "step": 1140,
+      "train_runtime": 6470.2105,
+      "train_tokens_per_second": 10173.811
+    },
+    {
+      "epoch": 0.2634606534744593,
+      "grad_norm": 0.8274907347598374,
+      "learning_rate": 9.578637361183922e-06,
+      "loss": 0.5116,
+      "num_input_tokens_seen": 66131288,
+      "step": 1145,
+      "train_runtime": 6500.0416,
+      "train_tokens_per_second": 10173.979
+    },
+    {
+      "epoch": 0.2646111366774045,
+      "grad_norm": 0.8841411663843471,
+      "learning_rate": 9.574999272856959e-06,
+      "loss": 0.5177,
+      "num_input_tokens_seen": 66429936,
+      "step": 1150,
+      "train_runtime": 6535.1099,
+      "train_tokens_per_second": 10165.083
+    },
+    {
+      "epoch": 0.26576161988034974,
+      "grad_norm": 0.8527610434739115,
+      "learning_rate": 9.57134624313215e-06,
+      "loss": 0.4827,
+      "num_input_tokens_seen": 66699896,
+      "step": 1155,
+      "train_runtime": 6561.1957,
+      "train_tokens_per_second": 10165.814
+    },
+    {
+      "epoch": 0.266912103083295,
+      "grad_norm": 0.8602135588586497,
+      "learning_rate": 9.567678283939848e-06,
+      "loss": 0.5392,
+      "num_input_tokens_seen": 66997128,
+      "step": 1160,
+      "train_runtime": 6583.4062,
+      "train_tokens_per_second": 10176.666
+    },
+    {
+      "epoch": 0.26806258628624025,
+      "grad_norm": 0.9020692419619663,
+      "learning_rate": 9.56399540725917e-06,
+      "loss": 0.5012,
+      "num_input_tokens_seen": 67305056,
+      "step": 1165,
+      "train_runtime": 6617.27,
+      "train_tokens_per_second": 10171.121
+    },
+    {
+      "epoch": 0.26921306948918544,
+      "grad_norm": 0.8713386446343888,
+      "learning_rate": 9.560297625117948e-06,
+      "loss": 0.5067,
+      "num_input_tokens_seen": 67610608,
+      "step": 1170,
+      "train_runtime": 6651.1453,
+      "train_tokens_per_second": 10165.258
+    },
+    {
+      "epoch": 0.2703635526921307,
+      "grad_norm": 0.8463862563715958,
+      "learning_rate": 9.55658494959269e-06,
+      "loss": 0.5064,
+      "num_input_tokens_seen": 67916744,
+      "step": 1175,
+      "train_runtime": 6674.8478,
+      "train_tokens_per_second": 10175.025
+    },
+    {
+      "epoch": 0.27151403589507594,
+      "grad_norm": 0.8319534801889948,
+      "learning_rate": 9.552857392808553e-06,
+      "loss": 0.4928,
+      "num_input_tokens_seen": 68218552,
+      "step": 1180,
+      "train_runtime": 6697.3348,
+      "train_tokens_per_second": 10185.925
+    },
+    {
+      "epoch": 0.2726645190980212,
+      "grad_norm": 0.974763759980284,
+      "learning_rate": 9.549114966939288e-06,
+      "loss": 0.5114,
+      "num_input_tokens_seen": 68493456,
+      "step": 1185,
+      "train_runtime": 6721.9398,
+      "train_tokens_per_second": 10189.537
+    },
+    {
+      "epoch": 0.2738150023009664,
+      "grad_norm": 0.8447836618247923,
+      "learning_rate": 9.54535768420721e-06,
+      "loss": 0.5307,
+      "num_input_tokens_seen": 68777712,
+      "step": 1190,
+      "train_runtime": 6748.906,
+      "train_tokens_per_second": 10190.942
+    },
+    {
+      "epoch": 0.27496548550391164,
+      "grad_norm": 0.8623895865635152,
+      "learning_rate": 9.541585556883149e-06,
+      "loss": 0.4903,
+      "num_input_tokens_seen": 69061728,
+      "step": 1195,
+      "train_runtime": 6775.7991,
+      "train_tokens_per_second": 10192.411
+    },
+    {
+      "epoch": 0.2761159687068569,
+      "grad_norm": 0.9666677082405928,
+      "learning_rate": 9.537798597286423e-06,
+      "loss": 0.5247,
+      "num_input_tokens_seen": 69349600,
+      "step": 1200,
+      "train_runtime": 6801.9955,
+      "train_tokens_per_second": 10195.479
+    },
+    {
+      "epoch": 0.2772664519098021,
+      "grad_norm": 0.8385156495292243,
+      "learning_rate": 9.533996817784786e-06,
+      "loss": 0.488,
+      "num_input_tokens_seen": 69642376,
+      "step": 1205,
+      "train_runtime": 6830.7269,
+      "train_tokens_per_second": 10195.456
+    },
+    {
+      "epoch": 0.27841693511274734,
+      "grad_norm": 0.8594888066180233,
+      "learning_rate": 9.53018023079439e-06,
+      "loss": 0.5096,
+      "num_input_tokens_seen": 69911424,
+      "step": 1210,
+      "train_runtime": 6852.8952,
+      "train_tokens_per_second": 10201.735
+    },
+    {
+      "epoch": 0.2795674183156926,
+      "grad_norm": 0.8910058057694898,
+      "learning_rate": 9.526348848779755e-06,
+      "loss": 0.511,
+      "num_input_tokens_seen": 70189280,
+      "step": 1215,
+      "train_runtime": 6874.8196,
+      "train_tokens_per_second": 10209.618
+    },
+    {
+      "epoch": 0.28071790151863785,
+      "grad_norm": 0.7715503686160624,
+      "learning_rate": 9.522502684253709e-06,
+      "loss": 0.5007,
+      "num_input_tokens_seen": 70479744,
+      "step": 1220,
+      "train_runtime": 6901.2635,
+      "train_tokens_per_second": 10212.586
+    },
+    {
+      "epoch": 0.28186838472158304,
+      "grad_norm": 1.0238917901671047,
+      "learning_rate": 9.518641749777366e-06,
+      "loss": 0.5132,
+      "num_input_tokens_seen": 70780752,
+      "step": 1225,
+      "train_runtime": 6929.8774,
+      "train_tokens_per_second": 10213.853
+    },
+    {
+      "epoch": 0.2830188679245283,
+      "grad_norm": 0.8070415695934842,
+      "learning_rate": 9.514766057960074e-06,
+      "loss": 0.4982,
+      "num_input_tokens_seen": 71075608,
+      "step": 1230,
+      "train_runtime": 6956.4593,
+      "train_tokens_per_second": 10217.21
+    },
+    {
+      "epoch": 0.28416935112747355,
+      "grad_norm": 0.8421623711775764,
+      "learning_rate": 9.510875621459377e-06,
+      "loss": 0.5176,
+      "num_input_tokens_seen": 71347528,
+      "step": 1235,
+      "train_runtime": 6986.7735,
+      "train_tokens_per_second": 10211.799
+    },
+    {
+      "epoch": 0.2853198343304188,
+      "grad_norm": 0.9083619177816767,
+      "learning_rate": 9.50697045298097e-06,
+      "loss": 0.5153,
+      "num_input_tokens_seen": 71639624,
+      "step": 1240,
+      "train_runtime": 7010.6472,
+      "train_tokens_per_second": 10218.689
+    },
+    {
+      "epoch": 0.286470317533364,
+      "grad_norm": 0.7679679372305114,
+      "learning_rate": 9.50305056527867e-06,
+      "loss": 0.521,
+      "num_input_tokens_seen": 71923168,
+      "step": 1245,
+      "train_runtime": 7035.8837,
+      "train_tokens_per_second": 10222.336
+    },
+    {
+      "epoch": 0.28762080073630925,
+      "grad_norm": 0.8710459658578291,
+      "learning_rate": 9.499115971154353e-06,
+      "loss": 0.5112,
+      "num_input_tokens_seen": 72200456,
+      "step": 1250,
+      "train_runtime": 7063.5343,
+      "train_tokens_per_second": 10221.576
+    },
+    {
+      "epoch": 0.2887712839392545,
+      "grad_norm": 0.7943813551736619,
+      "learning_rate": 9.495166683457933e-06,
+      "loss": 0.5016,
+      "num_input_tokens_seen": 72491568,
+      "step": 1255,
+      "train_runtime": 7089.8361,
+      "train_tokens_per_second": 10224.717
+    },
+    {
+      "epoch": 0.28992176714219975,
+      "grad_norm": 0.8797699127178307,
+      "learning_rate": 9.491202715087308e-06,
+      "loss": 0.49,
+      "num_input_tokens_seen": 72777968,
+      "step": 1260,
+      "train_runtime": 7121.428,
+      "train_tokens_per_second": 10219.575
+    },
+    {
+      "epoch": 0.29107225034514494,
+      "grad_norm": 0.9023865497885425,
+      "learning_rate": 9.487224078988326e-06,
+      "loss": 0.501,
+      "num_input_tokens_seen": 73074264,
+      "step": 1265,
+      "train_runtime": 7147.6328,
+      "train_tokens_per_second": 10223.562
+    },
+    {
+      "epoch": 0.2922227335480902,
+      "grad_norm": 0.9578682270749407,
+      "learning_rate": 9.483230788154731e-06,
+      "loss": 0.4805,
+      "num_input_tokens_seen": 73360608,
+      "step": 1270,
+      "train_runtime": 7171.8049,
+      "train_tokens_per_second": 10229.03
+    },
+    {
+      "epoch": 0.29337321675103545,
+      "grad_norm": 1.1821370098151063,
+      "learning_rate": 9.479222855628132e-06,
+      "loss": 0.4987,
+      "num_input_tokens_seen": 73655784,
+      "step": 1275,
+      "train_runtime": 7196.1609,
+      "train_tokens_per_second": 10235.428
+    },
+    {
+      "epoch": 0.29452369995398064,
+      "grad_norm": 0.7981758625712299,
+      "learning_rate": 9.475200294497954e-06,
+      "loss": 0.5319,
+      "num_input_tokens_seen": 73946608,
+      "step": 1280,
+      "train_runtime": 7226.2695,
+      "train_tokens_per_second": 10233.027
+    },
+    {
+      "epoch": 0.2956741831569259,
+      "grad_norm": 0.8323507226728601,
+      "learning_rate": 9.471163117901398e-06,
+      "loss": 0.551,
+      "num_input_tokens_seen": 74239448,
+      "step": 1285,
+      "train_runtime": 7262.1316,
+      "train_tokens_per_second": 10222.818
+    },
+    {
+      "epoch": 0.29682466635987115,
+      "grad_norm": 0.8479855362711718,
+      "learning_rate": 9.4671113390234e-06,
+      "loss": 0.4762,
+      "num_input_tokens_seen": 74514256,
+      "step": 1290,
+      "train_runtime": 7291.6135,
+      "train_tokens_per_second": 10219.172
+    },
+    {
+      "epoch": 0.2979751495628164,
+      "grad_norm": 0.7445067560048461,
+      "learning_rate": 9.463044971096582e-06,
+      "loss": 0.5103,
+      "num_input_tokens_seen": 74815944,
+      "step": 1295,
+      "train_runtime": 7324.7247,
+      "train_tokens_per_second": 10214.165
+    },
+    {
+      "epoch": 0.2991256327657616,
+      "grad_norm": 0.9822358688412866,
+      "learning_rate": 9.458964027401213e-06,
+      "loss": 0.4826,
+      "num_input_tokens_seen": 75118400,
+      "step": 1300,
+      "train_runtime": 7356.0441,
+      "train_tokens_per_second": 10211.793
+    },
+    {
+      "epoch": 0.30027611596870685,
+      "grad_norm": 0.8949844105848246,
+      "learning_rate": 9.454868521265164e-06,
+      "loss": 0.4906,
+      "num_input_tokens_seen": 75399040,
+      "step": 1305,
+      "train_runtime": 7379.1711,
+      "train_tokens_per_second": 10217.82
+    },
+    {
+      "epoch": 0.3014265991716521,
+      "grad_norm": 0.7788854371754802,
+      "learning_rate": 9.450758466063865e-06,
+      "loss": 0.5007,
+      "num_input_tokens_seen": 75709464,
+      "step": 1310,
+      "train_runtime": 7407.8573,
+      "train_tokens_per_second": 10220.157
+    },
+    {
+      "epoch": 0.30257708237459735,
+      "grad_norm": 0.810020708546766,
+      "learning_rate": 9.446633875220265e-06,
+      "loss": 0.5176,
+      "num_input_tokens_seen": 75994304,
+      "step": 1315,
+      "train_runtime": 7435.1491,
+      "train_tokens_per_second": 10220.952
+    },
+    {
+      "epoch": 0.30372756557754255,
+      "grad_norm": 0.7418404404260861,
+      "learning_rate": 9.44249476220478e-06,
+      "loss": 0.4694,
+      "num_input_tokens_seen": 76278112,
+      "step": 1320,
+      "train_runtime": 7468.3411,
+      "train_tokens_per_second": 10213.528
+    },
+    {
+      "epoch": 0.3048780487804878,
+      "grad_norm": 1.0209783284998575,
+      "learning_rate": 9.438341140535256e-06,
+      "loss": 0.4961,
+      "num_input_tokens_seen": 76550608,
+      "step": 1325,
+      "train_runtime": 7499.1978,
+      "train_tokens_per_second": 10207.84
+    },
+    {
+      "epoch": 0.30602853198343305,
+      "grad_norm": 0.8161232599970687,
+      "learning_rate": 9.434173023776923e-06,
+      "loss": 0.5075,
+      "num_input_tokens_seen": 76839112,
+      "step": 1330,
+      "train_runtime": 7534.2379,
+      "train_tokens_per_second": 10198.658
+    },
+    {
+      "epoch": 0.3071790151863783,
+      "grad_norm": 0.7985224836834903,
+      "learning_rate": 9.429990425542348e-06,
+      "loss": 0.5097,
+      "num_input_tokens_seen": 77138224,
+      "step": 1335,
+      "train_runtime": 7566.9649,
+      "train_tokens_per_second": 10194.077
+    },
+    {
+      "epoch": 0.3083294983893235,
+      "grad_norm": 0.8273254424014069,
+      "learning_rate": 9.425793359491396e-06,
+      "loss": 0.4911,
+      "num_input_tokens_seen": 77412552,
+      "step": 1340,
+      "train_runtime": 7591.6794,
+      "train_tokens_per_second": 10197.026
+    },
+    {
+      "epoch": 0.30947998159226875,
+      "grad_norm": 0.985494868290556,
+      "learning_rate": 9.42158183933118e-06,
+      "loss": 0.4665,
+      "num_input_tokens_seen": 77674592,
+      "step": 1345,
+      "train_runtime": 7620.5669,
+      "train_tokens_per_second": 10192.758
+    },
+    {
+      "epoch": 0.310630464795214,
+      "grad_norm": 0.8903968768493296,
+      "learning_rate": 9.417355878816016e-06,
+      "loss": 0.5213,
+      "num_input_tokens_seen": 77954544,
+      "step": 1350,
+      "train_runtime": 7644.747,
+      "train_tokens_per_second": 10197.138
+    },
+    {
+      "epoch": 0.31178094799815925,
+      "grad_norm": 1.0042328913471668,
+      "learning_rate": 9.413115491747389e-06,
+      "loss": 0.5066,
+      "num_input_tokens_seen": 78224344,
+      "step": 1355,
+      "train_runtime": 7671.598,
+      "train_tokens_per_second": 10196.617
+    },
+    {
+      "epoch": 0.31293143120110445,
+      "grad_norm": 0.8962245250602733,
+      "learning_rate": 9.40886069197389e-06,
+      "loss": 0.4957,
+      "num_input_tokens_seen": 78475912,
+      "step": 1360,
+      "train_runtime": 7698.5203,
+      "train_tokens_per_second": 10193.636
+    },
+    {
+      "epoch": 0.3140819144040497,
+      "grad_norm": 0.8126134467342647,
+      "learning_rate": 9.404591493391181e-06,
+      "loss": 0.534,
+      "num_input_tokens_seen": 78779632,
+      "step": 1365,
+      "train_runtime": 7726.1117,
+      "train_tokens_per_second": 10196.543
+    },
+    {
+      "epoch": 0.31523239760699495,
+      "grad_norm": 0.903215943787808,
+      "learning_rate": 9.400307909941955e-06,
+      "loss": 0.5328,
+      "num_input_tokens_seen": 79092424,
+      "step": 1370,
+      "train_runtime": 7753.6621,
+      "train_tokens_per_second": 10200.654
+    },
+    {
+      "epoch": 0.31638288080994015,
+      "grad_norm": 0.8146451223637345,
+      "learning_rate": 9.396009955615882e-06,
+      "loss": 0.5079,
+      "num_input_tokens_seen": 79375896,
+      "step": 1375,
+      "train_runtime": 7777.7158,
+      "train_tokens_per_second": 10205.554
+    },
+    {
+      "epoch": 0.3175333640128854,
+      "grad_norm": 0.7959786615282506,
+      "learning_rate": 9.391697644449562e-06,
+      "loss": 0.4942,
+      "num_input_tokens_seen": 79665592,
+      "step": 1380,
+      "train_runtime": 7812.8177,
+      "train_tokens_per_second": 10196.781
+    },
+    {
+      "epoch": 0.31868384721583065,
+      "grad_norm": 0.8073568819185224,
+      "learning_rate": 9.387370990526485e-06,
+      "loss": 0.5285,
+      "num_input_tokens_seen": 79951400,
+      "step": 1385,
+      "train_runtime": 7844.4306,
+      "train_tokens_per_second": 10192.123
+    },
+    {
+      "epoch": 0.3198343304187759,
+      "grad_norm": 0.7678738542714731,
+      "learning_rate": 9.383030007976984e-06,
+      "loss": 0.501,
+      "num_input_tokens_seen": 80245312,
+      "step": 1390,
+      "train_runtime": 7881.4929,
+      "train_tokens_per_second": 10181.486
+    },
+    {
+      "epoch": 0.3209848136217211,
+      "grad_norm": 0.7809024100447964,
+      "learning_rate": 9.378674710978185e-06,
+      "loss": 0.5229,
+      "num_input_tokens_seen": 80516736,
+      "step": 1395,
+      "train_runtime": 7906.5282,
+      "train_tokens_per_second": 10183.577
+    },
+    {
+      "epoch": 0.32213529682466635,
+      "grad_norm": 0.8420742985273151,
+      "learning_rate": 9.374305113753965e-06,
+      "loss": 0.4861,
+      "num_input_tokens_seen": 80815536,
+      "step": 1400,
+      "train_runtime": 7933.9969,
+      "train_tokens_per_second": 10185.98
+    },
+    {
+      "epoch": 0.3232857800276116,
+      "grad_norm": 0.8209967109663324,
+      "learning_rate": 9.369921230574905e-06,
+      "loss": 0.5071,
+      "num_input_tokens_seen": 81110840,
+      "step": 1405,
+      "train_runtime": 7970.1363,
+      "train_tokens_per_second": 10176.845
+    },
+    {
+      "epoch": 0.32443626323055685,
+      "grad_norm": 0.9475093315328245,
+      "learning_rate": 9.365523075758239e-06,
+      "loss": 0.4732,
+      "num_input_tokens_seen": 81384464,
+      "step": 1410,
+      "train_runtime": 8004.037,
+      "train_tokens_per_second": 10167.927
+    },
+    {
+      "epoch": 0.32558674643350205,
+      "grad_norm": 0.7683124077862844,
+      "learning_rate": 9.361110663667812e-06,
+      "loss": 0.4698,
+      "num_input_tokens_seen": 81676592,
+      "step": 1415,
+      "train_runtime": 8034.1733,
+      "train_tokens_per_second": 10166.148
+    },
+    {
+      "epoch": 0.3267372296364473,
+      "grad_norm": 0.8627919657643328,
+      "learning_rate": 9.356684008714034e-06,
+      "loss": 0.5167,
+      "num_input_tokens_seen": 81975800,
+      "step": 1420,
+      "train_runtime": 8069.0752,
+      "train_tokens_per_second": 10159.256
+    },
+    {
+      "epoch": 0.32788771283939255,
+      "grad_norm": 0.8029788348035442,
+      "learning_rate": 9.352243125353825e-06,
+      "loss": 0.5102,
+      "num_input_tokens_seen": 82285192,
+      "step": 1425,
+      "train_runtime": 8094.1338,
+      "train_tokens_per_second": 10166.028
+    },
+    {
+      "epoch": 0.3290381960423378,
+      "grad_norm": 0.8984349588208196,
+      "learning_rate": 9.347788028090578e-06,
+      "loss": 0.5067,
+      "num_input_tokens_seen": 82573320,
+      "step": 1430,
+      "train_runtime": 8121.438,
+      "train_tokens_per_second": 10167.327
+    },
+    {
+      "epoch": 0.330188679245283,
+      "grad_norm": 0.9886272850459229,
+      "learning_rate": 9.343318731474104e-06,
+      "loss": 0.4906,
+      "num_input_tokens_seen": 82858592,
+      "step": 1435,
+      "train_runtime": 8150.6776,
+      "train_tokens_per_second": 10165.853
+    },
+    {
+      "epoch": 0.33133916244822825,
+      "grad_norm": 0.7998620321166807,
+      "learning_rate": 9.338835250100591e-06,
+      "loss": 0.4973,
+      "num_input_tokens_seen": 83159248,
+      "step": 1440,
+      "train_runtime": 8180.0384,
+      "train_tokens_per_second": 10166.119
+    },
+    {
+      "epoch": 0.3324896456511735,
+      "grad_norm": 0.9083337869822152,
+      "learning_rate": 9.334337598612549e-06,
+      "loss": 0.5106,
+      "num_input_tokens_seen": 83449904,
+      "step": 1445,
+      "train_runtime": 8203.6067,
+      "train_tokens_per_second": 10172.343
+    },
+    {
+      "epoch": 0.33364012885411876,
+      "grad_norm": 0.9205428401854006,
+      "learning_rate": 9.329825791698767e-06,
+      "loss": 0.5079,
+      "num_input_tokens_seen": 83742368,
+      "step": 1450,
+      "train_runtime": 8229.1336,
+      "train_tokens_per_second": 10176.329
+    },
+    {
+      "epoch": 0.33479061205706395,
+      "grad_norm": 0.7806542399066583,
+      "learning_rate": 9.325299844094264e-06,
+      "loss": 0.4849,
+      "num_input_tokens_seen": 84023864,
+      "step": 1455,
+      "train_runtime": 8256.7496,
+      "train_tokens_per_second": 10176.385
+    },
+    {
+      "epoch": 0.3359410952600092,
+      "grad_norm": 0.8379461550430989,
+      "learning_rate": 9.320759770580241e-06,
+      "loss": 0.5186,
+      "num_input_tokens_seen": 84310600,
+      "step": 1460,
+      "train_runtime": 8284.153,
+      "train_tokens_per_second": 10177.335
+    },
+    {
+      "epoch": 0.33709157846295446,
+      "grad_norm": 0.8881990242480046,
+      "learning_rate": 9.316205585984035e-06,
+      "loss": 0.4651,
+      "num_input_tokens_seen": 84576528,
+      "step": 1465,
+      "train_runtime": 8315.4582,
+      "train_tokens_per_second": 10171.0
+    },
+    {
+      "epoch": 0.33824206166589965,
+      "grad_norm": 0.8654069585450914,
+      "learning_rate": 9.311637305179064e-06,
+      "loss": 0.4783,
+      "num_input_tokens_seen": 84857120,
+      "step": 1470,
+      "train_runtime": 8344.222,
+      "train_tokens_per_second": 10169.566
+    },
+    {
+      "epoch": 0.3393925448688449,
+      "grad_norm": 0.8354479678978177,
+      "learning_rate": 9.307054943084783e-06,
+      "loss": 0.4771,
+      "num_input_tokens_seen": 85110176,
+      "step": 1475,
+      "train_runtime": 8372.3411,
+      "train_tokens_per_second": 10165.636
+    },
+    {
+      "epoch": 0.34054302807179015,
+      "grad_norm": 0.7623020912089269,
+      "learning_rate": 9.302458514666634e-06,
+      "loss": 0.4976,
+      "num_input_tokens_seen": 85418624,
+      "step": 1480,
+      "train_runtime": 8405.2204,
+      "train_tokens_per_second": 10162.568
+    },
+    {
+      "epoch": 0.3416935112747354,
+      "grad_norm": 0.8619052543414016,
+      "learning_rate": 9.297848034936007e-06,
+      "loss": 0.5278,
+      "num_input_tokens_seen": 85686728,
+      "step": 1485,
+      "train_runtime": 8432.4434,
+      "train_tokens_per_second": 10161.554
+    },
+    {
+      "epoch": 0.3428439944776806,
+      "grad_norm": 0.8470575340587089,
+      "learning_rate": 9.293223518950169e-06,
+      "loss": 0.4772,
+      "num_input_tokens_seen": 85957864,
+      "step": 1490,
+      "train_runtime": 8462.2544,
+      "train_tokens_per_second": 10157.797
+    },
+    {
+      "epoch": 0.34399447768062585,
+      "grad_norm": 0.6714798632698927,
+      "learning_rate": 9.288584981812234e-06,
+      "loss": 0.4879,
+      "num_input_tokens_seen": 86264248,
+      "step": 1495,
+      "train_runtime": 8490.177,
+      "train_tokens_per_second": 10160.477
+    },
+    {
+      "epoch": 0.3451449608835711,
+      "grad_norm": 0.8225447772019122,
+      "learning_rate": 9.283932438671108e-06,
+      "loss": 0.4659,
+      "num_input_tokens_seen": 86528600,
+      "step": 1500,
+      "train_runtime": 8524.3509,
+      "train_tokens_per_second": 10150.755
+    },
+    {
+      "epoch": 0.34629544408651636,
+      "grad_norm": 0.7444033455751631,
+      "learning_rate": 9.279265904721438e-06,
+      "loss": 0.4989,
+      "num_input_tokens_seen": 86794344,
+      "step": 1505,
+      "train_runtime": 8556.1058,
+      "train_tokens_per_second": 10144.141
+    },
+    {
+      "epoch": 0.34744592728946155,
+      "grad_norm": 0.9984685674970626,
+      "learning_rate": 9.274585395203562e-06,
+      "loss": 0.5599,
+      "num_input_tokens_seen": 87053224,
+      "step": 1510,
+      "train_runtime": 8586.9798,
+      "train_tokens_per_second": 10137.816
+    },
+    {
+      "epoch": 0.3485964104924068,
+      "grad_norm": 0.856341562648223,
+      "learning_rate": 9.269890925403461e-06,
+      "loss": 0.5091,
+      "num_input_tokens_seen": 87355952,
+      "step": 1515,
+      "train_runtime": 8608.1161,
+      "train_tokens_per_second": 10148.092
+    },
+    {
+      "epoch": 0.34974689369535206,
+      "grad_norm": 0.7674044577049813,
+      "learning_rate": 9.265182510652713e-06,
+      "loss": 0.5122,
+      "num_input_tokens_seen": 87631696,
+      "step": 1520,
+      "train_runtime": 8634.3608,
+      "train_tokens_per_second": 10149.182
+    },
+    {
+      "epoch": 0.3508973768982973,
+      "grad_norm": 1.0077279107041603,
+      "learning_rate": 9.26046016632843e-06,
+      "loss": 0.5053,
+      "num_input_tokens_seen": 87916656,
+      "step": 1525,
+      "train_runtime": 8663.5799,
+      "train_tokens_per_second": 10147.844
+    },
+    {
+      "epoch": 0.3520478601012425,
+      "grad_norm": 0.7723845326426332,
+      "learning_rate": 9.255723907853226e-06,
+      "loss": 0.5148,
+      "num_input_tokens_seen": 88208128,
+      "step": 1530,
+      "train_runtime": 8688.3022,
+      "train_tokens_per_second": 10152.516
+    },
+    {
+      "epoch": 0.35319834330418776,
+      "grad_norm": 0.8964281541524628,
+      "learning_rate": 9.250973750695146e-06,
+      "loss": 0.4843,
+      "num_input_tokens_seen": 88493832,
+      "step": 1535,
+      "train_runtime": 8716.2163,
+      "train_tokens_per_second": 10152.781
+    },
+    {
+      "epoch": 0.354348826507133,
+      "grad_norm": 0.7740377952419387,
+      "learning_rate": 9.24620971036764e-06,
+      "loss": 0.494,
+      "num_input_tokens_seen": 88780400,
+      "step": 1540,
+      "train_runtime": 8735.1376,
+      "train_tokens_per_second": 10163.595
+    },
+    {
+      "epoch": 0.35549930971007826,
+      "grad_norm": 0.8008116238459994,
+      "learning_rate": 9.241431802429484e-06,
+      "loss": 0.4824,
+      "num_input_tokens_seen": 89063104,
+      "step": 1545,
+      "train_runtime": 8766.5357,
+      "train_tokens_per_second": 10159.441
+    },
+    {
+      "epoch": 0.35664979291302346,
+      "grad_norm": 0.7974800275186695,
+      "learning_rate": 9.236640042484755e-06,
+      "loss": 0.4811,
+      "num_input_tokens_seen": 89361360,
+      "step": 1550,
+      "train_runtime": 8789.1347,
+      "train_tokens_per_second": 10167.253
+    },
+    {
+      "epoch": 0.3578002761159687,
+      "grad_norm": 0.8209975037406593,
+      "learning_rate": 9.231834446182765e-06,
+      "loss": 0.4619,
+      "num_input_tokens_seen": 89650512,
+      "step": 1555,
+      "train_runtime": 8815.5705,
+      "train_tokens_per_second": 10169.564
+    },
+    {
+      "epoch": 0.35895075931891396,
+      "grad_norm": 0.8473721710035395,
+      "learning_rate": 9.227015029218013e-06,
+      "loss": 0.4957,
+      "num_input_tokens_seen": 89946416,
+      "step": 1560,
+      "train_runtime": 8846.1045,
+      "train_tokens_per_second": 10167.912
+    },
+    {
+      "epoch": 0.36010124252185916,
+      "grad_norm": 0.845357182986521,
+      "learning_rate": 9.222181807330135e-06,
+      "loss": 0.5138,
+      "num_input_tokens_seen": 90243064,
+      "step": 1565,
+      "train_runtime": 8878.9933,
+      "train_tokens_per_second": 10163.659
+    },
+    {
+      "epoch": 0.3612517257248044,
+      "grad_norm": 0.7948846373363363,
+      "learning_rate": 9.217334796303856e-06,
+      "loss": 0.5122,
+      "num_input_tokens_seen": 90542880,
+      "step": 1570,
+      "train_runtime": 8901.8321,
+      "train_tokens_per_second": 10171.263
+    },
+    {
+      "epoch": 0.36240220892774966,
+      "grad_norm": 0.776324188506098,
+      "learning_rate": 9.212474011968929e-06,
+      "loss": 0.4868,
+      "num_input_tokens_seen": 90809912,
+      "step": 1575,
+      "train_runtime": 8931.0992,
+      "train_tokens_per_second": 10167.832
+    },
+    {
+      "epoch": 0.3635526921306949,
+      "grad_norm": 0.7127479046810715,
+      "learning_rate": 9.20759947020009e-06,
+      "loss": 0.4938,
+      "num_input_tokens_seen": 91072408,
+      "step": 1580,
+      "train_runtime": 8959.0188,
+      "train_tokens_per_second": 10165.444
+    },
+    {
+      "epoch": 0.3647031753336401,
+      "grad_norm": 0.8113373015609233,
+      "learning_rate": 9.202711186917011e-06,
+      "loss": 0.4758,
+      "num_input_tokens_seen": 91352816,
+      "step": 1585,
+      "train_runtime": 8988.4394,
+      "train_tokens_per_second": 10163.368
+    },
+    {
+      "epoch": 0.36585365853658536,
+      "grad_norm": 0.7755700281043933,
+      "learning_rate": 9.197809178084234e-06,
+      "loss": 0.5089,
+      "num_input_tokens_seen": 91642600,
+      "step": 1590,
+      "train_runtime": 9019.7813,
+      "train_tokens_per_second": 10160.18
+    },
+    {
+      "epoch": 0.3670041417395306,
+      "grad_norm": 0.8620007903891606,
+      "learning_rate": 9.192893459711132e-06,
+      "loss": 0.5086,
+      "num_input_tokens_seen": 91906056,
+      "step": 1595,
+      "train_runtime": 9050.0924,
+      "train_tokens_per_second": 10155.262
+    },
+    {
+      "epoch": 0.36815462494247586,
+      "grad_norm": 0.7720599475275643,
+      "learning_rate": 9.187964047851851e-06,
+      "loss": 0.4763,
+      "num_input_tokens_seen": 92213304,
+      "step": 1600,
+      "train_runtime": 9077.6668,
+      "train_tokens_per_second": 10158.26
+    },
+    {
+      "epoch": 0.36930510814542106,
+      "grad_norm": 0.8502250990709589,
+      "learning_rate": 9.183020958605258e-06,
+      "loss": 0.4967,
+      "num_input_tokens_seen": 92511664,
+      "step": 1605,
+      "train_runtime": 9111.3891,
+      "train_tokens_per_second": 10153.409
+    },
+    {
+      "epoch": 0.3704555913483663,
+      "grad_norm": 0.8692902553821374,
+      "learning_rate": 9.178064208114888e-06,
+      "loss": 0.5154,
+      "num_input_tokens_seen": 92796224,
+      "step": 1610,
+      "train_runtime": 9134.7043,
+      "train_tokens_per_second": 10158.646
+    },
+    {
+      "epoch": 0.37160607455131156,
+      "grad_norm": 0.8209046656556779,
+      "learning_rate": 9.17309381256889e-06,
+      "loss": 0.4949,
+      "num_input_tokens_seen": 93098464,
+      "step": 1615,
+      "train_runtime": 9162.1328,
+      "train_tokens_per_second": 10161.222
+    },
+    {
+      "epoch": 0.3727565577542568,
+      "grad_norm": 1.1004934006560652,
+      "learning_rate": 9.168109788199984e-06,
+      "loss": 0.488,
+      "num_input_tokens_seen": 93372024,
+      "step": 1620,
+      "train_runtime": 9189.5831,
+      "train_tokens_per_second": 10160.638
+    },
+    {
+      "epoch": 0.373907040957202,
+      "grad_norm": 0.8140661051943932,
+      "learning_rate": 9.16311215128539e-06,
+      "loss": 0.5361,
+      "num_input_tokens_seen": 93656216,
+      "step": 1625,
+      "train_runtime": 9214.4343,
+      "train_tokens_per_second": 10164.077
+    },
+    {
+      "epoch": 0.37505752416014726,
+      "grad_norm": 0.8392995045139348,
+      "learning_rate": 9.15810091814679e-06,
+      "loss": 0.5301,
+      "num_input_tokens_seen": 93960456,
+      "step": 1630,
+      "train_runtime": 9242.959,
+      "train_tokens_per_second": 10165.625
+    },
+    {
+      "epoch": 0.3762080073630925,
+      "grad_norm": 0.9149355884956788,
+      "learning_rate": 9.153076105150274e-06,
+      "loss": 0.4838,
+      "num_input_tokens_seen": 94263192,
+      "step": 1635,
+      "train_runtime": 9267.3133,
+      "train_tokens_per_second": 10171.577
+    },
+    {
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.8911020381572199,
+      "learning_rate": 9.148037728706272e-06,
+      "loss": 0.4872,
+      "num_input_tokens_seen": 94545432,
+      "step": 1640,
+      "train_runtime": 9294.0892,
+      "train_tokens_per_second": 10172.641
+    },
+    {
+      "epoch": 0.37850897376898296,
+      "grad_norm": 0.8747480669703357,
+      "learning_rate": 9.142985805269516e-06,
+      "loss": 0.4689,
+      "num_input_tokens_seen": 94842696,
+      "step": 1645,
+      "train_runtime": 9328.552,
+      "train_tokens_per_second": 10166.926
+    },
+    {
+      "epoch": 0.3796594569719282,
+      "grad_norm": 0.6755045650540324,
+      "learning_rate": 9.137920351338984e-06,
+      "loss": 0.4925,
+      "num_input_tokens_seen": 95127440,
+      "step": 1650,
+      "train_runtime": 9361.4985,
+      "train_tokens_per_second": 10161.561
+    },
+    {
+      "epoch": 0.38080994017487346,
+      "grad_norm": 0.9610528606284288,
+      "learning_rate": 9.13284138345784e-06,
+      "loss": 0.5047,
+      "num_input_tokens_seen": 95407824,
+      "step": 1655,
+      "train_runtime": 9391.6731,
+      "train_tokens_per_second": 10158.768
+    },
+    {
+      "epoch": 0.38196042337781866,
+      "grad_norm": 0.878506064665154,
+      "learning_rate": 9.12774891821338e-06,
+      "loss": 0.5032,
+      "num_input_tokens_seen": 95688480,
+      "step": 1660,
+      "train_runtime": 9423.4782,
+      "train_tokens_per_second": 10154.263
+    },
+    {
+      "epoch": 0.3831109065807639,
+      "grad_norm": 0.7340835930629078,
+      "learning_rate": 9.122642972236983e-06,
+      "loss": 0.472,
+      "num_input_tokens_seen": 95971992,
+      "step": 1665,
+      "train_runtime": 9447.8768,
+      "train_tokens_per_second": 10158.049
+    },
+    {
+      "epoch": 0.38426138978370916,
+      "grad_norm": 0.8771279120887797,
+      "learning_rate": 9.11752356220406e-06,
+      "loss": 0.4837,
+      "num_input_tokens_seen": 96260256,
+      "step": 1670,
+      "train_runtime": 9475.3415,
+      "train_tokens_per_second": 10159.028
+    },
+    {
+      "epoch": 0.3854118729866544,
+      "grad_norm": 0.7100733302538136,
+      "learning_rate": 9.112390704833983e-06,
+      "loss": 0.4806,
+      "num_input_tokens_seen": 96555680,
+      "step": 1675,
+      "train_runtime": 9507.4467,
+      "train_tokens_per_second": 10155.795
+    },
+    {
+      "epoch": 0.3865623561895996,
+      "grad_norm": 0.9783418100611737,
+      "learning_rate": 9.107244416890052e-06,
+      "loss": 0.4764,
+      "num_input_tokens_seen": 96860304,
+      "step": 1680,
+      "train_runtime": 9539.275,
+      "train_tokens_per_second": 10153.843
+    },
+    {
+      "epoch": 0.38771283939254486,
+      "grad_norm": 0.8244535443023979,
+      "learning_rate": 9.102084715179423e-06,
+      "loss": 0.4907,
+      "num_input_tokens_seen": 97143480,
+      "step": 1685,
+      "train_runtime": 9572.5579,
+      "train_tokens_per_second": 10148.121
+    },
+    {
+      "epoch": 0.3888633225954901,
+      "grad_norm": 0.8272767548688784,
+      "learning_rate": 9.09691161655306e-06,
+      "loss": 0.4942,
+      "num_input_tokens_seen": 97452528,
+      "step": 1690,
+      "train_runtime": 9596.4818,
+      "train_tokens_per_second": 10155.027
+    },
+    {
+      "epoch": 0.39001380579843536,
+      "grad_norm": 0.7491659554601956,
+      "learning_rate": 9.091725137905685e-06,
+      "loss": 0.4683,
+      "num_input_tokens_seen": 97756368,
+      "step": 1695,
+      "train_runtime": 9627.8119,
+      "train_tokens_per_second": 10153.539
+    },
+    {
+      "epoch": 0.39116428900138056,
+      "grad_norm": 0.8051426986760922,
+      "learning_rate": 9.086525296175712e-06,
+      "loss": 0.5034,
+      "num_input_tokens_seen": 98043912,
+      "step": 1700,
+      "train_runtime": 9655.7616,
+      "train_tokens_per_second": 10153.928
+    },
+    {
+      "epoch": 0.3923147722043258,
+      "grad_norm": 0.8509474387331862,
+      "learning_rate": 9.081312108345199e-06,
+      "loss": 0.5078,
+      "num_input_tokens_seen": 98330352,
+      "step": 1705,
+      "train_runtime": 9685.745,
+      "train_tokens_per_second": 10152.069
+    },
+    {
+      "epoch": 0.39346525540727106,
+      "grad_norm": 0.925129334692823,
+      "learning_rate": 9.076085591439792e-06,
+      "loss": 0.4668,
+      "num_input_tokens_seen": 98606544,
+      "step": 1710,
+      "train_runtime": 9708.5601,
+      "train_tokens_per_second": 10156.66
+    },
+    {
+      "epoch": 0.3946157386102163,
+      "grad_norm": 0.7887521853691624,
+      "learning_rate": 9.070845762528666e-06,
+      "loss": 0.5038,
+      "num_input_tokens_seen": 98895744,
+      "step": 1715,
+      "train_runtime": 9742.9044,
+      "train_tokens_per_second": 10150.54
+    },
+    {
+      "epoch": 0.3957662218131615,
+      "grad_norm": 1.1172418740613836,
+      "learning_rate": 9.065592638724474e-06,
+      "loss": 0.4819,
+      "num_input_tokens_seen": 99157968,
+      "step": 1720,
+      "train_runtime": 9767.1886,
+      "train_tokens_per_second": 10152.15
+    },
+    {
+      "epoch": 0.39691670501610676,
+      "grad_norm": 0.7985339076918294,
+      "learning_rate": 9.060326237183286e-06,
+      "loss": 0.4997,
+      "num_input_tokens_seen": 99455784,
+      "step": 1725,
+      "train_runtime": 9794.3162,
+      "train_tokens_per_second": 10154.439
+    },
+    {
+      "epoch": 0.398067188219052,
+      "grad_norm": 0.7772430807117556,
+      "learning_rate": 9.055046575104535e-06,
+      "loss": 0.4829,
+      "num_input_tokens_seen": 99746440,
+      "step": 1730,
+      "train_runtime": 9817.6133,
+      "train_tokens_per_second": 10159.948
+    },
+    {
+      "epoch": 0.3992176714219972,
+      "grad_norm": 0.7395267344580544,
+      "learning_rate": 9.049753669730968e-06,
+      "loss": 0.4663,
+      "num_input_tokens_seen": 100046496,
+      "step": 1735,
+      "train_runtime": 9847.7541,
+      "train_tokens_per_second": 10159.321
+    },
+    {
+      "epoch": 0.40036815462494246,
+      "grad_norm": 0.9404635766285109,
+      "learning_rate": 9.044447538348572e-06,
+      "loss": 0.5242,
+      "num_input_tokens_seen": 100367200,
+      "step": 1740,
+      "train_runtime": 9879.2292,
+      "train_tokens_per_second": 10159.416
+    },
+    {
+      "epoch": 0.4015186378278877,
+      "grad_norm": 0.7703539868202706,
+      "learning_rate": 9.039128198286538e-06,
+      "loss": 0.4817,
+      "num_input_tokens_seen": 100643328,
+      "step": 1745,
+      "train_runtime": 9908.0398,
+      "train_tokens_per_second": 10157.744
+    },
+    {
+      "epoch": 0.40266912103083297,
+      "grad_norm": 0.8497129852992968,
+      "learning_rate": 9.033795666917191e-06,
+      "loss": 0.5163,
+      "num_input_tokens_seen": 100927144,
+      "step": 1750,
+      "train_runtime": 9935.4909,
+      "train_tokens_per_second": 10158.244
+    },
+    {
+      "epoch": 0.40381960423377816,
+      "grad_norm": 0.8996213374517024,
+      "learning_rate": 9.028449961655938e-06,
+      "loss": 0.4839,
+      "num_input_tokens_seen": 101218728,
+      "step": 1755,
+      "train_runtime": 9958.656,
+      "train_tokens_per_second": 10163.894
+    },
+    {
+      "epoch": 0.4049700874367234,
+      "grad_norm": 0.8778558062897669,
+      "learning_rate": 9.023091099961208e-06,
+      "loss": 0.4718,
+      "num_input_tokens_seen": 101492304,
+      "step": 1760,
+      "train_runtime": 9983.2142,
+      "train_tokens_per_second": 10166.295
+    },
+    {
+      "epoch": 0.40612057063966867,
+      "grad_norm": 0.8835279001102162,
+      "learning_rate": 9.0177190993344e-06,
+      "loss": 0.5046,
+      "num_input_tokens_seen": 101794800,
+      "step": 1765,
+      "train_runtime": 10013.769,
+      "train_tokens_per_second": 10165.483
+    },
+    {
+      "epoch": 0.4072710538426139,
+      "grad_norm": 0.8419111334304098,
+      "learning_rate": 9.012333977319822e-06,
+      "loss": 0.5037,
+      "num_input_tokens_seen": 102082944,
+      "step": 1770,
+      "train_runtime": 10035.0196,
+      "train_tokens_per_second": 10172.67
+    },
+    {
+      "epoch": 0.4084215370455591,
+      "grad_norm": 0.8807554798578316,
+      "learning_rate": 9.00693575150464e-06,
+      "loss": 0.5265,
+      "num_input_tokens_seen": 102378496,
+      "step": 1775,
+      "train_runtime": 10055.4876,
+      "train_tokens_per_second": 10181.356
+    },
+    {
+      "epoch": 0.40957202024850436,
+      "grad_norm": 0.6627155991685542,
+      "learning_rate": 9.001524439518802e-06,
+      "loss": 0.483,
+      "num_input_tokens_seen": 102669008,
+      "step": 1780,
+      "train_runtime": 10079.2893,
+      "train_tokens_per_second": 10186.136
+    },
+    {
+      "epoch": 0.4107225034514496,
+      "grad_norm": 1.0204784762057804,
+      "learning_rate": 8.996100059035012e-06,
+      "loss": 0.4844,
+      "num_input_tokens_seen": 102977136,
+      "step": 1785,
+      "train_runtime": 10103.5683,
+      "train_tokens_per_second": 10192.155
+    },
+    {
+      "epoch": 0.41187298665439487,
+      "grad_norm": 0.7280441697484784,
+      "learning_rate": 8.990662627768641e-06,
+      "loss": 0.5004,
+      "num_input_tokens_seen": 103291848,
+      "step": 1790,
+      "train_runtime": 10127.7528,
+      "train_tokens_per_second": 10198.891
+    },
+    {
+      "epoch": 0.41302346985734006,
+      "grad_norm": 0.7594650132144262,
+      "learning_rate": 8.985212163477682e-06,
+      "loss": 0.4927,
+      "num_input_tokens_seen": 103551400,
+      "step": 1795,
+      "train_runtime": 10160.371,
+      "train_tokens_per_second": 10191.695
+    },
+    {
+      "epoch": 0.4141739530602853,
+      "grad_norm": 0.9033946589639509,
+      "learning_rate": 8.979748683962702e-06,
+      "loss": 0.5224,
+      "num_input_tokens_seen": 103841104,
+      "step": 1800,
+      "train_runtime": 10188.5935,
+      "train_tokens_per_second": 10191.898
+    },
+    {
+      "epoch": 0.41532443626323057,
+      "grad_norm": 0.8077550941924575,
+      "learning_rate": 8.974272207066767e-06,
+      "loss": 0.4942,
+      "num_input_tokens_seen": 104123440,
+      "step": 1805,
+      "train_runtime": 10215.8944,
+      "train_tokens_per_second": 10192.298
+    },
+    {
+      "epoch": 0.4164749194661758,
+      "grad_norm": 0.7331253385486215,
+      "learning_rate": 8.968782750675393e-06,
+      "loss": 0.4894,
+      "num_input_tokens_seen": 104426360,
+      "step": 1810,
+      "train_runtime": 10244.2139,
+      "train_tokens_per_second": 10193.692
+    },
+    {
+      "epoch": 0.417625402669121,
+      "grad_norm": 0.8516524552969844,
+      "learning_rate": 8.963280332716481e-06,
+      "loss": 0.5301,
+      "num_input_tokens_seen": 104714984,
+      "step": 1815,
+      "train_runtime": 10272.6587,
+      "train_tokens_per_second": 10193.562
+    },
+    {
+      "epoch": 0.41877588587206627,
+      "grad_norm": 0.7920696330662705,
+      "learning_rate": 8.957764971160271e-06,
+      "loss": 0.489,
+      "num_input_tokens_seen": 105014560,
+      "step": 1820,
+      "train_runtime": 10302.9006,
+      "train_tokens_per_second": 10192.718
+    },
+    {
+      "epoch": 0.4199263690750115,
+      "grad_norm": 0.8150752443622702,
+      "learning_rate": 8.95223668401927e-06,
+      "loss": 0.4659,
+      "num_input_tokens_seen": 105289936,
+      "step": 1825,
+      "train_runtime": 10332.1274,
+      "train_tokens_per_second": 10190.538
+    },
+    {
+      "epoch": 0.4210768522779567,
+      "grad_norm": 0.8946327761842672,
+      "learning_rate": 8.946695489348201e-06,
+      "loss": 0.4552,
+      "num_input_tokens_seen": 105567264,
+      "step": 1830,
+      "train_runtime": 10358.5039,
+      "train_tokens_per_second": 10191.362
+    },
+    {
+      "epoch": 0.42222733548090197,
+      "grad_norm": 0.7718692733602033,
+      "learning_rate": 8.941141405243939e-06,
+      "loss": 0.5175,
+      "num_input_tokens_seen": 105879120,
+      "step": 1835,
+      "train_runtime": 10386.0015,
+      "train_tokens_per_second": 10194.406
+    },
+    {
+      "epoch": 0.4233778186838472,
+      "grad_norm": 0.7643680382482771,
+      "learning_rate": 8.935574449845451e-06,
+      "loss": 0.4933,
+      "num_input_tokens_seen": 106162384,
+      "step": 1840,
+      "train_runtime": 10409.0603,
+      "train_tokens_per_second": 10199.036
+    },
+    {
+      "epoch": 0.42452830188679247,
+      "grad_norm": 0.8016215206700045,
+      "learning_rate": 8.929994641333748e-06,
+      "loss": 0.4865,
+      "num_input_tokens_seen": 106458192,
+      "step": 1845,
+      "train_runtime": 10437.8813,
+      "train_tokens_per_second": 10199.215
+    },
+    {
+      "epoch": 0.42567878508973767,
+      "grad_norm": 0.7431259516196493,
+      "learning_rate": 8.924401997931818e-06,
+      "loss": 0.5095,
+      "num_input_tokens_seen": 106745944,
+      "step": 1850,
+      "train_runtime": 10471.9065,
+      "train_tokens_per_second": 10193.554
+    },
+    {
+      "epoch": 0.4268292682926829,
+      "grad_norm": 0.8832083238963427,
+      "learning_rate": 8.918796537904555e-06,
+      "loss": 0.4961,
+      "num_input_tokens_seen": 107021280,
+      "step": 1855,
+      "train_runtime": 10494.2932,
+      "train_tokens_per_second": 10198.046
+    },
+    {
+      "epoch": 0.42797975149562817,
+      "grad_norm": 0.7464777452059294,
+      "learning_rate": 8.913178279558721e-06,
+      "loss": 0.4964,
+      "num_input_tokens_seen": 107330360,
+      "step": 1860,
+      "train_runtime": 10517.2702,
+      "train_tokens_per_second": 10205.154
+    },
+    {
+      "epoch": 0.4291302346985734,
+      "grad_norm": 0.7677865664568925,
+      "learning_rate": 8.907547241242874e-06,
+      "loss": 0.4929,
+      "num_input_tokens_seen": 107630272,
+      "step": 1865,
+      "train_runtime": 10542.8472,
+      "train_tokens_per_second": 10208.843
+    },
+    {
+      "epoch": 0.4302807179015186,
+      "grad_norm": 0.7929138801024914,
+      "learning_rate": 8.901903441347307e-06,
+      "loss": 0.4813,
+      "num_input_tokens_seen": 107931736,
+      "step": 1870,
+      "train_runtime": 10572.3215,
+      "train_tokens_per_second": 10208.896
+    },
+    {
+      "epoch": 0.43143120110446387,
+      "grad_norm": 0.7591247701210149,
+      "learning_rate": 8.896246898303991e-06,
+      "loss": 0.4917,
+      "num_input_tokens_seen": 108217464,
+      "step": 1875,
+      "train_runtime": 10599.6544,
+      "train_tokens_per_second": 10209.528
+    },
+    {
+      "epoch": 0.4325816843074091,
+      "grad_norm": 0.800096472307934,
+      "learning_rate": 8.890577630586518e-06,
+      "loss": 0.5023,
+      "num_input_tokens_seen": 108500672,
+      "step": 1880,
+      "train_runtime": 10627.4395,
+      "train_tokens_per_second": 10209.484
+    },
+    {
+      "epoch": 0.43373216751035437,
+      "grad_norm": 0.8361309526339042,
+      "learning_rate": 8.884895656710034e-06,
+      "loss": 0.4943,
+      "num_input_tokens_seen": 108756872,
+      "step": 1885,
+      "train_runtime": 10655.8693,
+      "train_tokens_per_second": 10206.288
+    },
+    {
+      "epoch": 0.43488265071329957,
+      "grad_norm": 0.7639558542791104,
+      "learning_rate": 8.879200995231183e-06,
+      "loss": 0.496,
+      "num_input_tokens_seen": 109050592,
+      "step": 1890,
+      "train_runtime": 10687.0924,
+      "train_tokens_per_second": 10203.953
+    },
+    {
+      "epoch": 0.4360331339162448,
+      "grad_norm": 0.8022055739665632,
+      "learning_rate": 8.873493664748047e-06,
+      "loss": 0.4948,
+      "num_input_tokens_seen": 109335920,
+      "step": 1895,
+      "train_runtime": 10711.3628,
+      "train_tokens_per_second": 10207.471
+    },
+    {
+      "epoch": 0.43718361711919007,
+      "grad_norm": 0.8047205901243916,
+      "learning_rate": 8.867773683900078e-06,
+      "loss": 0.4761,
+      "num_input_tokens_seen": 109616280,
+      "step": 1900,
+      "train_runtime": 10734.5778,
+      "train_tokens_per_second": 10211.513
+    },
+    {
+      "epoch": 0.4383341003221353,
+      "grad_norm": 0.8450136082533041,
+      "learning_rate": 8.862041071368048e-06,
+      "loss": 0.4832,
+      "num_input_tokens_seen": 109882296,
+      "step": 1905,
+      "train_runtime": 10761.7042,
+      "train_tokens_per_second": 10210.492
+    },
+    {
+      "epoch": 0.4394845835250805,
+      "grad_norm": 0.7483749354096689,
+      "learning_rate": 8.85629584587398e-06,
+      "loss": 0.4903,
+      "num_input_tokens_seen": 110182312,
+      "step": 1910,
+      "train_runtime": 10788.5314,
+      "train_tokens_per_second": 10212.911
+    },
+    {
+      "epoch": 0.44063506672802577,
+      "grad_norm": 0.7600919668006907,
+      "learning_rate": 8.850538026181092e-06,
+      "loss": 0.5056,
+      "num_input_tokens_seen": 110474032,
+      "step": 1915,
+      "train_runtime": 10813.5307,
+      "train_tokens_per_second": 10216.278
+    },
+    {
+      "epoch": 0.441785549930971,
+      "grad_norm": 0.7205956072824878,
+      "learning_rate": 8.844767631093729e-06,
+      "loss": 0.4743,
+      "num_input_tokens_seen": 110782488,
+      "step": 1920,
+      "train_runtime": 10836.2682,
+      "train_tokens_per_second": 10223.306
+    },
+    {
+      "epoch": 0.4429360331339162,
+      "grad_norm": 0.754079906452407,
+      "learning_rate": 8.838984679457308e-06,
+      "loss": 0.4924,
+      "num_input_tokens_seen": 111067664,
+      "step": 1925,
+      "train_runtime": 10861.8697,
+      "train_tokens_per_second": 10225.465
+    },
+    {
+      "epoch": 0.44408651633686147,
+      "grad_norm": 0.6810528023769105,
+      "learning_rate": 8.833189190158255e-06,
+      "loss": 0.5022,
+      "num_input_tokens_seen": 111370800,
+      "step": 1930,
+      "train_runtime": 10883.9755,
+      "train_tokens_per_second": 10232.548
+    },
+    {
+      "epoch": 0.4452369995398067,
+      "grad_norm": 0.9180785091187725,
+      "learning_rate": 8.827381182123941e-06,
+      "loss": 0.4736,
+      "num_input_tokens_seen": 111656088,
+      "step": 1935,
+      "train_runtime": 10915.1296,
+      "train_tokens_per_second": 10229.479
+    },
+    {
+      "epoch": 0.446387482742752,
+      "grad_norm": 0.8649692435077119,
+      "learning_rate": 8.821560674322625e-06,
+      "loss": 0.4777,
+      "num_input_tokens_seen": 111948688,
+      "step": 1940,
+      "train_runtime": 10943.9794,
+      "train_tokens_per_second": 10229.249
+    },
+    {
+      "epoch": 0.44753796594569717,
+      "grad_norm": 1.0359820802640534,
+      "learning_rate": 8.815727685763383e-06,
+      "loss": 0.5169,
+      "num_input_tokens_seen": 112255024,
+      "step": 1945,
+      "train_runtime": 10970.5054,
+      "train_tokens_per_second": 10232.439
+    },
+    {
+      "epoch": 0.4486884491486424,
+      "grad_norm": 0.8135596729479354,
+      "learning_rate": 8.809882235496053e-06,
+      "loss": 0.4691,
+      "num_input_tokens_seen": 112536168,
+      "step": 1950,
+      "train_runtime": 10997.0893,
+      "train_tokens_per_second": 10233.269
+    },
+    {
+      "epoch": 0.4498389323515877,
+      "grad_norm": 0.8451951357686893,
+      "learning_rate": 8.804024342611178e-06,
+      "loss": 0.522,
+      "num_input_tokens_seen": 112819560,
+      "step": 1955,
+      "train_runtime": 11023.3677,
+      "train_tokens_per_second": 10234.582
+    },
+    {
+      "epoch": 0.4509894155545329,
+      "grad_norm": 0.7064128413569595,
+      "learning_rate": 8.79815402623993e-06,
+      "loss": 0.4841,
+      "num_input_tokens_seen": 113134440,
+      "step": 1960,
+      "train_runtime": 11055.8664,
+      "train_tokens_per_second": 10232.978
+    },
+    {
+      "epoch": 0.4521398987574781,
+      "grad_norm": 0.7456951715507291,
+      "learning_rate": 8.792271305554054e-06,
+      "loss": 0.4611,
+      "num_input_tokens_seen": 113428616,
+      "step": 1965,
+      "train_runtime": 11088.8261,
+      "train_tokens_per_second": 10229.091
+    },
+    {
+      "epoch": 0.45329038196042337,
+      "grad_norm": 0.7306979441149495,
+      "learning_rate": 8.786376199765811e-06,
+      "loss": 0.482,
+      "num_input_tokens_seen": 113710920,
+      "step": 1970,
+      "train_runtime": 11113.605,
+      "train_tokens_per_second": 10231.686
+    },
+    {
+      "epoch": 0.4544408651633686,
+      "grad_norm": 0.7273509965279797,
+      "learning_rate": 8.780468728127911e-06,
+      "loss": 0.4573,
+      "num_input_tokens_seen": 113969264,
+      "step": 1975,
+      "train_runtime": 11138.9888,
+      "train_tokens_per_second": 10231.563
+    },
+    {
+      "epoch": 0.4555913483663139,
+      "grad_norm": 0.7224060271667133,
+      "learning_rate": 8.774548909933443e-06,
+      "loss": 0.4985,
+      "num_input_tokens_seen": 114251680,
+      "step": 1980,
+      "train_runtime": 11167.6792,
+      "train_tokens_per_second": 10230.566
+    },
+    {
+      "epoch": 0.45674183156925907,
+      "grad_norm": 0.7832432875924432,
+      "learning_rate": 8.768616764515822e-06,
+      "loss": 0.4778,
+      "num_input_tokens_seen": 114541440,
+      "step": 1985,
+      "train_runtime": 11199.9428,
+      "train_tokens_per_second": 10226.966
+    },
+    {
+      "epoch": 0.4578923147722043,
+      "grad_norm": 0.8035409086682139,
+      "learning_rate": 8.762672311248723e-06,
+      "loss": 0.4982,
+      "num_input_tokens_seen": 114842176,
+      "step": 1990,
+      "train_runtime": 11233.6382,
+      "train_tokens_per_second": 10223.062
+    },
+    {
+      "epoch": 0.4590427979751496,
+      "grad_norm": 0.8753556384563861,
+      "learning_rate": 8.756715569546017e-06,
+      "loss": 0.5014,
+      "num_input_tokens_seen": 115119224,
+      "step": 1995,
+      "train_runtime": 11259.8938,
+      "train_tokens_per_second": 10223.829
+    },
+    {
+      "epoch": 0.4601932811780948,
+      "grad_norm": 0.9060524997435464,
+      "learning_rate": 8.750746558861705e-06,
+      "loss": 0.508,
+      "num_input_tokens_seen": 115400488,
+      "step": 2000,
+      "train_runtime": 11289.0661,
+      "train_tokens_per_second": 10222.324
+    },
+    {
+      "epoch": 0.46134376438104,
+      "grad_norm": 1.08331497130723,
+      "learning_rate": 8.744765298689859e-06,
+      "loss": 0.5029,
+      "num_input_tokens_seen": 115707904,
+      "step": 2005,
+      "train_runtime": 11315.3708,
+      "train_tokens_per_second": 10225.728
+    },
+    {
+      "epoch": 0.4624942475839853,
+      "grad_norm": 0.852244888408279,
+      "learning_rate": 8.738771808564555e-06,
+      "loss": 0.4977,
+      "num_input_tokens_seen": 116009480,
+      "step": 2010,
+      "train_runtime": 11337.9827,
+      "train_tokens_per_second": 10231.933
+    },
+    {
+      "epoch": 0.4636447307869305,
+      "grad_norm": 0.8801531776220354,
+      "learning_rate": 8.732766108059814e-06,
+      "loss": 0.5036,
+      "num_input_tokens_seen": 116258440,
+      "step": 2015,
+      "train_runtime": 11361.9896,
+      "train_tokens_per_second": 10232.226
+    },
+    {
+      "epoch": 0.4647952139898757,
+      "grad_norm": 0.8818857416522501,
+      "learning_rate": 8.72674821678953e-06,
+      "loss": 0.4634,
+      "num_input_tokens_seen": 116551304,
+      "step": 2020,
+      "train_runtime": 11390.6423,
+      "train_tokens_per_second": 10232.198
+    },
+    {
+      "epoch": 0.465945697192821,
+      "grad_norm": 0.7815363667154609,
+      "learning_rate": 8.720718154407413e-06,
+      "loss": 0.4869,
+      "num_input_tokens_seen": 116808024,
+      "step": 2025,
+      "train_runtime": 11414.0788,
+      "train_tokens_per_second": 10233.68
+    },
+    {
+      "epoch": 0.4670961803957662,
+      "grad_norm": 0.7619129847096116,
+      "learning_rate": 8.714675940606923e-06,
+      "loss": 0.5017,
+      "num_input_tokens_seen": 117083792,
+      "step": 2030,
+      "train_runtime": 11447.7587,
+      "train_tokens_per_second": 10227.661
+    },
+    {
+      "epoch": 0.4682466635987115,
+      "grad_norm": 0.8607969496032352,
+      "learning_rate": 8.708621595121204e-06,
+      "loss": 0.4603,
+      "num_input_tokens_seen": 117341256,
+      "step": 2035,
+      "train_runtime": 11472.0431,
+      "train_tokens_per_second": 10228.453
+    },
+    {
+      "epoch": 0.4693971468016567,
+      "grad_norm": 0.7953233339855286,
+      "learning_rate": 8.70255513772302e-06,
+      "loss": 0.4557,
+      "num_input_tokens_seen": 117630560,
+      "step": 2040,
+      "train_runtime": 11496.3602,
+      "train_tokens_per_second": 10231.983
+    },
+    {
+      "epoch": 0.4705476300046019,
+      "grad_norm": 0.841730288113026,
+      "learning_rate": 8.696476588224691e-06,
+      "loss": 0.5005,
+      "num_input_tokens_seen": 117913968,
+      "step": 2045,
+      "train_runtime": 11525.4664,
+      "train_tokens_per_second": 10230.733
+    },
+    {
+      "epoch": 0.4716981132075472,
+      "grad_norm": 0.7601641907826765,
+      "learning_rate": 8.690385966478033e-06,
+      "loss": 0.4607,
+      "num_input_tokens_seen": 118181200,
+      "step": 2050,
+      "train_runtime": 11553.8974,
+      "train_tokens_per_second": 10228.687
+    },
+    {
+      "epoch": 0.4728485964104924,
+      "grad_norm": 0.8849370905174317,
+      "learning_rate": 8.684283292374282e-06,
+      "loss": 0.5111,
+      "num_input_tokens_seen": 118467640,
+      "step": 2055,
+      "train_runtime": 11583.2757,
+      "train_tokens_per_second": 10227.473
+    },
+    {
+      "epoch": 0.4739990796134376,
+      "grad_norm": 0.7499151326964894,
+      "learning_rate": 8.678168585844036e-06,
+      "loss": 0.495,
+      "num_input_tokens_seen": 118767528,
+      "step": 2060,
+      "train_runtime": 11613.3622,
+      "train_tokens_per_second": 10226.8
+    },
+    {
+      "epoch": 0.4751495628163829,
+      "grad_norm": 0.7916308368070114,
+      "learning_rate": 8.672041866857198e-06,
+      "loss": 0.4731,
+      "num_input_tokens_seen": 119084184,
+      "step": 2065,
+      "train_runtime": 11645.7183,
+      "train_tokens_per_second": 10225.577
+    },
+    {
+      "epoch": 0.4763000460193281,
+      "grad_norm": 0.7252974077104789,
+      "learning_rate": 8.665903155422888e-06,
+      "loss": 0.4957,
+      "num_input_tokens_seen": 119359992,
+      "step": 2070,
+      "train_runtime": 11672.5532,
+      "train_tokens_per_second": 10225.697
+    },
+    {
+      "epoch": 0.4774505292222734,
+      "grad_norm": 0.9529916071608601,
+      "learning_rate": 8.659752471589407e-06,
+      "loss": 0.4927,
+      "num_input_tokens_seen": 119668280,
+      "step": 2075,
+      "train_runtime": 11700.0068,
+      "train_tokens_per_second": 10228.052
+    },
+    {
+      "epoch": 0.4786010124252186,
+      "grad_norm": 0.9259644620575997,
+      "learning_rate": 8.653589835444145e-06,
+      "loss": 0.4722,
+      "num_input_tokens_seen": 119942576,
+      "step": 2080,
+      "train_runtime": 11727.6809,
+      "train_tokens_per_second": 10227.306
+    },
+    {
+      "epoch": 0.4797514956281638,
+      "grad_norm": 0.9106165272510715,
+      "learning_rate": 8.647415267113533e-06,
+      "loss": 0.5061,
+      "num_input_tokens_seen": 120243896,
+      "step": 2085,
+      "train_runtime": 11755.8633,
+      "train_tokens_per_second": 10228.419
+    },
+    {
+      "epoch": 0.4809019788311091,
+      "grad_norm": 0.7444099095516836,
+      "learning_rate": 8.641228786762968e-06,
+      "loss": 0.5266,
+      "num_input_tokens_seen": 120533376,
+      "step": 2090,
+      "train_runtime": 11783.3156,
+      "train_tokens_per_second": 10229.156
+    },
+    {
+      "epoch": 0.48205246203405433,
+      "grad_norm": 0.7653164771627062,
+      "learning_rate": 8.635030414596752e-06,
+      "loss": 0.4608,
+      "num_input_tokens_seen": 120838320,
+      "step": 2095,
+      "train_runtime": 11802.3125,
+      "train_tokens_per_second": 10238.529
+    },
+    {
+      "epoch": 0.4832029452369995,
+      "grad_norm": 0.7628511766595231,
+      "learning_rate": 8.628820170858024e-06,
+      "loss": 0.49,
+      "num_input_tokens_seen": 121135168,
+      "step": 2100,
+      "train_runtime": 11828.6582,
+      "train_tokens_per_second": 10240.821
+    },
+    {
+      "epoch": 0.4843534284399448,
+      "grad_norm": 0.8496669986485003,
+      "learning_rate": 8.622598075828692e-06,
+      "loss": 0.5016,
+      "num_input_tokens_seen": 121421144,
+      "step": 2105,
+      "train_runtime": 11852.8461,
+      "train_tokens_per_second": 10244.05
+    },
+    {
+      "epoch": 0.48550391164289003,
+      "grad_norm": 0.8193592219537081,
+      "learning_rate": 8.616364149829374e-06,
+      "loss": 0.4903,
+      "num_input_tokens_seen": 121708664,
+      "step": 2110,
+      "train_runtime": 11874.4432,
+      "train_tokens_per_second": 10249.631
+    },
+    {
+      "epoch": 0.4866543948458352,
+      "grad_norm": 0.7398419486938662,
+      "learning_rate": 8.610118413219321e-06,
+      "loss": 0.4537,
+      "num_input_tokens_seen": 121972512,
+      "step": 2115,
+      "train_runtime": 11895.9407,
+      "train_tokens_per_second": 10253.289
+    },
+    {
+      "epoch": 0.4878048780487805,
+      "grad_norm": 0.9009737817117937,
+      "learning_rate": 8.603860886396359e-06,
+      "loss": 0.5128,
+      "num_input_tokens_seen": 122250560,
+      "step": 2120,
+      "train_runtime": 11927.6004,
+      "train_tokens_per_second": 10249.384
+    },
+    {
+      "epoch": 0.48895536125172573,
+      "grad_norm": 0.7704585692936685,
+      "learning_rate": 8.597591589796816e-06,
+      "loss": 0.4897,
+      "num_input_tokens_seen": 122547240,
+      "step": 2125,
+      "train_runtime": 11949.0398,
+      "train_tokens_per_second": 10255.823
+    },
+    {
+      "epoch": 0.490105844454671,
+      "grad_norm": 0.7305469263461605,
+      "learning_rate": 8.591310543895464e-06,
+      "loss": 0.466,
+      "num_input_tokens_seen": 122810280,
+      "step": 2130,
+      "train_runtime": 11970.7161,
+      "train_tokens_per_second": 10259.226
+    },
+    {
+      "epoch": 0.4912563276576162,
+      "grad_norm": 0.8875924601257352,
+      "learning_rate": 8.585017769205444e-06,
+      "loss": 0.4976,
+      "num_input_tokens_seen": 123114600,
+      "step": 2135,
+      "train_runtime": 11996.5071,
+      "train_tokens_per_second": 10262.537
+    },
+    {
+      "epoch": 0.49240681086056143,
+      "grad_norm": 0.8426868414073386,
+      "learning_rate": 8.5787132862782e-06,
+      "loss": 0.5123,
+      "num_input_tokens_seen": 123422264,
+      "step": 2140,
+      "train_runtime": 12018.997,
+      "train_tokens_per_second": 10268.932
+    },
+    {
+      "epoch": 0.4935572940635067,
+      "grad_norm": 0.962430143327387,
+      "learning_rate": 8.57239711570342e-06,
+      "loss": 0.4922,
+      "num_input_tokens_seen": 123699704,
+      "step": 2145,
+      "train_runtime": 12053.1521,
+      "train_tokens_per_second": 10262.851
+    },
+    {
+      "epoch": 0.49470777726645193,
+      "grad_norm": 0.8038947300397576,
+      "learning_rate": 8.566069278108953e-06,
+      "loss": 0.5064,
+      "num_input_tokens_seen": 123975840,
+      "step": 2150,
+      "train_runtime": 12081.6789,
+      "train_tokens_per_second": 10261.475
+    },
+    {
+      "epoch": 0.4958582604693971,
+      "grad_norm": 0.8138395583449975,
+      "learning_rate": 8.559729794160757e-06,
+      "loss": 0.4745,
+      "num_input_tokens_seen": 124238232,
+      "step": 2155,
+      "train_runtime": 12106.6914,
+      "train_tokens_per_second": 10261.948
+    },
+    {
+      "epoch": 0.4970087436723424,
+      "grad_norm": 0.8089185520605962,
+      "learning_rate": 8.553378684562829e-06,
+      "loss": 0.5046,
+      "num_input_tokens_seen": 124519112,
+      "step": 2160,
+      "train_runtime": 12133.061,
+      "train_tokens_per_second": 10262.794
+    },
+    {
+      "epoch": 0.49815922687528763,
+      "grad_norm": 0.7888355523186774,
+      "learning_rate": 8.547015970057125e-06,
+      "loss": 0.4553,
+      "num_input_tokens_seen": 124815704,
+      "step": 2165,
+      "train_runtime": 12162.7938,
+      "train_tokens_per_second": 10262.092
+    },
+    {
+      "epoch": 0.4993097100782329,
+      "grad_norm": 0.8071837985792354,
+      "learning_rate": 8.540641671423508e-06,
+      "loss": 0.4793,
+      "num_input_tokens_seen": 125107624,
+      "step": 2170,
+      "train_runtime": 12187.7851,
+      "train_tokens_per_second": 10265.001
+    },
+    {
+      "epoch": 0.5004601932811781,
+      "grad_norm": 0.7859025201445226,
+      "learning_rate": 8.534255809479673e-06,
+      "loss": 0.5093,
+      "num_input_tokens_seen": 125396960,
+      "step": 2175,
+      "train_runtime": 12215.409,
+      "train_tokens_per_second": 10265.474
+    },
+    {
+      "epoch": 0.5016106764841234,
+      "grad_norm": 0.8786657811518497,
+      "learning_rate": 8.527858405081073e-06,
+      "loss": 0.4769,
+      "num_input_tokens_seen": 125696680,
+      "step": 2180,
+      "train_runtime": 12235.2377,
+      "train_tokens_per_second": 10273.334
+    },
+    {
+      "epoch": 0.5027611596870686,
+      "grad_norm": 0.7889399326584061,
+      "learning_rate": 8.521449479120866e-06,
+      "loss": 0.481,
+      "num_input_tokens_seen": 125981000,
+      "step": 2185,
+      "train_runtime": 12267.2893,
+      "train_tokens_per_second": 10269.669
+    },
+    {
+      "epoch": 0.5039116428900138,
+      "grad_norm": 0.8708485192457405,
+      "learning_rate": 8.515029052529834e-06,
+      "loss": 0.5143,
+      "num_input_tokens_seen": 126263512,
+      "step": 2190,
+      "train_runtime": 12297.9886,
+      "train_tokens_per_second": 10267.005
+    },
+    {
+      "epoch": 0.5050621260929591,
+      "grad_norm": 0.8595125647389966,
+      "learning_rate": 8.508597146276315e-06,
+      "loss": 0.466,
+      "num_input_tokens_seen": 126557792,
+      "step": 2195,
+      "train_runtime": 12323.9832,
+      "train_tokens_per_second": 10269.228
+    },
+    {
+      "epoch": 0.5062126092959043,
+      "grad_norm": 0.9379791505013956,
+      "learning_rate": 8.502153781366147e-06,
+      "loss": 0.5225,
+      "num_input_tokens_seen": 126847672,
+      "step": 2200,
+      "train_runtime": 12348.2791,
+      "train_tokens_per_second": 10272.498
+    },
+    {
+      "epoch": 0.5073630924988495,
+      "grad_norm": 0.8949370095877753,
+      "learning_rate": 8.495698978842582e-06,
+      "loss": 0.4762,
+      "num_input_tokens_seen": 127145176,
+      "step": 2205,
+      "train_runtime": 12371.0889,
+      "train_tokens_per_second": 10277.606
+    },
+    {
+      "epoch": 0.5085135757017948,
+      "grad_norm": 0.7526055059242389,
+      "learning_rate": 8.48923275978623e-06,
+      "loss": 0.5082,
+      "num_input_tokens_seen": 127440832,
+      "step": 2210,
+      "train_runtime": 12390.1045,
+      "train_tokens_per_second": 10285.695
+    },
+    {
+      "epoch": 0.50966405890474,
+      "grad_norm": 0.7255045527622521,
+      "learning_rate": 8.482755145314987e-06,
+      "loss": 0.481,
+      "num_input_tokens_seen": 127716280,
+      "step": 2215,
+      "train_runtime": 12419.0748,
+      "train_tokens_per_second": 10283.88
+    },
+    {
+      "epoch": 0.5108145421076852,
+      "grad_norm": 0.9154067158195597,
+      "learning_rate": 8.47626615658396e-06,
+      "loss": 0.4835,
+      "num_input_tokens_seen": 128006912,
+      "step": 2220,
+      "train_runtime": 12441.4475,
+      "train_tokens_per_second": 10288.748
+    },
+    {
+      "epoch": 0.5119650253106305,
+      "grad_norm": 0.8471779481838955,
+      "learning_rate": 8.46976581478541e-06,
+      "loss": 0.5208,
+      "num_input_tokens_seen": 128305064,
+      "step": 2225,
+      "train_runtime": 12470.8197,
+      "train_tokens_per_second": 10288.423
+    },
+    {
+      "epoch": 0.5131155085135757,
+      "grad_norm": 0.8996654266253066,
+      "learning_rate": 8.46325414114867e-06,
+      "loss": 0.4832,
+      "num_input_tokens_seen": 128582696,
+      "step": 2230,
+      "train_runtime": 12499.7285,
+      "train_tokens_per_second": 10286.839
+    },
+    {
+      "epoch": 0.514265991716521,
+      "grad_norm": 0.8373632887709677,
+      "learning_rate": 8.456731156940085e-06,
+      "loss": 0.4973,
+      "num_input_tokens_seen": 128869904,
+      "step": 2235,
+      "train_runtime": 12521.5608,
+      "train_tokens_per_second": 10291.84
+    },
+    {
+      "epoch": 0.5154164749194662,
+      "grad_norm": 0.7468342048174506,
+      "learning_rate": 8.450196883462936e-06,
+      "loss": 0.4799,
+      "num_input_tokens_seen": 129176016,
+      "step": 2240,
+      "train_runtime": 12555.4622,
+      "train_tokens_per_second": 10288.432
+    },
+    {
+      "epoch": 0.5165669581224114,
+      "grad_norm": 0.7581631616058633,
+      "learning_rate": 8.443651342057377e-06,
+      "loss": 0.4749,
+      "num_input_tokens_seen": 129483280,
+      "step": 2245,
+      "train_runtime": 12590.5339,
+      "train_tokens_per_second": 10284.177
+    },
+    {
+      "epoch": 0.5177174413253567,
+      "grad_norm": 0.8308350722197435,
+      "learning_rate": 8.437094554100358e-06,
+      "loss": 0.4681,
+      "num_input_tokens_seen": 129746824,
+      "step": 2250,
+      "train_runtime": 12614.1366,
+      "train_tokens_per_second": 10285.827
+    },
+    {
+      "epoch": 0.5188679245283019,
+      "grad_norm": 0.8549694681913604,
+      "learning_rate": 8.43052654100556e-06,
+      "loss": 0.4598,
+      "num_input_tokens_seen": 130042184,
+      "step": 2255,
+      "train_runtime": 12643.4704,
+      "train_tokens_per_second": 10285.324
+    },
+    {
+      "epoch": 0.5200184077312471,
+      "grad_norm": 0.9174978967728845,
+      "learning_rate": 8.423947324223327e-06,
+      "loss": 0.4669,
+      "num_input_tokens_seen": 130310872,
+      "step": 2260,
+      "train_runtime": 12671.3128,
+      "train_tokens_per_second": 10283.928
+    },
+    {
+      "epoch": 0.5211688909341924,
+      "grad_norm": 0.8016937005916187,
+      "learning_rate": 8.417356925240587e-06,
+      "loss": 0.4823,
+      "num_input_tokens_seen": 130629640,
+      "step": 2265,
+      "train_runtime": 12707.6737,
+      "train_tokens_per_second": 10279.587
+    },
+    {
+      "epoch": 0.5223193741371376,
+      "grad_norm": 0.8130695215406989,
+      "learning_rate": 8.410755365580797e-06,
+      "loss": 0.4767,
+      "num_input_tokens_seen": 130887560,
+      "step": 2270,
+      "train_runtime": 12729.6771,
+      "train_tokens_per_second": 10282.08
+    },
+    {
+      "epoch": 0.5234698573400829,
+      "grad_norm": 0.7461677865423545,
+      "learning_rate": 8.40414266680385e-06,
+      "loss": 0.469,
+      "num_input_tokens_seen": 131169360,
+      "step": 2275,
+      "train_runtime": 12761.1887,
+      "train_tokens_per_second": 10278.773
+    },
+    {
+      "epoch": 0.5246203405430281,
+      "grad_norm": 0.8556515189828957,
+      "learning_rate": 8.39751885050603e-06,
+      "loss": 0.5019,
+      "num_input_tokens_seen": 131460536,
+      "step": 2280,
+      "train_runtime": 12790.5689,
+      "train_tokens_per_second": 10277.927
+    },
+    {
+      "epoch": 0.5257708237459733,
+      "grad_norm": 0.9208648521489012,
+      "learning_rate": 8.390883938319922e-06,
+      "loss": 0.4976,
+      "num_input_tokens_seen": 131738856,
+      "step": 2285,
+      "train_runtime": 12816.1889,
+      "train_tokens_per_second": 10279.098
+    },
+    {
+      "epoch": 0.5269213069489186,
+      "grad_norm": 1.0411751308147925,
+      "learning_rate": 8.384237951914359e-06,
+      "loss": 0.4927,
+      "num_input_tokens_seen": 132027704,
+      "step": 2290,
+      "train_runtime": 12847.8662,
+      "train_tokens_per_second": 10276.236
+    },
+    {
+      "epoch": 0.5280717901518638,
+      "grad_norm": 0.8067060955920581,
+      "learning_rate": 8.377580912994325e-06,
+      "loss": 0.4556,
+      "num_input_tokens_seen": 132301264,
+      "step": 2295,
+      "train_runtime": 12872.1189,
+      "train_tokens_per_second": 10278.126
+    },
+    {
+      "epoch": 0.529222273354809,
+      "grad_norm": 0.8672196279381705,
+      "learning_rate": 8.370912843300914e-06,
+      "loss": 0.4738,
+      "num_input_tokens_seen": 132592584,
+      "step": 2300,
+      "train_runtime": 12893.0263,
+      "train_tokens_per_second": 10284.054
+    },
+    {
+      "epoch": 0.5303727565577543,
+      "grad_norm": 0.769152145901695,
+      "learning_rate": 8.36423376461124e-06,
+      "loss": 0.4717,
+      "num_input_tokens_seen": 132897248,
+      "step": 2305,
+      "train_runtime": 12920.8464,
+      "train_tokens_per_second": 10285.491
+    },
+    {
+      "epoch": 0.5315232397606995,
+      "grad_norm": 0.9246883528855874,
+      "learning_rate": 8.357543698738371e-06,
+      "loss": 0.5172,
+      "num_input_tokens_seen": 133165816,
+      "step": 2310,
+      "train_runtime": 12943.1746,
+      "train_tokens_per_second": 10288.497
+    },
+    {
+      "epoch": 0.5326737229636447,
+      "grad_norm": 0.826725739890336,
+      "learning_rate": 8.350842667531256e-06,
+      "loss": 0.4837,
+      "num_input_tokens_seen": 133471208,
+      "step": 2315,
+      "train_runtime": 12970.0721,
+      "train_tokens_per_second": 10290.707
+    },
+    {
+      "epoch": 0.53382420616659,
+      "grad_norm": 0.7440415387626675,
+      "learning_rate": 8.344130692874661e-06,
+      "loss": 0.4843,
+      "num_input_tokens_seen": 133781688,
+      "step": 2320,
+      "train_runtime": 12993.9646,
+      "train_tokens_per_second": 10295.679
+    },
+    {
+      "epoch": 0.5349746893695352,
+      "grad_norm": 0.7949999731027044,
+      "learning_rate": 8.337407796689088e-06,
+      "loss": 0.4608,
+      "num_input_tokens_seen": 134069136,
+      "step": 2325,
+      "train_runtime": 13014.7998,
+      "train_tokens_per_second": 10301.283
+    },
+    {
+      "epoch": 0.5361251725724805,
+      "grad_norm": 0.888947186410822,
+      "learning_rate": 8.330674000930708e-06,
+      "loss": 0.4999,
+      "num_input_tokens_seen": 134350312,
+      "step": 2330,
+      "train_runtime": 13042.0043,
+      "train_tokens_per_second": 10301.355
+    },
+    {
+      "epoch": 0.5372756557754257,
+      "grad_norm": 0.7708731732954736,
+      "learning_rate": 8.323929327591288e-06,
+      "loss": 0.4799,
+      "num_input_tokens_seen": 134637152,
+      "step": 2335,
+      "train_runtime": 13069.9878,
+      "train_tokens_per_second": 10301.245
+    },
+    {
+      "epoch": 0.5384261389783709,
+      "grad_norm": 0.7148356916758405,
+      "learning_rate": 8.317173798698118e-06,
+      "loss": 0.4954,
+      "num_input_tokens_seen": 134920752,
+      "step": 2340,
+      "train_runtime": 13095.2541,
+      "train_tokens_per_second": 10303.027
+    },
+    {
+      "epoch": 0.5395766221813162,
+      "grad_norm": 0.669236840716778,
+      "learning_rate": 8.310407436313947e-06,
+      "loss": 0.4896,
+      "num_input_tokens_seen": 135213512,
+      "step": 2345,
+      "train_runtime": 13127.8791,
+      "train_tokens_per_second": 10299.723
+    },
+    {
+      "epoch": 0.5407271053842614,
+      "grad_norm": 0.814514337034522,
+      "learning_rate": 8.303630262536903e-06,
+      "loss": 0.4628,
+      "num_input_tokens_seen": 135492168,
+      "step": 2350,
+      "train_runtime": 13159.8976,
+      "train_tokens_per_second": 10295.838
+    },
+    {
+      "epoch": 0.5418775885872066,
+      "grad_norm": 0.7142111277383983,
+      "learning_rate": 8.296842299500418e-06,
+      "loss": 0.4479,
+      "num_input_tokens_seen": 135795752,
+      "step": 2355,
+      "train_runtime": 13185.7083,
+      "train_tokens_per_second": 10298.707
+    },
+    {
+      "epoch": 0.5430280717901519,
+      "grad_norm": 0.7008684182726419,
+      "learning_rate": 8.290043569373163e-06,
+      "loss": 0.4716,
+      "num_input_tokens_seen": 136110656,
+      "step": 2360,
+      "train_runtime": 13208.0395,
+      "train_tokens_per_second": 10305.137
+    },
+    {
+      "epoch": 0.5441785549930971,
+      "grad_norm": 1.1712254340750716,
+      "learning_rate": 8.283234094358976e-06,
+      "loss": 0.5236,
+      "num_input_tokens_seen": 136378136,
+      "step": 2365,
+      "train_runtime": 13236.7343,
+      "train_tokens_per_second": 10303.005
+    },
+    {
+      "epoch": 0.5453290381960424,
+      "grad_norm": 0.8251085835576891,
+      "learning_rate": 8.276413896696784e-06,
+      "loss": 0.5001,
+      "num_input_tokens_seen": 136683800,
+      "step": 2370,
+      "train_runtime": 13268.4228,
+      "train_tokens_per_second": 10301.435
+    },
+    {
+      "epoch": 0.5464795213989876,
+      "grad_norm": 0.7488694353780002,
+      "learning_rate": 8.26958299866053e-06,
+      "loss": 0.499,
+      "num_input_tokens_seen": 136966896,
+      "step": 2375,
+      "train_runtime": 13299.5805,
+      "train_tokens_per_second": 10298.588
+    },
+    {
+      "epoch": 0.5476300046019328,
+      "grad_norm": 0.8428015493890508,
+      "learning_rate": 8.262741422559111e-06,
+      "loss": 0.4771,
+      "num_input_tokens_seen": 137237536,
+      "step": 2380,
+      "train_runtime": 13333.934,
+      "train_tokens_per_second": 10292.351
+    },
+    {
+      "epoch": 0.5487804878048781,
+      "grad_norm": 0.7928701793328784,
+      "learning_rate": 8.255889190736288e-06,
+      "loss": 0.5099,
+      "num_input_tokens_seen": 137527024,
+      "step": 2385,
+      "train_runtime": 13360.2048,
+      "train_tokens_per_second": 10293.781
+    },
+    {
+      "epoch": 0.5499309710078233,
+      "grad_norm": 0.8569843583238851,
+      "learning_rate": 8.249026325570628e-06,
+      "loss": 0.4706,
+      "num_input_tokens_seen": 137806776,
+      "step": 2390,
+      "train_runtime": 13391.066,
+      "train_tokens_per_second": 10290.949
+    },
+    {
+      "epoch": 0.5510814542107685,
+      "grad_norm": 0.8152874773802933,
+      "learning_rate": 8.242152849475425e-06,
+      "loss": 0.4776,
+      "num_input_tokens_seen": 138089336,
+      "step": 2395,
+      "train_runtime": 13413.7664,
+      "train_tokens_per_second": 10294.598
+    },
+    {
+      "epoch": 0.5522319374137138,
+      "grad_norm": 2.6140744255171993,
+      "learning_rate": 8.235268784898625e-06,
+      "loss": 0.4645,
+      "num_input_tokens_seen": 138371104,
+      "step": 2400,
+      "train_runtime": 13445.9525,
+      "train_tokens_per_second": 10290.911
+    },
+    {
+      "epoch": 0.553382420616659,
+      "grad_norm": 0.7995411909834026,
+      "learning_rate": 8.228374154322755e-06,
+      "loss": 0.4795,
+      "num_input_tokens_seen": 138662208,
+      "step": 2405,
+      "train_runtime": 13471.7142,
+      "train_tokens_per_second": 10292.841
+    },
+    {
+      "epoch": 0.5545329038196042,
+      "grad_norm": 0.9164760149468357,
+      "learning_rate": 8.221468980264853e-06,
+      "loss": 0.4588,
+      "num_input_tokens_seen": 138953448,
+      "step": 2410,
+      "train_runtime": 13498.1656,
+      "train_tokens_per_second": 10294.247
+    },
+    {
+      "epoch": 0.5556833870225495,
+      "grad_norm": 0.8566169683914129,
+      "learning_rate": 8.214553285276383e-06,
+      "loss": 0.4925,
+      "num_input_tokens_seen": 139246560,
+      "step": 2415,
+      "train_runtime": 13522.654,
+      "train_tokens_per_second": 10297.28
+    },
+    {
+      "epoch": 0.5568338702254947,
+      "grad_norm": 0.8426147596969124,
+      "learning_rate": 8.207627091943178e-06,
+      "loss": 0.4689,
+      "num_input_tokens_seen": 139502560,
+      "step": 2420,
+      "train_runtime": 13551.4589,
+      "train_tokens_per_second": 10294.283
+    },
+    {
+      "epoch": 0.55798435342844,
+      "grad_norm": 0.7084894478578524,
+      "learning_rate": 8.20069042288535e-06,
+      "loss": 0.4783,
+      "num_input_tokens_seen": 139797712,
+      "step": 2425,
+      "train_runtime": 13583.4693,
+      "train_tokens_per_second": 10291.753
+    },
+    {
+      "epoch": 0.5591348366313852,
+      "grad_norm": 0.8783424668874638,
+      "learning_rate": 8.193743300757227e-06,
+      "loss": 0.4737,
+      "num_input_tokens_seen": 140073056,
+      "step": 2430,
+      "train_runtime": 13608.9992,
+      "train_tokens_per_second": 10292.679
+    },
+    {
+      "epoch": 0.5602853198343304,
+      "grad_norm": 0.8585422412489129,
+      "learning_rate": 8.186785748247279e-06,
+      "loss": 0.4992,
+      "num_input_tokens_seen": 140341936,
+      "step": 2435,
+      "train_runtime": 13637.8392,
+      "train_tokens_per_second": 10290.628
+    },
+    {
+      "epoch": 0.5614358030372757,
+      "grad_norm": 0.845230118232004,
+      "learning_rate": 8.17981778807803e-06,
+      "loss": 0.4682,
+      "num_input_tokens_seen": 140642840,
+      "step": 2440,
+      "train_runtime": 13660.0548,
+      "train_tokens_per_second": 10295.921
+    },
+    {
+      "epoch": 0.5625862862402209,
+      "grad_norm": 1.2434885493077155,
+      "learning_rate": 8.172839443006006e-06,
+      "loss": 0.4659,
+      "num_input_tokens_seen": 140925704,
+      "step": 2445,
+      "train_runtime": 13688.66,
+      "train_tokens_per_second": 10295.069
+    },
+    {
+      "epoch": 0.5637367694431661,
+      "grad_norm": 0.8005649421382797,
+      "learning_rate": 8.16585073582164e-06,
+      "loss": 0.4829,
+      "num_input_tokens_seen": 141231280,
+      "step": 2450,
+      "train_runtime": 13717.2703,
+      "train_tokens_per_second": 10295.874
+    },
+    {
+      "epoch": 0.5648872526461114,
+      "grad_norm": 0.941448076970814,
+      "learning_rate": 8.158851689349213e-06,
+      "loss": 0.5128,
+      "num_input_tokens_seen": 141520048,
+      "step": 2455,
+      "train_runtime": 13749.5228,
+      "train_tokens_per_second": 10292.724
+    },
+    {
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.8374354407793042,
+      "learning_rate": 8.151842326446768e-06,
+      "loss": 0.5008,
+      "num_input_tokens_seen": 141810928,
+      "step": 2460,
+      "train_runtime": 13780.5098,
+      "train_tokens_per_second": 10290.688
+    },
+    {
+      "epoch": 0.5671882190520019,
+      "grad_norm": 0.7313258086391339,
+      "learning_rate": 8.14482267000604e-06,
+      "loss": 0.4601,
+      "num_input_tokens_seen": 142092912,
+      "step": 2465,
+      "train_runtime": 13808.9513,
+      "train_tokens_per_second": 10289.913
+    },
+    {
+      "epoch": 0.5683387022549471,
+      "grad_norm": 0.8334382151904006,
+      "learning_rate": 8.137792742952385e-06,
+      "loss": 0.4774,
+      "num_input_tokens_seen": 142382768,
+      "step": 2470,
+      "train_runtime": 13835.8784,
+      "train_tokens_per_second": 10290.837
+    },
+    {
+      "epoch": 0.5694891854578923,
+      "grad_norm": 0.7712356636304243,
+      "learning_rate": 8.130752568244703e-06,
+      "loss": 0.4565,
+      "num_input_tokens_seen": 142658824,
+      "step": 2475,
+      "train_runtime": 13860.4024,
+      "train_tokens_per_second": 10292.546
+    },
+    {
+      "epoch": 0.5706396686608376,
+      "grad_norm": 0.7352320192508478,
+      "learning_rate": 8.123702168875351e-06,
+      "loss": 0.4758,
+      "num_input_tokens_seen": 142947232,
+      "step": 2480,
+      "train_runtime": 13886.0748,
+      "train_tokens_per_second": 10294.286
+    },
+    {
+      "epoch": 0.5717901518637828,
+      "grad_norm": 0.8070847398761061,
+      "learning_rate": 8.116641567870093e-06,
+      "loss": 0.5029,
+      "num_input_tokens_seen": 143258096,
+      "step": 2485,
+      "train_runtime": 13922.1319,
+      "train_tokens_per_second": 10289.954
+    },
+    {
+      "epoch": 0.572940635066728,
+      "grad_norm": 1.7659228261154032,
+      "learning_rate": 8.109570788288004e-06,
+      "loss": 0.4989,
+      "num_input_tokens_seen": 143536376,
+      "step": 2490,
+      "train_runtime": 13953.8597,
+      "train_tokens_per_second": 10286.5
+    },
+    {
+      "epoch": 0.5740911182696733,
+      "grad_norm": 0.737341709892772,
+      "learning_rate": 8.102489853221394e-06,
+      "loss": 0.4872,
+      "num_input_tokens_seen": 143832568,
+      "step": 2495,
+      "train_runtime": 13981.8511,
+      "train_tokens_per_second": 10287.09
+    },
+    {
+      "epoch": 0.5752416014726185,
+      "grad_norm": 1.5817475992230332,
+      "learning_rate": 8.095398785795752e-06,
+      "loss": 0.4663,
+      "num_input_tokens_seen": 144120632,
+      "step": 2500,
+      "train_runtime": 14013.3093,
+      "train_tokens_per_second": 10284.554
+    },
+    {
+      "epoch": 0.5763920846755637,
+      "grad_norm": 0.7607111233972024,
+      "learning_rate": 8.088297609169648e-06,
+      "loss": 0.4943,
+      "num_input_tokens_seen": 144433408,
+      "step": 2505,
+      "train_runtime": 14039.6702,
+      "train_tokens_per_second": 10287.521
+    },
+    {
+      "epoch": 0.577542567878509,
+      "grad_norm": 0.8203251060352914,
+      "learning_rate": 8.081186346534673e-06,
+      "loss": 0.4857,
+      "num_input_tokens_seen": 144686328,
+      "step": 2510,
+      "train_runtime": 14064.943,
+      "train_tokens_per_second": 10287.018
+    },
+    {
+      "epoch": 0.5786930510814542,
+      "grad_norm": 0.8845586211026448,
+      "learning_rate": 8.074065021115358e-06,
+      "loss": 0.4798,
+      "num_input_tokens_seen": 144958960,
+      "step": 2515,
+      "train_runtime": 14089.9172,
+      "train_tokens_per_second": 10288.134
+    },
+    {
+      "epoch": 0.5798435342843995,
+      "grad_norm": 0.7491762146132817,
+      "learning_rate": 8.066933656169093e-06,
+      "loss": 0.4396,
+      "num_input_tokens_seen": 145238056,
+      "step": 2520,
+      "train_runtime": 14118.6139,
+      "train_tokens_per_second": 10286.991
+    },
+    {
+      "epoch": 0.5809940174873447,
+      "grad_norm": 1.0515507556507233,
+      "learning_rate": 8.059792274986062e-06,
+      "loss": 0.4885,
+      "num_input_tokens_seen": 145510864,
+      "step": 2525,
+      "train_runtime": 14142.621,
+      "train_tokens_per_second": 10288.819
+    },
+    {
+      "epoch": 0.5821445006902899,
+      "grad_norm": 0.818886767280279,
+      "learning_rate": 8.052640900889155e-06,
+      "loss": 0.482,
+      "num_input_tokens_seen": 145809688,
+      "step": 2530,
+      "train_runtime": 14171.9453,
+      "train_tokens_per_second": 10288.615
+    },
+    {
+      "epoch": 0.5832949838932352,
+      "grad_norm": 0.8769858204573509,
+      "learning_rate": 8.045479557233901e-06,
+      "loss": 0.4703,
+      "num_input_tokens_seen": 146106360,
+      "step": 2535,
+      "train_runtime": 14197.2849,
+      "train_tokens_per_second": 10291.148
+    },
+    {
+      "epoch": 0.5844454670961804,
+      "grad_norm": 0.7460645176963251,
+      "learning_rate": 8.03830826740839e-06,
+      "loss": 0.4733,
+      "num_input_tokens_seen": 146391240,
+      "step": 2540,
+      "train_runtime": 14224.5531,
+      "train_tokens_per_second": 10291.447
+    },
+    {
+      "epoch": 0.5855959502991256,
+      "grad_norm": 0.7848638257252075,
+      "learning_rate": 8.031127054833192e-06,
+      "loss": 0.4788,
+      "num_input_tokens_seen": 146670760,
+      "step": 2545,
+      "train_runtime": 14253.0718,
+      "train_tokens_per_second": 10290.467
+    },
+    {
+      "epoch": 0.5867464335020709,
+      "grad_norm": 0.8325851371204545,
+      "learning_rate": 8.02393594296128e-06,
+      "loss": 0.4651,
+      "num_input_tokens_seen": 146930896,
+      "step": 2550,
+      "train_runtime": 14277.6652,
+      "train_tokens_per_second": 10290.961
+    },
+    {
+      "epoch": 0.5878969167050161,
+      "grad_norm": 0.8121274999421523,
+      "learning_rate": 8.016734955277968e-06,
+      "loss": 0.4559,
+      "num_input_tokens_seen": 147170360,
+      "step": 2555,
+      "train_runtime": 14304.2441,
+      "train_tokens_per_second": 10288.58
+    },
+    {
+      "epoch": 0.5890473999079613,
+      "grad_norm": 0.7713628312397729,
+      "learning_rate": 8.009524115300812e-06,
+      "loss": 0.4379,
+      "num_input_tokens_seen": 147471152,
+      "step": 2560,
+      "train_runtime": 14339.365,
+      "train_tokens_per_second": 10284.357
+    },
+    {
+      "epoch": 0.5901978831109066,
+      "grad_norm": 0.8181328108878305,
+      "learning_rate": 8.002303446579549e-06,
+      "loss": 0.4668,
+      "num_input_tokens_seen": 147750248,
+      "step": 2565,
+      "train_runtime": 14370.5852,
+      "train_tokens_per_second": 10281.436
+    },
+    {
+      "epoch": 0.5913483663138518,
+      "grad_norm": 0.8243589244872646,
+      "learning_rate": 7.995072972696019e-06,
+      "loss": 0.5205,
+      "num_input_tokens_seen": 148045680,
+      "step": 2570,
+      "train_runtime": 14394.0612,
+      "train_tokens_per_second": 10285.192
+    },
+    {
+      "epoch": 0.5924988495167971,
+      "grad_norm": 0.7703839835001517,
+      "learning_rate": 7.987832717264075e-06,
+      "loss": 0.4678,
+      "num_input_tokens_seen": 148335592,
+      "step": 2575,
+      "train_runtime": 14422.4783,
+      "train_tokens_per_second": 10285.028
+    },
+    {
+      "epoch": 0.5936493327197423,
+      "grad_norm": 0.7856537927194422,
+      "learning_rate": 7.980582703929526e-06,
+      "loss": 0.475,
+      "num_input_tokens_seen": 148610168,
+      "step": 2580,
+      "train_runtime": 14456.994,
+      "train_tokens_per_second": 10279.465
+    },
+    {
+      "epoch": 0.5947998159226875,
+      "grad_norm": 0.7331569332643155,
+      "learning_rate": 7.973322956370043e-06,
+      "loss": 0.4811,
+      "num_input_tokens_seen": 148920112,
+      "step": 2585,
+      "train_runtime": 14478.842,
+      "train_tokens_per_second": 10285.361
+    },
+    {
+      "epoch": 0.5959502991256328,
+      "grad_norm": 0.7650306448874212,
+      "learning_rate": 7.966053498295088e-06,
+      "loss": 0.4624,
+      "num_input_tokens_seen": 149188552,
+      "step": 2590,
+      "train_runtime": 14510.3998,
+      "train_tokens_per_second": 10281.491
+    },
+    {
+      "epoch": 0.597100782328578,
+      "grad_norm": 0.8715629075406254,
+      "learning_rate": 7.958774353445839e-06,
+      "loss": 0.49,
+      "num_input_tokens_seen": 149471752,
+      "step": 2595,
+      "train_runtime": 14537.6645,
+      "train_tokens_per_second": 10281.689
+    },
+    {
+      "epoch": 0.5982512655315232,
+      "grad_norm": 0.7044538339093126,
+      "learning_rate": 7.951485545595108e-06,
+      "loss": 0.4777,
+      "num_input_tokens_seen": 149762392,
+      "step": 2600,
+      "train_runtime": 14562.4152,
+      "train_tokens_per_second": 10284.173
+    },
+    {
+      "epoch": 0.5994017487344685,
+      "grad_norm": 0.7022848233009673,
+      "learning_rate": 7.944187098547263e-06,
+      "loss": 0.4824,
+      "num_input_tokens_seen": 150038216,
+      "step": 2605,
+      "train_runtime": 14588.5751,
+      "train_tokens_per_second": 10284.638
+    },
+    {
+      "epoch": 0.6005522319374137,
+      "grad_norm": 0.8519837914025554,
+      "learning_rate": 7.936879036138159e-06,
+      "loss": 0.4925,
+      "num_input_tokens_seen": 150314152,
+      "step": 2610,
+      "train_runtime": 14617.0006,
+      "train_tokens_per_second": 10283.515
+    },
+    {
+      "epoch": 0.601702715140359,
+      "grad_norm": 0.9426180657822917,
+      "learning_rate": 7.929561382235047e-06,
+      "loss": 0.475,
+      "num_input_tokens_seen": 150591984,
+      "step": 2615,
+      "train_runtime": 14646.2873,
+      "train_tokens_per_second": 10281.922
+    },
+    {
+      "epoch": 0.6028531983433042,
+      "grad_norm": 0.730806866308507,
+      "learning_rate": 7.922234160736506e-06,
+      "loss": 0.5011,
+      "num_input_tokens_seen": 150887568,
+      "step": 2620,
+      "train_runtime": 14673.0129,
+      "train_tokens_per_second": 10283.339
+    },
+    {
+      "epoch": 0.6040036815462494,
+      "grad_norm": 0.7065189186379442,
+      "learning_rate": 7.914897395572362e-06,
+      "loss": 0.4817,
+      "num_input_tokens_seen": 151196152,
+      "step": 2625,
+      "train_runtime": 14709.8283,
+      "train_tokens_per_second": 10278.58
+    },
+    {
+      "epoch": 0.6051541647491947,
+      "grad_norm": 0.8697540000279285,
+      "learning_rate": 7.907551110703606e-06,
+      "loss": 0.4732,
+      "num_input_tokens_seen": 151504544,
+      "step": 2630,
+      "train_runtime": 14736.8939,
+      "train_tokens_per_second": 10280.629
+    },
+    {
+      "epoch": 0.6063046479521399,
+      "grad_norm": 0.70468070846114,
+      "learning_rate": 7.900195330122323e-06,
+      "loss": 0.4933,
+      "num_input_tokens_seen": 151802824,
+      "step": 2635,
+      "train_runtime": 14765.1312,
+      "train_tokens_per_second": 10281.17
+    },
+    {
+      "epoch": 0.6074551311550851,
+      "grad_norm": 0.7193794370004378,
+      "learning_rate": 7.89283007785161e-06,
+      "loss": 0.4498,
+      "num_input_tokens_seen": 152093648,
+      "step": 2640,
+      "train_runtime": 14791.269,
+      "train_tokens_per_second": 10282.664
+    },
+    {
+      "epoch": 0.6086056143580304,
+      "grad_norm": 0.7068332016637432,
+      "learning_rate": 7.885455377945492e-06,
+      "loss": 0.4905,
+      "num_input_tokens_seen": 152384384,
+      "step": 2645,
+      "train_runtime": 14819.3831,
+      "train_tokens_per_second": 10282.775
+    },
+    {
+      "epoch": 0.6097560975609756,
+      "grad_norm": 0.7270294977407151,
+      "learning_rate": 7.878071254488856e-06,
+      "loss": 0.4764,
+      "num_input_tokens_seen": 152693016,
+      "step": 2650,
+      "train_runtime": 14844.7978,
+      "train_tokens_per_second": 10285.961
+    },
+    {
+      "epoch": 0.6109065807639208,
+      "grad_norm": 2.1614713302819744,
+      "learning_rate": 7.87067773159736e-06,
+      "loss": 0.4667,
+      "num_input_tokens_seen": 152963328,
+      "step": 2655,
+      "train_runtime": 14875.4365,
+      "train_tokens_per_second": 10282.947
+    },
+    {
+      "epoch": 0.6120570639668661,
+      "grad_norm": 0.7856742266883362,
+      "learning_rate": 7.863274833417362e-06,
+      "loss": 0.4814,
+      "num_input_tokens_seen": 153254040,
+      "step": 2660,
+      "train_runtime": 14906.177,
+      "train_tokens_per_second": 10281.244
+    },
+    {
+      "epoch": 0.6132075471698113,
+      "grad_norm": 0.7210873958258907,
+      "learning_rate": 7.85586258412584e-06,
+      "loss": 0.4647,
+      "num_input_tokens_seen": 153546032,
+      "step": 2665,
+      "train_runtime": 14928.5622,
+      "train_tokens_per_second": 10285.386
+    },
+    {
+      "epoch": 0.6143580303727566,
+      "grad_norm": 0.8593699114532515,
+      "learning_rate": 7.848441007930305e-06,
+      "loss": 0.4882,
+      "num_input_tokens_seen": 153824856,
+      "step": 2670,
+      "train_runtime": 14958.9035,
+      "train_tokens_per_second": 10283.164
+    },
+    {
+      "epoch": 0.6155085135757018,
+      "grad_norm": 0.7658888484510346,
+      "learning_rate": 7.841010129068736e-06,
+      "loss": 0.4691,
+      "num_input_tokens_seen": 154125600,
+      "step": 2675,
+      "train_runtime": 14983.8854,
+      "train_tokens_per_second": 10286.09
+    },
+    {
+      "epoch": 0.616658996778647,
+      "grad_norm": 0.707603839221591,
+      "learning_rate": 7.83356997180949e-06,
+      "loss": 0.465,
+      "num_input_tokens_seen": 154418104,
+      "step": 2680,
+      "train_runtime": 15008.6045,
+      "train_tokens_per_second": 10288.638
+    },
+    {
+      "epoch": 0.6178094799815923,
+      "grad_norm": 1.785362468053383,
+      "learning_rate": 7.826120560451228e-06,
+      "loss": 0.4916,
+      "num_input_tokens_seen": 154676208,
+      "step": 2685,
+      "train_runtime": 15037.0578,
+      "train_tokens_per_second": 10286.335
+    },
+    {
+      "epoch": 0.6189599631845375,
+      "grad_norm": 0.8039408761153178,
+      "learning_rate": 7.818661919322827e-06,
+      "loss": 0.48,
+      "num_input_tokens_seen": 154973512,
+      "step": 2690,
+      "train_runtime": 15066.7182,
+      "train_tokens_per_second": 10285.817
+    },
+    {
+      "epoch": 0.6201104463874827,
+      "grad_norm": 0.782525252231477,
+      "learning_rate": 7.811194072783318e-06,
+      "loss": 0.4617,
+      "num_input_tokens_seen": 155281256,
+      "step": 2695,
+      "train_runtime": 15093.589,
+      "train_tokens_per_second": 10287.895
+    },
+    {
+      "epoch": 0.621260929590428,
+      "grad_norm": 0.8355791332912936,
+      "learning_rate": 7.803717045221787e-06,
+      "loss": 0.4824,
+      "num_input_tokens_seen": 155583608,
+      "step": 2700,
+      "train_runtime": 15120.1599,
+      "train_tokens_per_second": 10289.812
+    },
+    {
+      "epoch": 0.6224114127933732,
+      "grad_norm": 0.7445617894859013,
+      "learning_rate": 7.79623086105731e-06,
+      "loss": 0.4613,
+      "num_input_tokens_seen": 155862336,
+      "step": 2705,
+      "train_runtime": 15145.2224,
+      "train_tokens_per_second": 10291.188
+    },
+    {
+      "epoch": 0.6235618959963185,
+      "grad_norm": 0.8414736653365485,
+      "learning_rate": 7.788735544738865e-06,
+      "loss": 0.5013,
+      "num_input_tokens_seen": 156153936,
+      "step": 2710,
+      "train_runtime": 15173.2791,
+      "train_tokens_per_second": 10291.377
+    },
+    {
+      "epoch": 0.6247123791992637,
+      "grad_norm": 0.7626440557612986,
+      "learning_rate": 7.781231120745252e-06,
+      "loss": 0.4886,
+      "num_input_tokens_seen": 156438400,
+      "step": 2715,
+      "train_runtime": 15197.086,
+      "train_tokens_per_second": 10293.973
+    },
+    {
+      "epoch": 0.6258628624022089,
+      "grad_norm": 1.224384718123704,
+      "learning_rate": 7.773717613585022e-06,
+      "loss": 0.4668,
+      "num_input_tokens_seen": 156727872,
+      "step": 2720,
+      "train_runtime": 15225.2944,
+      "train_tokens_per_second": 10293.914
+    },
+    {
+      "epoch": 0.6270133456051542,
+      "grad_norm": 0.8127655130055158,
+      "learning_rate": 7.766195047796378e-06,
+      "loss": 0.497,
+      "num_input_tokens_seen": 157036480,
+      "step": 2725,
+      "train_runtime": 15257.6424,
+      "train_tokens_per_second": 10292.316
+    },
+    {
+      "epoch": 0.6281638288080994,
+      "grad_norm": 0.7379688476961642,
+      "learning_rate": 7.758663447947126e-06,
+      "loss": 0.4956,
+      "num_input_tokens_seen": 157352424,
+      "step": 2730,
+      "train_runtime": 15278.8381,
+      "train_tokens_per_second": 10298.717
+    },
+    {
+      "epoch": 0.6293143120110446,
+      "grad_norm": 0.726882494646803,
+      "learning_rate": 7.751122838634561e-06,
+      "loss": 0.4772,
+      "num_input_tokens_seen": 157662144,
+      "step": 2735,
+      "train_runtime": 15305.0216,
+      "train_tokens_per_second": 10301.334
+    },
+    {
+      "epoch": 0.6304647952139899,
+      "grad_norm": 0.6883281947666869,
+      "learning_rate": 7.743573244485407e-06,
+      "loss": 0.4711,
+      "num_input_tokens_seen": 157957504,
+      "step": 2740,
+      "train_runtime": 15340.3157,
+      "train_tokens_per_second": 10296.887
+    },
+    {
+      "epoch": 0.6316152784169351,
+      "grad_norm": 0.8921517207391104,
+      "learning_rate": 7.736014690155732e-06,
+      "loss": 0.4877,
+      "num_input_tokens_seen": 158218584,
+      "step": 2745,
+      "train_runtime": 15368.2994,
+      "train_tokens_per_second": 10295.126
+    },
+    {
+      "epoch": 0.6327657616198803,
+      "grad_norm": 0.7091802726635639,
+      "learning_rate": 7.728447200330867e-06,
+      "loss": 0.4879,
+      "num_input_tokens_seen": 158502904,
+      "step": 2750,
+      "train_runtime": 15400.138,
+      "train_tokens_per_second": 10292.304
+    },
+    {
+      "epoch": 0.6339162448228256,
+      "grad_norm": 0.7619802399813204,
+      "learning_rate": 7.72087079972532e-06,
+      "loss": 0.4528,
+      "num_input_tokens_seen": 158779224,
+      "step": 2755,
+      "train_runtime": 15426.1328,
+      "train_tokens_per_second": 10292.873
+    },
+    {
+      "epoch": 0.6350667280257708,
+      "grad_norm": 0.7675457864693662,
+      "learning_rate": 7.713285513082712e-06,
+      "loss": 0.4763,
+      "num_input_tokens_seen": 159084808,
+      "step": 2760,
+      "train_runtime": 15455.2122,
+      "train_tokens_per_second": 10293.279
+    },
+    {
+      "epoch": 0.6362172112287161,
+      "grad_norm": 0.8231660869702927,
+      "learning_rate": 7.705691365175672e-06,
+      "loss": 0.4381,
+      "num_input_tokens_seen": 159353976,
+      "step": 2765,
+      "train_runtime": 15488.8266,
+      "train_tokens_per_second": 10288.318
+    },
+    {
+      "epoch": 0.6373676944316613,
+      "grad_norm": 0.792394074858804,
+      "learning_rate": 7.698088380805776e-06,
+      "loss": 0.4752,
+      "num_input_tokens_seen": 159663960,
+      "step": 2770,
+      "train_runtime": 15511.2691,
+      "train_tokens_per_second": 10293.417
+    },
+    {
+      "epoch": 0.6385181776346065,
+      "grad_norm": 0.7664011052571271,
+      "learning_rate": 7.690476584803457e-06,
+      "loss": 0.4793,
+      "num_input_tokens_seen": 159931840,
+      "step": 2775,
+      "train_runtime": 15542.0803,
+      "train_tokens_per_second": 10290.247
+    },
+    {
+      "epoch": 0.6396686608375518,
+      "grad_norm": 0.871565718038199,
+      "learning_rate": 7.68285600202793e-06,
+      "loss": 0.4792,
+      "num_input_tokens_seen": 160198128,
+      "step": 2780,
+      "train_runtime": 15565.9527,
+      "train_tokens_per_second": 10291.572
+    },
+    {
+      "epoch": 0.640819144040497,
+      "grad_norm": 0.6629992029953665,
+      "learning_rate": 7.675226657367095e-06,
+      "loss": 0.4593,
+      "num_input_tokens_seen": 160503400,
+      "step": 2785,
+      "train_runtime": 15591.9891,
+      "train_tokens_per_second": 10293.966
+    },
+    {
+      "epoch": 0.6419696272434422,
+      "grad_norm": 0.7786272539521918,
+      "learning_rate": 7.667588575737481e-06,
+      "loss": 0.4656,
+      "num_input_tokens_seen": 160787848,
+      "step": 2790,
+      "train_runtime": 15617.9615,
+      "train_tokens_per_second": 10295.06
+    },
+    {
+      "epoch": 0.6431201104463875,
+      "grad_norm": 0.6874046880638893,
+      "learning_rate": 7.659941782084145e-06,
+      "loss": 0.4855,
+      "num_input_tokens_seen": 161100808,
+      "step": 2795,
+      "train_runtime": 15646.4669,
+      "train_tokens_per_second": 10296.306
+    },
+    {
+      "epoch": 0.6442705936493327,
+      "grad_norm": 0.8440148528448385,
+      "learning_rate": 7.652286301380589e-06,
+      "loss": 0.4855,
+      "num_input_tokens_seen": 161401664,
+      "step": 2800,
+      "train_runtime": 15678.3117,
+      "train_tokens_per_second": 10294.582
+    },
+    {
+      "epoch": 0.645421076852278,
+      "grad_norm": 0.8125490610247502,
+      "learning_rate": 7.644622158628701e-06,
+      "loss": 0.4618,
+      "num_input_tokens_seen": 161697824,
+      "step": 2805,
+      "train_runtime": 15701.9182,
+      "train_tokens_per_second": 10297.966
+    },
+    {
+      "epoch": 0.6465715600552232,
+      "grad_norm": 0.8423141738607763,
+      "learning_rate": 7.636949378858647e-06,
+      "loss": 0.4881,
+      "num_input_tokens_seen": 162008296,
+      "step": 2810,
+      "train_runtime": 15734.7061,
+      "train_tokens_per_second": 10296.239
+    },
+    {
+      "epoch": 0.6477220432581684,
+      "grad_norm": 0.8137872371898669,
+      "learning_rate": 7.629267987128806e-06,
+      "loss": 0.4591,
+      "num_input_tokens_seen": 162299296,
+      "step": 2815,
+      "train_runtime": 15762.6228,
+      "train_tokens_per_second": 10296.465
+    },
+    {
+      "epoch": 0.6488725264611137,
+      "grad_norm": 0.8196326229087935,
+      "learning_rate": 7.6215780085256744e-06,
+      "loss": 0.5078,
+      "num_input_tokens_seen": 162566888,
+      "step": 2820,
+      "train_runtime": 15793.971,
+      "train_tokens_per_second": 10292.971
+    },
+    {
+      "epoch": 0.6500230096640589,
+      "grad_norm": 0.772116835624322,
+      "learning_rate": 7.613879468163804e-06,
+      "loss": 0.4846,
+      "num_input_tokens_seen": 162861424,
+      "step": 2825,
+      "train_runtime": 15815.5915,
+      "train_tokens_per_second": 10297.523
+    },
+    {
+      "epoch": 0.6511734928670041,
+      "grad_norm": 0.8149364019337492,
+      "learning_rate": 7.6061723911857e-06,
+      "loss": 0.4793,
+      "num_input_tokens_seen": 163145280,
+      "step": 2830,
+      "train_runtime": 15843.8986,
+      "train_tokens_per_second": 10297.041
+    },
+    {
+      "epoch": 0.6523239760699494,
+      "grad_norm": 0.7460478889975525,
+      "learning_rate": 7.598456802761751e-06,
+      "loss": 0.4445,
+      "num_input_tokens_seen": 163441080,
+      "step": 2835,
+      "train_runtime": 15870.8654,
+      "train_tokens_per_second": 10298.183
+    },
+    {
+      "epoch": 0.6534744592728946,
+      "grad_norm": 0.8281016659249562,
+      "learning_rate": 7.59073272809014e-06,
+      "loss": 0.4855,
+      "num_input_tokens_seen": 163725680,
+      "step": 2840,
+      "train_runtime": 15905.0885,
+      "train_tokens_per_second": 10293.918
+    },
+    {
+      "epoch": 0.6546249424758398,
+      "grad_norm": 0.9851645488361443,
+      "learning_rate": 7.583000192396768e-06,
+      "loss": 0.4981,
+      "num_input_tokens_seen": 164011696,
+      "step": 2845,
+      "train_runtime": 15931.6996,
+      "train_tokens_per_second": 10294.677
+    },
+    {
+      "epoch": 0.6557754256787851,
+      "grad_norm": 0.8587627130319571,
+      "learning_rate": 7.575259220935166e-06,
+      "loss": 0.4716,
+      "num_input_tokens_seen": 164314056,
+      "step": 2850,
+      "train_runtime": 15964.6825,
+      "train_tokens_per_second": 10292.347
+    },
+    {
+      "epoch": 0.6569259088817303,
+      "grad_norm": 0.7320988242650495,
+      "learning_rate": 7.567509838986418e-06,
+      "loss": 0.4579,
+      "num_input_tokens_seen": 164613256,
+      "step": 2855,
+      "train_runtime": 15992.9105,
+      "train_tokens_per_second": 10292.889
+    },
+    {
+      "epoch": 0.6580763920846756,
+      "grad_norm": 0.7919951938004727,
+      "learning_rate": 7.559752071859072e-06,
+      "loss": 0.4777,
+      "num_input_tokens_seen": 164884144,
+      "step": 2860,
+      "train_runtime": 16020.7185,
+      "train_tokens_per_second": 10291.932
+    },
+    {
+      "epoch": 0.6592268752876208,
+      "grad_norm": 0.8242557946649387,
+      "learning_rate": 7.551985944889068e-06,
+      "loss": 0.4614,
+      "num_input_tokens_seen": 165177552,
+      "step": 2865,
+      "train_runtime": 16049.2102,
+      "train_tokens_per_second": 10291.943
+    },
+    {
+      "epoch": 0.660377358490566,
+      "grad_norm": 0.9913185807446188,
+      "learning_rate": 7.544211483439639e-06,
+      "loss": 0.4601,
+      "num_input_tokens_seen": 165457808,
+      "step": 2870,
+      "train_runtime": 16078.293,
+      "train_tokens_per_second": 10290.757
+    },
+    {
+      "epoch": 0.6615278416935113,
+      "grad_norm": 0.7950093944399945,
+      "learning_rate": 7.536428712901242e-06,
+      "loss": 0.4888,
+      "num_input_tokens_seen": 165719248,
+      "step": 2875,
+      "train_runtime": 16105.0687,
+      "train_tokens_per_second": 10289.881
+    },
+    {
+      "epoch": 0.6626783248964565,
+      "grad_norm": 0.7606567026444985,
+      "learning_rate": 7.528637658691471e-06,
+      "loss": 0.4643,
+      "num_input_tokens_seen": 166023760,
+      "step": 2880,
+      "train_runtime": 16129.6915,
+      "train_tokens_per_second": 10293.052
+    },
+    {
+      "epoch": 0.6638288080994017,
+      "grad_norm": 0.7613630545037862,
+      "learning_rate": 7.520838346254975e-06,
+      "loss": 0.4782,
+      "num_input_tokens_seen": 166311440,
+      "step": 2885,
+      "train_runtime": 16161.252,
+      "train_tokens_per_second": 10290.752
+    },
+    {
+      "epoch": 0.664979291302347,
+      "grad_norm": 0.9432510044320658,
+      "learning_rate": 7.513030801063367e-06,
+      "loss": 0.4633,
+      "num_input_tokens_seen": 166576208,
+      "step": 2890,
+      "train_runtime": 16187.3807,
+      "train_tokens_per_second": 10290.498
+    },
+    {
+      "epoch": 0.6661297745052922,
+      "grad_norm": 0.8118383212974096,
+      "learning_rate": 7.505215048615155e-06,
+      "loss": 0.4811,
+      "num_input_tokens_seen": 166880784,
+      "step": 2895,
+      "train_runtime": 16210.1909,
+      "train_tokens_per_second": 10294.807
+    },
+    {
+      "epoch": 0.6672802577082375,
+      "grad_norm": 0.7932428123434949,
+      "learning_rate": 7.497391114435646e-06,
+      "loss": 0.4763,
+      "num_input_tokens_seen": 167165752,
+      "step": 2900,
+      "train_runtime": 16232.2705,
+      "train_tokens_per_second": 10298.359
+    },
+    {
+      "epoch": 0.6684307409111827,
+      "grad_norm": 0.8636575653479661,
+      "learning_rate": 7.489559024076869e-06,
+      "loss": 0.4575,
+      "num_input_tokens_seen": 167435976,
+      "step": 2905,
+      "train_runtime": 16258.1474,
+      "train_tokens_per_second": 10298.589
+    },
+    {
+      "epoch": 0.6695812241141279,
+      "grad_norm": 0.8433378196935765,
+      "learning_rate": 7.481718803117489e-06,
+      "loss": 0.4707,
+      "num_input_tokens_seen": 167722072,
+      "step": 2910,
+      "train_runtime": 16286.3196,
+      "train_tokens_per_second": 10298.341
+    },
+    {
+      "epoch": 0.6707317073170732,
+      "grad_norm": 0.7151813514006768,
+      "learning_rate": 7.473870477162726e-06,
+      "loss": 0.4814,
+      "num_input_tokens_seen": 167979720,
+      "step": 2915,
+      "train_runtime": 16308.3504,
+      "train_tokens_per_second": 10300.228
+    },
+    {
+      "epoch": 0.6718821905200184,
+      "grad_norm": 0.8903148294675728,
+      "learning_rate": 7.466014071844271e-06,
+      "loss": 0.4436,
+      "num_input_tokens_seen": 168270080,
+      "step": 2920,
+      "train_runtime": 16335.0199,
+      "train_tokens_per_second": 10301.186
+    },
+    {
+      "epoch": 0.6730326737229636,
+      "grad_norm": 0.7288330605630249,
+      "learning_rate": 7.458149612820198e-06,
+      "loss": 0.4716,
+      "num_input_tokens_seen": 168544208,
+      "step": 2925,
+      "train_runtime": 16360.9357,
+      "train_tokens_per_second": 10301.624
+    },
+    {
+      "epoch": 0.6741831569259089,
+      "grad_norm": 0.842791517345471,
+      "learning_rate": 7.450277125774884e-06,
+      "loss": 0.4629,
+      "num_input_tokens_seen": 168837328,
+      "step": 2930,
+      "train_runtime": 16390.173,
+      "train_tokens_per_second": 10301.132
+    },
+    {
+      "epoch": 0.6753336401288541,
+      "grad_norm": 0.9913387669168312,
+      "learning_rate": 7.442396636418929e-06,
+      "loss": 0.4877,
+      "num_input_tokens_seen": 169131456,
+      "step": 2935,
+      "train_runtime": 16418.1572,
+      "train_tokens_per_second": 10301.488
+    },
+    {
+      "epoch": 0.6764841233317993,
+      "grad_norm": 0.7425296134159732,
+      "learning_rate": 7.434508170489061e-06,
+      "loss": 0.495,
+      "num_input_tokens_seen": 169413392,
+      "step": 2940,
+      "train_runtime": 16446.132,
+      "train_tokens_per_second": 10301.109
+    },
+    {
+      "epoch": 0.6776346065347446,
+      "grad_norm": 0.8041605441667194,
+      "learning_rate": 7.426611753748063e-06,
+      "loss": 0.4571,
+      "num_input_tokens_seen": 169691840,
+      "step": 2945,
+      "train_runtime": 16476.8458,
+      "train_tokens_per_second": 10298.806
+    },
+    {
+      "epoch": 0.6787850897376898,
+      "grad_norm": 0.7830557095790432,
+      "learning_rate": 7.418707411984684e-06,
+      "loss": 0.4682,
+      "num_input_tokens_seen": 169990464,
+      "step": 2950,
+      "train_runtime": 16505.5574,
+      "train_tokens_per_second": 10298.984
+    },
+    {
+      "epoch": 0.6799355729406351,
+      "grad_norm": 0.7303823685171997,
+      "learning_rate": 7.410795171013553e-06,
+      "loss": 0.4915,
+      "num_input_tokens_seen": 170290048,
+      "step": 2955,
+      "train_runtime": 16531.5644,
+      "train_tokens_per_second": 10300.903
+    },
+    {
+      "epoch": 0.6810860561435803,
+      "grad_norm": 0.8336568631551833,
+      "learning_rate": 7.402875056675099e-06,
+      "loss": 0.4804,
+      "num_input_tokens_seen": 170567240,
+      "step": 2960,
+      "train_runtime": 16553.4489,
+      "train_tokens_per_second": 10304.03
+    },
+    {
+      "epoch": 0.6822365393465255,
+      "grad_norm": 0.8219932058263919,
+      "learning_rate": 7.394947094835464e-06,
+      "loss": 0.4694,
+      "num_input_tokens_seen": 170867528,
+      "step": 2965,
+      "train_runtime": 16583.3364,
+      "train_tokens_per_second": 10303.568
+    },
+    {
+      "epoch": 0.6833870225494708,
+      "grad_norm": 0.7567510838852517,
+      "learning_rate": 7.387011311386417e-06,
+      "loss": 0.4737,
+      "num_input_tokens_seen": 171160328,
+      "step": 2970,
+      "train_runtime": 16616.1562,
+      "train_tokens_per_second": 10300.838
+    },
+    {
+      "epoch": 0.684537505752416,
+      "grad_norm": 0.7700466144506856,
+      "learning_rate": 7.379067732245275e-06,
+      "loss": 0.4727,
+      "num_input_tokens_seen": 171461360,
+      "step": 2975,
+      "train_runtime": 16640.8127,
+      "train_tokens_per_second": 10303.665
+    },
+    {
+      "epoch": 0.6856879889553612,
+      "grad_norm": 0.7532154286354551,
+      "learning_rate": 7.371116383354813e-06,
+      "loss": 0.4581,
+      "num_input_tokens_seen": 171743240,
+      "step": 2980,
+      "train_runtime": 16669.4797,
+      "train_tokens_per_second": 10302.855
+    },
+    {
+      "epoch": 0.6868384721583065,
+      "grad_norm": 0.7314518122957149,
+      "learning_rate": 7.363157290683177e-06,
+      "loss": 0.4644,
+      "num_input_tokens_seen": 172019072,
+      "step": 2985,
+      "train_runtime": 16694.3937,
+      "train_tokens_per_second": 10304.002
+    },
+    {
+      "epoch": 0.6879889553612517,
+      "grad_norm": 0.7569621620319172,
+      "learning_rate": 7.355190480223813e-06,
+      "loss": 0.456,
+      "num_input_tokens_seen": 172288360,
+      "step": 2990,
+      "train_runtime": 16722.324,
+      "train_tokens_per_second": 10302.896
+    },
+    {
+      "epoch": 0.689139438564197,
+      "grad_norm": 0.7581639272626943,
+      "learning_rate": 7.347215977995363e-06,
+      "loss": 0.4835,
+      "num_input_tokens_seen": 172582536,
+      "step": 2995,
+      "train_runtime": 16752.1607,
+      "train_tokens_per_second": 10302.106
+    },
+    {
+      "epoch": 0.6902899217671422,
+      "grad_norm": 0.7117120743178253,
+      "learning_rate": 7.339233810041594e-06,
+      "loss": 0.4622,
+      "num_input_tokens_seen": 172844272,
+      "step": 3000,
+      "train_runtime": 16782.0132,
+      "train_tokens_per_second": 10299.377
+    },
+    {
+      "epoch": 0.6914404049700874,
+      "grad_norm": 1.4929587094258303,
+      "learning_rate": 7.331244002431308e-06,
+      "loss": 0.4769,
+      "num_input_tokens_seen": 173126088,
+      "step": 3005,
+      "train_runtime": 16811.5302,
+      "train_tokens_per_second": 10298.057
+    },
+    {
+      "epoch": 0.6925908881730327,
+      "grad_norm": 0.7663591356345228,
+      "learning_rate": 7.3232465812582535e-06,
+      "loss": 0.4696,
+      "num_input_tokens_seen": 173442752,
+      "step": 3010,
+      "train_runtime": 16842.8313,
+      "train_tokens_per_second": 10297.719
+    },
+    {
+      "epoch": 0.6937413713759779,
+      "grad_norm": 0.8075084990591824,
+      "learning_rate": 7.315241572641051e-06,
+      "loss": 0.4925,
+      "num_input_tokens_seen": 173714992,
+      "step": 3015,
+      "train_runtime": 16865.2743,
+      "train_tokens_per_second": 10300.158
+    },
+    {
+      "epoch": 0.6948918545789231,
+      "grad_norm": 0.8135101494750264,
+      "learning_rate": 7.307229002723094e-06,
+      "loss": 0.4994,
+      "num_input_tokens_seen": 174013240,
+      "step": 3020,
+      "train_runtime": 16897.445,
+      "train_tokens_per_second": 10298.198
+    },
+    {
+      "epoch": 0.6960423377818684,
+      "grad_norm": 0.7753295916060926,
+      "learning_rate": 7.299208897672474e-06,
+      "loss": 0.483,
+      "num_input_tokens_seen": 174301272,
+      "step": 3025,
+      "train_runtime": 16935.3127,
+      "train_tokens_per_second": 10292.179
+    },
+    {
+      "epoch": 0.6971928209848136,
+      "grad_norm": 0.7973421124963277,
+      "learning_rate": 7.291181283681891e-06,
+      "loss": 0.449,
+      "num_input_tokens_seen": 174599016,
+      "step": 3030,
+      "train_runtime": 16967.3424,
+      "train_tokens_per_second": 10290.298
+    },
+    {
+      "epoch": 0.6983433041877588,
+      "grad_norm": 0.8684379624229391,
+      "learning_rate": 7.283146186968566e-06,
+      "loss": 0.4888,
+      "num_input_tokens_seen": 174869104,
+      "step": 3035,
+      "train_runtime": 16995.7722,
+      "train_tokens_per_second": 10288.977
+    },
+    {
+      "epoch": 0.6994937873907041,
+      "grad_norm": 0.6864840050720713,
+      "learning_rate": 7.275103633774161e-06,
+      "loss": 0.4908,
+      "num_input_tokens_seen": 175171480,
+      "step": 3040,
+      "train_runtime": 17022.6585,
+      "train_tokens_per_second": 10290.489
+    },
+    {
+      "epoch": 0.7006442705936493,
+      "grad_norm": 0.7085976512571162,
+      "learning_rate": 7.267053650364686e-06,
+      "loss": 0.481,
+      "num_input_tokens_seen": 175477944,
+      "step": 3045,
+      "train_runtime": 17047.0868,
+      "train_tokens_per_second": 10293.72
+    },
+    {
+      "epoch": 0.7017947537965946,
+      "grad_norm": 0.7895214598435649,
+      "learning_rate": 7.258996263030421e-06,
+      "loss": 0.5042,
+      "num_input_tokens_seen": 175773552,
+      "step": 3050,
+      "train_runtime": 17077.9511,
+      "train_tokens_per_second": 10292.426
+    },
+    {
+      "epoch": 0.7029452369995398,
+      "grad_norm": 0.7317686123494026,
+      "learning_rate": 7.250931498085825e-06,
+      "loss": 0.4843,
+      "num_input_tokens_seen": 176087440,
+      "step": 3055,
+      "train_runtime": 17113.1325,
+      "train_tokens_per_second": 10289.609
+    },
+    {
+      "epoch": 0.704095720202485,
+      "grad_norm": 0.8609960363249958,
+      "learning_rate": 7.242859381869452e-06,
+      "loss": 0.4643,
+      "num_input_tokens_seen": 176363000,
+      "step": 3060,
+      "train_runtime": 17142.3389,
+      "train_tokens_per_second": 10288.153
+    },
+    {
+      "epoch": 0.7052462034054303,
+      "grad_norm": 0.9464107250444789,
+      "learning_rate": 7.23477994074386e-06,
+      "loss": 0.4845,
+      "num_input_tokens_seen": 176645944,
+      "step": 3065,
+      "train_runtime": 17176.1113,
+      "train_tokens_per_second": 10284.397
+    },
+    {
+      "epoch": 0.7063966866083755,
+      "grad_norm": 0.7664679510250674,
+      "learning_rate": 7.226693201095536e-06,
+      "loss": 0.465,
+      "num_input_tokens_seen": 176940216,
+      "step": 3070,
+      "train_runtime": 17203.8161,
+      "train_tokens_per_second": 10284.94
+    },
+    {
+      "epoch": 0.7075471698113207,
+      "grad_norm": 0.7776938880662367,
+      "learning_rate": 7.218599189334799e-06,
+      "loss": 0.5085,
+      "num_input_tokens_seen": 177238984,
+      "step": 3075,
+      "train_runtime": 17237.4117,
+      "train_tokens_per_second": 10282.227
+    },
+    {
+      "epoch": 0.708697653014266,
+      "grad_norm": 0.7338595862518222,
+      "learning_rate": 7.210497931895719e-06,
+      "loss": 0.4775,
+      "num_input_tokens_seen": 177517304,
+      "step": 3080,
+      "train_runtime": 17265.7353,
+      "train_tokens_per_second": 10281.48
+    },
+    {
+      "epoch": 0.7098481362172112,
+      "grad_norm": 0.7719790326824635,
+      "learning_rate": 7.202389455236029e-06,
+      "loss": 0.4749,
+      "num_input_tokens_seen": 177804888,
+      "step": 3085,
+      "train_runtime": 17293.6245,
+      "train_tokens_per_second": 10281.528
+    },
+    {
+      "epoch": 0.7109986194201565,
+      "grad_norm": 1.1454292693103056,
+      "learning_rate": 7.19427378583704e-06,
+      "loss": 0.433,
+      "num_input_tokens_seen": 178099480,
+      "step": 3090,
+      "train_runtime": 17319.3989,
+      "train_tokens_per_second": 10283.237
+    },
+    {
+      "epoch": 0.7121491026231017,
+      "grad_norm": 0.7488455310054369,
+      "learning_rate": 7.18615095020355e-06,
+      "loss": 0.462,
+      "num_input_tokens_seen": 178397224,
+      "step": 3095,
+      "train_runtime": 17344.6061,
+      "train_tokens_per_second": 10285.458
+    },
+    {
+      "epoch": 0.7132995858260469,
+      "grad_norm": 0.7251997627567377,
+      "learning_rate": 7.178020974863768e-06,
+      "loss": 0.4803,
+      "num_input_tokens_seen": 178690200,
+      "step": 3100,
+      "train_runtime": 17376.6431,
+      "train_tokens_per_second": 10283.356
+    },
+    {
+      "epoch": 0.7144500690289922,
+      "grad_norm": 0.7558022281306584,
+      "learning_rate": 7.169883886369216e-06,
+      "loss": 0.4475,
+      "num_input_tokens_seen": 178957712,
+      "step": 3105,
+      "train_runtime": 17400.8645,
+      "train_tokens_per_second": 10284.415
+    },
+    {
+      "epoch": 0.7156005522319374,
+      "grad_norm": 0.7528333312602612,
+      "learning_rate": 7.161739711294644e-06,
+      "loss": 0.4789,
+      "num_input_tokens_seen": 179250712,
+      "step": 3110,
+      "train_runtime": 17436.2348,
+      "train_tokens_per_second": 10280.357
+    },
+    {
+      "epoch": 0.7167510354348826,
+      "grad_norm": 0.7632583560857564,
+      "learning_rate": 7.1535884762379524e-06,
+      "loss": 0.4806,
+      "num_input_tokens_seen": 179549328,
+      "step": 3115,
+      "train_runtime": 17463.317,
+      "train_tokens_per_second": 10281.513
+    },
+    {
+      "epoch": 0.7179015186378279,
+      "grad_norm": 0.7219479397305609,
+      "learning_rate": 7.145430207820094e-06,
+      "loss": 0.4998,
+      "num_input_tokens_seen": 179834672,
+      "step": 3120,
+      "train_runtime": 17497.9541,
+      "train_tokens_per_second": 10277.469
+    },
+    {
+      "epoch": 0.7190520018407731,
+      "grad_norm": 0.7119452724768526,
+      "learning_rate": 7.137264932684993e-06,
+      "loss": 0.4596,
+      "num_input_tokens_seen": 180138000,
+      "step": 3125,
+      "train_runtime": 17528.4955,
+      "train_tokens_per_second": 10276.866
+    },
+    {
+      "epoch": 0.7202024850437183,
+      "grad_norm": 0.8500675623026189,
+      "learning_rate": 7.1290926774994586e-06,
+      "loss": 0.4769,
+      "num_input_tokens_seen": 180437856,
+      "step": 3130,
+      "train_runtime": 17558.5267,
+      "train_tokens_per_second": 10276.367
+    },
+    {
+      "epoch": 0.7213529682466636,
+      "grad_norm": 0.7751269141987924,
+      "learning_rate": 7.12091346895309e-06,
+      "loss": 0.4804,
+      "num_input_tokens_seen": 180743416,
+      "step": 3135,
+      "train_runtime": 17581.6486,
+      "train_tokens_per_second": 10280.231
+    },
+    {
+      "epoch": 0.7225034514496088,
+      "grad_norm": 0.7745039328994962,
+      "learning_rate": 7.112727333758203e-06,
+      "loss": 0.4333,
+      "num_input_tokens_seen": 181027056,
+      "step": 3140,
+      "train_runtime": 17612.4485,
+      "train_tokens_per_second": 10278.358
+    },
+    {
+      "epoch": 0.7236539346525541,
+      "grad_norm": 0.7991455480012907,
+      "learning_rate": 7.104534298649733e-06,
+      "loss": 0.4835,
+      "num_input_tokens_seen": 181324016,
+      "step": 3145,
+      "train_runtime": 17639.5502,
+      "train_tokens_per_second": 10279.401
+    },
+    {
+      "epoch": 0.7248044178554993,
+      "grad_norm": 0.750867936078803,
+      "learning_rate": 7.096334390385141e-06,
+      "loss": 0.4471,
+      "num_input_tokens_seen": 181619400,
+      "step": 3150,
+      "train_runtime": 17664.3706,
+      "train_tokens_per_second": 10281.68
+    },
+    {
+      "epoch": 0.7259549010584445,
+      "grad_norm": 0.7945326243407761,
+      "learning_rate": 7.088127635744349e-06,
+      "loss": 0.4843,
+      "num_input_tokens_seen": 181913480,
+      "step": 3155,
+      "train_runtime": 17690.085,
+      "train_tokens_per_second": 10283.358
+    },
+    {
+      "epoch": 0.7271053842613898,
+      "grad_norm": 0.952506094342453,
+      "learning_rate": 7.079914061529629e-06,
+      "loss": 0.4688,
+      "num_input_tokens_seen": 182198280,
+      "step": 3160,
+      "train_runtime": 17717.9126,
+      "train_tokens_per_second": 10283.281
+    },
+    {
+      "epoch": 0.728255867464335,
+      "grad_norm": 0.9218085516261377,
+      "learning_rate": 7.071693694565526e-06,
+      "loss": 0.4731,
+      "num_input_tokens_seen": 182485912,
+      "step": 3165,
+      "train_runtime": 17745.913,
+      "train_tokens_per_second": 10283.264
+    },
+    {
+      "epoch": 0.7294063506672802,
+      "grad_norm": 0.8799284278057512,
+      "learning_rate": 7.0634665616987754e-06,
+      "loss": 0.503,
+      "num_input_tokens_seen": 182770832,
+      "step": 3170,
+      "train_runtime": 17772.0874,
+      "train_tokens_per_second": 10284.151
+    },
+    {
+      "epoch": 0.7305568338702255,
+      "grad_norm": 0.8371983616336449,
+      "learning_rate": 7.055232689798197e-06,
+      "loss": 0.4575,
+      "num_input_tokens_seen": 183066064,
+      "step": 3175,
+      "train_runtime": 17804.0594,
+      "train_tokens_per_second": 10282.265
+    },
+    {
+      "epoch": 0.7317073170731707,
+      "grad_norm": 0.8964654844549433,
+      "learning_rate": 7.0469921057546345e-06,
+      "loss": 0.4703,
+      "num_input_tokens_seen": 183360032,
+      "step": 3180,
+      "train_runtime": 17834.4927,
+      "train_tokens_per_second": 10281.203
+    },
+    {
+      "epoch": 0.732857800276116,
+      "grad_norm": 0.8635732925134377,
+      "learning_rate": 7.03874483648084e-06,
+      "loss": 0.4883,
+      "num_input_tokens_seen": 183660360,
+      "step": 3185,
+      "train_runtime": 17859.8703,
+      "train_tokens_per_second": 10283.41
+    },
+    {
+      "epoch": 0.7340082834790612,
+      "grad_norm": 0.7821569057825497,
+      "learning_rate": 7.030490908911405e-06,
+      "loss": 0.4995,
+      "num_input_tokens_seen": 183924464,
+      "step": 3190,
+      "train_runtime": 17883.8443,
+      "train_tokens_per_second": 10284.392
+    },
+    {
+      "epoch": 0.7351587666820064,
+      "grad_norm": 0.7437976281494045,
+      "learning_rate": 7.022230350002665e-06,
+      "loss": 0.4847,
+      "num_input_tokens_seen": 184221360,
+      "step": 3195,
+      "train_runtime": 17913.9921,
+      "train_tokens_per_second": 10283.658
+    },
+    {
+      "epoch": 0.7363092498849517,
+      "grad_norm": 0.74044577823954,
+      "learning_rate": 7.013963186732615e-06,
+      "loss": 0.4678,
+      "num_input_tokens_seen": 184528920,
+      "step": 3200,
+      "train_runtime": 17946.4578,
+      "train_tokens_per_second": 10282.192
+    },
+    {
+      "epoch": 0.7374597330878969,
+      "grad_norm": 0.817948099656526,
+      "learning_rate": 7.005689446100816e-06,
+      "loss": 0.4878,
+      "num_input_tokens_seen": 184829904,
+      "step": 3205,
+      "train_runtime": 17971.8264,
+      "train_tokens_per_second": 10284.425
+    },
+    {
+      "epoch": 0.7386102162908421,
+      "grad_norm": 0.9165634793793612,
+      "learning_rate": 6.997409155128309e-06,
+      "loss": 0.4731,
+      "num_input_tokens_seen": 185129024,
+      "step": 3210,
+      "train_runtime": 18001.3182,
+      "train_tokens_per_second": 10284.193
+    },
+    {
+      "epoch": 0.7397606994937874,
+      "grad_norm": 0.7481794358696198,
+      "learning_rate": 6.989122340857528e-06,
+      "loss": 0.4642,
+      "num_input_tokens_seen": 185427696,
+      "step": 3215,
+      "train_runtime": 18027.0218,
+      "train_tokens_per_second": 10286.097
+    },
+    {
+      "epoch": 0.7409111826967326,
+      "grad_norm": 0.8096419289723841,
+      "learning_rate": 6.980829030352216e-06,
+      "loss": 0.4854,
+      "num_input_tokens_seen": 185700760,
+      "step": 3220,
+      "train_runtime": 18059.4848,
+      "train_tokens_per_second": 10282.727
+    },
+    {
+      "epoch": 0.7420616658996778,
+      "grad_norm": 1.2192955771691543,
+      "learning_rate": 6.972529250697329e-06,
+      "loss": 0.4833,
+      "num_input_tokens_seen": 185980832,
+      "step": 3225,
+      "train_runtime": 18088.7954,
+      "train_tokens_per_second": 10281.549
+    },
+    {
+      "epoch": 0.7432121491026231,
+      "grad_norm": 0.7590656448097156,
+      "learning_rate": 6.964223028998947e-06,
+      "loss": 0.4799,
+      "num_input_tokens_seen": 186278576,
+      "step": 3230,
+      "train_runtime": 18114.0384,
+      "train_tokens_per_second": 10283.658
+    },
+    {
+      "epoch": 0.7443626323055683,
+      "grad_norm": 0.6782259572828151,
+      "learning_rate": 6.955910392384191e-06,
+      "loss": 0.4355,
+      "num_input_tokens_seen": 186544264,
+      "step": 3235,
+      "train_runtime": 18148.6879,
+      "train_tokens_per_second": 10278.664
+    },
+    {
+      "epoch": 0.7455131155085136,
+      "grad_norm": 0.7773537510189871,
+      "learning_rate": 6.947591368001138e-06,
+      "loss": 0.4766,
+      "num_input_tokens_seen": 186846880,
+      "step": 3240,
+      "train_runtime": 18170.3049,
+      "train_tokens_per_second": 10283.09
+    },
+    {
+      "epoch": 0.7466635987114588,
+      "grad_norm": 0.6961437520106416,
+      "learning_rate": 6.939265983018717e-06,
+      "loss": 0.4589,
+      "num_input_tokens_seen": 187137376,
+      "step": 3245,
+      "train_runtime": 18201.1998,
+      "train_tokens_per_second": 10281.596
+    },
+    {
+      "epoch": 0.747814081914404,
+      "grad_norm": 0.889601087181237,
+      "learning_rate": 6.930934264626639e-06,
+      "loss": 0.4822,
+      "num_input_tokens_seen": 187422128,
+      "step": 3250,
+      "train_runtime": 18236.771,
+      "train_tokens_per_second": 10277.155
+    },
+    {
+      "epoch": 0.7489645651173493,
+      "grad_norm": 0.8446338559276212,
+      "learning_rate": 6.922596240035291e-06,
+      "loss": 0.4686,
+      "num_input_tokens_seen": 187700672,
+      "step": 3255,
+      "train_runtime": 18267.1826,
+      "train_tokens_per_second": 10275.294
+    },
+    {
+      "epoch": 0.7501150483202945,
+      "grad_norm": 0.7835231632122263,
+      "learning_rate": 6.9142519364756625e-06,
+      "loss": 0.4574,
+      "num_input_tokens_seen": 187983248,
+      "step": 3260,
+      "train_runtime": 18300.2699,
+      "train_tokens_per_second": 10272.157
+    },
+    {
+      "epoch": 0.7512655315232397,
+      "grad_norm": 0.7239398558295543,
+      "learning_rate": 6.905901381199245e-06,
+      "loss": 0.4792,
+      "num_input_tokens_seen": 188285712,
+      "step": 3265,
+      "train_runtime": 18333.01,
+      "train_tokens_per_second": 10270.311
+    },
+    {
+      "epoch": 0.752416014726185,
+      "grad_norm": 0.8038938097258223,
+      "learning_rate": 6.897544601477949e-06,
+      "loss": 0.4861,
+      "num_input_tokens_seen": 188564736,
+      "step": 3270,
+      "train_runtime": 18369.0999,
+      "train_tokens_per_second": 10265.323
+    },
+    {
+      "epoch": 0.7535664979291302,
+      "grad_norm": 0.8048526935163539,
+      "learning_rate": 6.8891816246040136e-06,
+      "loss": 0.4706,
+      "num_input_tokens_seen": 188875728,
+      "step": 3275,
+      "train_runtime": 18403.2657,
+      "train_tokens_per_second": 10263.164
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.8323373142294449,
+      "learning_rate": 6.880812477889914e-06,
+      "loss": 0.4569,
+      "num_input_tokens_seen": 189163472,
+      "step": 3280,
+      "train_runtime": 18430.1521,
+      "train_tokens_per_second": 10263.804
+    },
+    {
+      "epoch": 0.7558674643350207,
+      "grad_norm": 0.8372395081407994,
+      "learning_rate": 6.872437188668279e-06,
+      "loss": 0.4546,
+      "num_input_tokens_seen": 189438352,
+      "step": 3285,
+      "train_runtime": 18457.8456,
+      "train_tokens_per_second": 10263.297
+    },
+    {
+      "epoch": 0.7570179475379659,
+      "grad_norm": 0.8418520048577535,
+      "learning_rate": 6.864055784291797e-06,
+      "loss": 0.4409,
+      "num_input_tokens_seen": 189714432,
+      "step": 3290,
+      "train_runtime": 18489.4519,
+      "train_tokens_per_second": 10260.684
+    },
+    {
+      "epoch": 0.7581684307409112,
+      "grad_norm": 0.7759317356615708,
+      "learning_rate": 6.855668292133126e-06,
+      "loss": 0.4754,
+      "num_input_tokens_seen": 190010264,
+      "step": 3295,
+      "train_runtime": 18517.582,
+      "train_tokens_per_second": 10261.073
+    },
+    {
+      "epoch": 0.7593189139438564,
+      "grad_norm": 0.867938883380217,
+      "learning_rate": 6.84727473958481e-06,
+      "loss": 0.4707,
+      "num_input_tokens_seen": 190310944,
+      "step": 3300,
+      "train_runtime": 18547.26,
+      "train_tokens_per_second": 10260.866
+    },
+    {
+      "epoch": 0.7604693971468016,
+      "grad_norm": 0.8517638102865541,
+      "learning_rate": 6.838875154059181e-06,
+      "loss": 0.4757,
+      "num_input_tokens_seen": 190605952,
+      "step": 3305,
+      "train_runtime": 18572.3653,
+      "train_tokens_per_second": 10262.88
+    },
+    {
+      "epoch": 0.7616198803497469,
+      "grad_norm": 0.7543432890917187,
+      "learning_rate": 6.830469562988275e-06,
+      "loss": 0.4383,
+      "num_input_tokens_seen": 190875640,
+      "step": 3310,
+      "train_runtime": 18601.3246,
+      "train_tokens_per_second": 10261.4
+    },
+    {
+      "epoch": 0.7627703635526921,
+      "grad_norm": 0.7843190013357402,
+      "learning_rate": 6.822057993823745e-06,
+      "loss": 0.4356,
+      "num_input_tokens_seen": 191166520,
+      "step": 3315,
+      "train_runtime": 18629.0324,
+      "train_tokens_per_second": 10261.752
+    },
+    {
+      "epoch": 0.7639208467556373,
+      "grad_norm": 0.865220992112316,
+      "learning_rate": 6.813640474036766e-06,
+      "loss": 0.4728,
+      "num_input_tokens_seen": 191475560,
+      "step": 3320,
+      "train_runtime": 18653.9287,
+      "train_tokens_per_second": 10264.624
+    },
+    {
+      "epoch": 0.7650713299585826,
+      "grad_norm": 0.7831887348579345,
+      "learning_rate": 6.805217031117941e-06,
+      "loss": 0.4673,
+      "num_input_tokens_seen": 191759024,
+      "step": 3325,
+      "train_runtime": 18691.9834,
+      "train_tokens_per_second": 10258.891
+    },
+    {
+      "epoch": 0.7662218131615278,
+      "grad_norm": 0.7466850732703669,
+      "learning_rate": 6.796787692577227e-06,
+      "loss": 0.4679,
+      "num_input_tokens_seen": 192078864,
+      "step": 3330,
+      "train_runtime": 18716.5426,
+      "train_tokens_per_second": 10262.518
+    },
+    {
+      "epoch": 0.7673722963644731,
+      "grad_norm": 0.8558930491441068,
+      "learning_rate": 6.788352485943828e-06,
+      "loss": 0.4597,
+      "num_input_tokens_seen": 192387824,
+      "step": 3335,
+      "train_runtime": 18743.7171,
+      "train_tokens_per_second": 10264.123
+    },
+    {
+      "epoch": 0.7685227795674183,
+      "grad_norm": 0.7917003722529835,
+      "learning_rate": 6.779911438766117e-06,
+      "loss": 0.4662,
+      "num_input_tokens_seen": 192676328,
+      "step": 3340,
+      "train_runtime": 18777.0272,
+      "train_tokens_per_second": 10261.28
+    },
+    {
+      "epoch": 0.7696732627703635,
+      "grad_norm": 1.0074499972157536,
+      "learning_rate": 6.7714645786115395e-06,
+      "loss": 0.4567,
+      "num_input_tokens_seen": 192941864,
+      "step": 3345,
+      "train_runtime": 18805.2801,
+      "train_tokens_per_second": 10259.984
+    },
+    {
+      "epoch": 0.7708237459733088,
+      "grad_norm": 0.7928724665344253,
+      "learning_rate": 6.763011933066524e-06,
+      "loss": 0.4483,
+      "num_input_tokens_seen": 193234784,
+      "step": 3350,
+      "train_runtime": 18830.8323,
+      "train_tokens_per_second": 10261.617
+    },
+    {
+      "epoch": 0.771974229176254,
+      "grad_norm": 0.7553156431888949,
+      "learning_rate": 6.754553529736398e-06,
+      "loss": 0.4555,
+      "num_input_tokens_seen": 193486056,
+      "step": 3355,
+      "train_runtime": 18862.5368,
+      "train_tokens_per_second": 10257.69
+    },
+    {
+      "epoch": 0.7731247123791992,
+      "grad_norm": 1.0599574166454613,
+      "learning_rate": 6.746089396245285e-06,
+      "loss": 0.4665,
+      "num_input_tokens_seen": 193770624,
+      "step": 3360,
+      "train_runtime": 18887.5404,
+      "train_tokens_per_second": 10259.177
+    },
+    {
+      "epoch": 0.7742751955821445,
+      "grad_norm": 0.8273833954715941,
+      "learning_rate": 6.737619560236035e-06,
+      "loss": 0.4612,
+      "num_input_tokens_seen": 194035760,
+      "step": 3365,
+      "train_runtime": 18914.566,
+      "train_tokens_per_second": 10258.536
+    },
+    {
+      "epoch": 0.7754256787850897,
+      "grad_norm": 0.8457205737609425,
+      "learning_rate": 6.729144049370111e-06,
+      "loss": 0.4808,
+      "num_input_tokens_seen": 194340688,
+      "step": 3370,
+      "train_runtime": 18947.2502,
+      "train_tokens_per_second": 10256.934
+    },
+    {
+      "epoch": 0.776576161988035,
+      "grad_norm": 0.793624613628056,
+      "learning_rate": 6.720662891327512e-06,
+      "loss": 0.4466,
+      "num_input_tokens_seen": 194608784,
+      "step": 3375,
+      "train_runtime": 18974.9218,
+      "train_tokens_per_second": 10256.105
+    },
+    {
+      "epoch": 0.7777266451909802,
+      "grad_norm": 0.7786838552620566,
+      "learning_rate": 6.712176113806681e-06,
+      "loss": 0.4852,
+      "num_input_tokens_seen": 194907448,
+      "step": 3380,
+      "train_runtime": 19001.0902,
+      "train_tokens_per_second": 10257.698
+    },
+    {
+      "epoch": 0.7788771283939254,
+      "grad_norm": 0.7023624991995246,
+      "learning_rate": 6.703683744524415e-06,
+      "loss": 0.4563,
+      "num_input_tokens_seen": 195175392,
+      "step": 3385,
+      "train_runtime": 19027.8979,
+      "train_tokens_per_second": 10257.328
+    },
+    {
+      "epoch": 0.7800276115968707,
+      "grad_norm": 0.7135932603579592,
+      "learning_rate": 6.695185811215772e-06,
+      "loss": 0.474,
+      "num_input_tokens_seen": 195457272,
+      "step": 3390,
+      "train_runtime": 19054.2994,
+      "train_tokens_per_second": 10257.909
+    },
+    {
+      "epoch": 0.7811780947998159,
+      "grad_norm": 0.8773328029570522,
+      "learning_rate": 6.686682341633979e-06,
+      "loss": 0.4866,
+      "num_input_tokens_seen": 195738632,
+      "step": 3395,
+      "train_runtime": 19080.3903,
+      "train_tokens_per_second": 10258.628
+    },
+    {
+      "epoch": 0.7823285780027611,
+      "grad_norm": 0.7693277106748914,
+      "learning_rate": 6.6781733635503485e-06,
+      "loss": 0.4528,
+      "num_input_tokens_seen": 196016320,
+      "step": 3400,
+      "train_runtime": 19103.3379,
+      "train_tokens_per_second": 10260.841
+    },
+    {
+      "epoch": 0.7834790612057064,
+      "grad_norm": 0.6998622424392189,
+      "learning_rate": 6.669658904754177e-06,
+      "loss": 0.4499,
+      "num_input_tokens_seen": 196315712,
+      "step": 3405,
+      "train_runtime": 19129.8601,
+      "train_tokens_per_second": 10262.266
+    },
+    {
+      "epoch": 0.7846295444086516,
+      "grad_norm": 0.7863959004416019,
+      "learning_rate": 6.661138993052667e-06,
+      "loss": 0.491,
+      "num_input_tokens_seen": 196599088,
+      "step": 3410,
+      "train_runtime": 19163.8446,
+      "train_tokens_per_second": 10258.854
+    },
+    {
+      "epoch": 0.7857800276115968,
+      "grad_norm": 0.7290474954275662,
+      "learning_rate": 6.6526136562708244e-06,
+      "loss": 0.479,
+      "num_input_tokens_seen": 196892736,
+      "step": 3415,
+      "train_runtime": 19186.0704,
+      "train_tokens_per_second": 10262.275
+    },
+    {
+      "epoch": 0.7869305108145421,
+      "grad_norm": 0.778715420245958,
+      "learning_rate": 6.644082922251373e-06,
+      "loss": 0.5084,
+      "num_input_tokens_seen": 197180408,
+      "step": 3420,
+      "train_runtime": 19212.5656,
+      "train_tokens_per_second": 10263.096
+    },
+    {
+      "epoch": 0.7880809940174873,
+      "grad_norm": 0.7537611256146215,
+      "learning_rate": 6.63554681885467e-06,
+      "loss": 0.4765,
+      "num_input_tokens_seen": 197483720,
+      "step": 3425,
+      "train_runtime": 19239.4687,
+      "train_tokens_per_second": 10264.51
+    },
+    {
+      "epoch": 0.7892314772204326,
+      "grad_norm": 0.902014061196049,
+      "learning_rate": 6.627005373958599e-06,
+      "loss": 0.4854,
+      "num_input_tokens_seen": 197781792,
+      "step": 3430,
+      "train_runtime": 19271.5405,
+      "train_tokens_per_second": 10262.895
+    },
+    {
+      "epoch": 0.7903819604233778,
+      "grad_norm": 0.7705454037094507,
+      "learning_rate": 6.618458615458494e-06,
+      "loss": 0.4865,
+      "num_input_tokens_seen": 198090664,
+      "step": 3435,
+      "train_runtime": 19294.8146,
+      "train_tokens_per_second": 10266.523
+    },
+    {
+      "epoch": 0.791532443626323,
+      "grad_norm": 0.799534002243912,
+      "learning_rate": 6.609906571267042e-06,
+      "loss": 0.4558,
+      "num_input_tokens_seen": 198367552,
+      "step": 3440,
+      "train_runtime": 19317.7679,
+      "train_tokens_per_second": 10268.658
+    },
+    {
+      "epoch": 0.7926829268292683,
+      "grad_norm": 0.7513657200485946,
+      "learning_rate": 6.601349269314188e-06,
+      "loss": 0.4784,
+      "num_input_tokens_seen": 198657840,
+      "step": 3445,
+      "train_runtime": 19339.63,
+      "train_tokens_per_second": 10272.06
+    },
+    {
+      "epoch": 0.7938334100322135,
+      "grad_norm": 0.8829792556637166,
+      "learning_rate": 6.592786737547057e-06,
+      "loss": 0.4596,
+      "num_input_tokens_seen": 198939664,
+      "step": 3450,
+      "train_runtime": 19359.6713,
+      "train_tokens_per_second": 10275.984
+    },
+    {
+      "epoch": 0.7949838932351587,
+      "grad_norm": 0.763916446153832,
+      "learning_rate": 6.584219003929846e-06,
+      "loss": 0.4559,
+      "num_input_tokens_seen": 199254392,
+      "step": 3455,
+      "train_runtime": 19391.8433,
+      "train_tokens_per_second": 10275.165
+    },
+    {
+      "epoch": 0.796134376438104,
+      "grad_norm": 0.7916239724155869,
+      "learning_rate": 6.575646096443744e-06,
+      "loss": 0.4621,
+      "num_input_tokens_seen": 199552224,
+      "step": 3460,
+      "train_runtime": 19423.5458,
+      "train_tokens_per_second": 10273.728
+    },
+    {
+      "epoch": 0.7972848596410492,
+      "grad_norm": 0.6832982760763032,
+      "learning_rate": 6.567068043086836e-06,
+      "loss": 0.4631,
+      "num_input_tokens_seen": 199845536,
+      "step": 3465,
+      "train_runtime": 19448.6193,
+      "train_tokens_per_second": 10275.564
+    },
+    {
+      "epoch": 0.7984353428439944,
+      "grad_norm": 0.7636143962089569,
+      "learning_rate": 6.558484871874013e-06,
+      "loss": 0.4714,
+      "num_input_tokens_seen": 200131376,
+      "step": 3470,
+      "train_runtime": 19475.7226,
+      "train_tokens_per_second": 10275.941
+    },
+    {
+      "epoch": 0.7995858260469397,
+      "grad_norm": 0.957687335822549,
+      "learning_rate": 6.549896610836884e-06,
+      "loss": 0.4709,
+      "num_input_tokens_seen": 200401448,
+      "step": 3475,
+      "train_runtime": 19498.6733,
+      "train_tokens_per_second": 10277.697
+    },
+    {
+      "epoch": 0.8007363092498849,
+      "grad_norm": 0.6793975928356231,
+      "learning_rate": 6.5413032880236725e-06,
+      "loss": 0.4675,
+      "num_input_tokens_seen": 200696440,
+      "step": 3480,
+      "train_runtime": 19530.5524,
+      "train_tokens_per_second": 10276.025
+    },
+    {
+      "epoch": 0.8018867924528302,
+      "grad_norm": 0.8460943950611907,
+      "learning_rate": 6.532704931499142e-06,
+      "loss": 0.464,
+      "num_input_tokens_seen": 200974408,
+      "step": 3485,
+      "train_runtime": 19559.8221,
+      "train_tokens_per_second": 10274.859
+    },
+    {
+      "epoch": 0.8030372756557754,
+      "grad_norm": 0.7777585297399622,
+      "learning_rate": 6.52410156934449e-06,
+      "loss": 0.4689,
+      "num_input_tokens_seen": 201236504,
+      "step": 3490,
+      "train_runtime": 19589.5383,
+      "train_tokens_per_second": 10272.652
+    },
+    {
+      "epoch": 0.8041877588587206,
+      "grad_norm": 0.8033541136400638,
+      "learning_rate": 6.515493229657265e-06,
+      "loss": 0.4557,
+      "num_input_tokens_seen": 201516912,
+      "step": 3495,
+      "train_runtime": 19611.7687,
+      "train_tokens_per_second": 10275.305
+    },
+    {
+      "epoch": 0.8053382420616659,
+      "grad_norm": 0.7959539756068433,
+      "learning_rate": 6.506879940551268e-06,
+      "loss": 0.4373,
+      "num_input_tokens_seen": 201829200,
+      "step": 3500,
+      "train_runtime": 19636.202,
+      "train_tokens_per_second": 10278.424
+    },
+    {
+      "epoch": 0.8064887252646111,
+      "grad_norm": 0.7251031884994366,
+      "learning_rate": 6.4982617301564665e-06,
+      "loss": 0.4393,
+      "num_input_tokens_seen": 202092272,
+      "step": 3505,
+      "train_runtime": 19664.9557,
+      "train_tokens_per_second": 10276.772
+    },
+    {
+      "epoch": 0.8076392084675563,
+      "grad_norm": 0.8023450849762833,
+      "learning_rate": 6.489638626618901e-06,
+      "loss": 0.4711,
+      "num_input_tokens_seen": 202368024,
+      "step": 3510,
+      "train_runtime": 19700.0432,
+      "train_tokens_per_second": 10272.466
+    },
+    {
+      "epoch": 0.8087896916705016,
+      "grad_norm": 0.813896984307446,
+      "learning_rate": 6.481010658100591e-06,
+      "loss": 0.4965,
+      "num_input_tokens_seen": 202635536,
+      "step": 3515,
+      "train_runtime": 19726.5329,
+      "train_tokens_per_second": 10272.233
+    },
+    {
+      "epoch": 0.8099401748734468,
+      "grad_norm": 0.7763604505632685,
+      "learning_rate": 6.4723778527794446e-06,
+      "loss": 0.4716,
+      "num_input_tokens_seen": 202916416,
+      "step": 3520,
+      "train_runtime": 19755.8335,
+      "train_tokens_per_second": 10271.215
+    },
+    {
+      "epoch": 0.8110906580763921,
+      "grad_norm": 0.913578819154624,
+      "learning_rate": 6.463740238849165e-06,
+      "loss": 0.4796,
+      "num_input_tokens_seen": 203186704,
+      "step": 3525,
+      "train_runtime": 19787.425,
+      "train_tokens_per_second": 10268.476
+    },
+    {
+      "epoch": 0.8122411412793373,
+      "grad_norm": 0.7709056313689515,
+      "learning_rate": 6.455097844519165e-06,
+      "loss": 0.4659,
+      "num_input_tokens_seen": 203478568,
+      "step": 3530,
+      "train_runtime": 19814.447,
+      "train_tokens_per_second": 10269.202
+    },
+    {
+      "epoch": 0.8133916244822825,
+      "grad_norm": 0.8238638008800891,
+      "learning_rate": 6.4464506980144645e-06,
+      "loss": 0.5016,
+      "num_input_tokens_seen": 203779512,
+      "step": 3535,
+      "train_runtime": 19836.6859,
+      "train_tokens_per_second": 10272.861
+    },
+    {
+      "epoch": 0.8145421076852278,
+      "grad_norm": 0.8603140089120698,
+      "learning_rate": 6.437798827575602e-06,
+      "loss": 0.4859,
+      "num_input_tokens_seen": 204077088,
+      "step": 3540,
+      "train_runtime": 19872.2869,
+      "train_tokens_per_second": 10269.431
+    },
+    {
+      "epoch": 0.815692590888173,
+      "grad_norm": 0.7582998518373074,
+      "learning_rate": 6.4291422614585505e-06,
+      "loss": 0.4494,
+      "num_input_tokens_seen": 204367688,
+      "step": 3545,
+      "train_runtime": 19895.6029,
+      "train_tokens_per_second": 10272.003
+    },
+    {
+      "epoch": 0.8168430740911182,
+      "grad_norm": 0.9450781630166978,
+      "learning_rate": 6.420481027934613e-06,
+      "loss": 0.4574,
+      "num_input_tokens_seen": 204659920,
+      "step": 3550,
+      "train_runtime": 19929.0474,
+      "train_tokens_per_second": 10269.428
+    },
+    {
+      "epoch": 0.8179935572940635,
+      "grad_norm": 0.797223900950099,
+      "learning_rate": 6.411815155290338e-06,
+      "loss": 0.479,
+      "num_input_tokens_seen": 204951672,
+      "step": 3555,
+      "train_runtime": 19959.9533,
+      "train_tokens_per_second": 10268.144
+    },
+    {
+      "epoch": 0.8191440404970087,
+      "grad_norm": 0.7347008194458451,
+      "learning_rate": 6.403144671827421e-06,
+      "loss": 0.5055,
+      "num_input_tokens_seen": 205241968,
+      "step": 3560,
+      "train_runtime": 19985.1656,
+      "train_tokens_per_second": 10269.716
+    },
+    {
+      "epoch": 0.8202945236999539,
+      "grad_norm": 0.771754589495352,
+      "learning_rate": 6.394469605862625e-06,
+      "loss": 0.4804,
+      "num_input_tokens_seen": 205523336,
+      "step": 3565,
+      "train_runtime": 20015.665,
+      "train_tokens_per_second": 10268.124
+    },
+    {
+      "epoch": 0.8214450069028992,
+      "grad_norm": 0.7458681461243873,
+      "learning_rate": 6.3857899857276685e-06,
+      "loss": 0.4638,
+      "num_input_tokens_seen": 205804168,
+      "step": 3570,
+      "train_runtime": 20044.1246,
+      "train_tokens_per_second": 10267.556
+    },
+    {
+      "epoch": 0.8225954901058444,
+      "grad_norm": 0.669162761336147,
+      "learning_rate": 6.377105839769149e-06,
+      "loss": 0.4643,
+      "num_input_tokens_seen": 206115352,
+      "step": 3575,
+      "train_runtime": 20077.3255,
+      "train_tokens_per_second": 10266.076
+    },
+    {
+      "epoch": 0.8237459733087897,
+      "grad_norm": 0.7296215081343778,
+      "learning_rate": 6.368417196348443e-06,
+      "loss": 0.4727,
+      "num_input_tokens_seen": 206418944,
+      "step": 3580,
+      "train_runtime": 20106.5965,
+      "train_tokens_per_second": 10266.23
+    },
+    {
+      "epoch": 0.8248964565117349,
+      "grad_norm": 0.7775050618474711,
+      "learning_rate": 6.3597240838416175e-06,
+      "loss": 0.4954,
+      "num_input_tokens_seen": 206724800,
+      "step": 3585,
+      "train_runtime": 20138.0728,
+      "train_tokens_per_second": 10265.372
+    },
+    {
+      "epoch": 0.8260469397146801,
+      "grad_norm": 0.8107603501029481,
+      "learning_rate": 6.351026530639333e-06,
+      "loss": 0.4621,
+      "num_input_tokens_seen": 207001296,
+      "step": 3590,
+      "train_runtime": 20160.6803,
+      "train_tokens_per_second": 10267.575
+    },
+    {
+      "epoch": 0.8271974229176254,
+      "grad_norm": 0.7748692171666179,
+      "learning_rate": 6.3423245651467505e-06,
+      "loss": 0.4813,
+      "num_input_tokens_seen": 207295272,
+      "step": 3595,
+      "train_runtime": 20192.1967,
+      "train_tokens_per_second": 10266.108
+    },
+    {
+      "epoch": 0.8283479061205706,
+      "grad_norm": 0.7917611557796362,
+      "learning_rate": 6.333618215783446e-06,
+      "loss": 0.4791,
+      "num_input_tokens_seen": 207600272,
+      "step": 3600,
+      "train_runtime": 20213.8297,
+      "train_tokens_per_second": 10270.21
+    },
+    {
+      "epoch": 0.8294983893235158,
+      "grad_norm": 0.7509418500695179,
+      "learning_rate": 6.32490751098331e-06,
+      "loss": 0.4471,
+      "num_input_tokens_seen": 207901800,
+      "step": 3605,
+      "train_runtime": 20238.9147,
+      "train_tokens_per_second": 10272.379
+    },
+    {
+      "epoch": 0.8306488725264611,
+      "grad_norm": 0.7737589952971572,
+      "learning_rate": 6.316192479194456e-06,
+      "loss": 0.4408,
+      "num_input_tokens_seen": 208149496,
+      "step": 3610,
+      "train_runtime": 20273.0639,
+      "train_tokens_per_second": 10267.293
+    },
+    {
+      "epoch": 0.8317993557294063,
+      "grad_norm": 0.8341720354726492,
+      "learning_rate": 6.307473148879133e-06,
+      "loss": 0.4825,
+      "num_input_tokens_seen": 208423816,
+      "step": 3615,
+      "train_runtime": 20308.7126,
+      "train_tokens_per_second": 10262.778
+    },
+    {
+      "epoch": 0.8329498389323516,
+      "grad_norm": 0.7789214419985493,
+      "learning_rate": 6.298749548513625e-06,
+      "loss": 0.472,
+      "num_input_tokens_seen": 208714688,
+      "step": 3620,
+      "train_runtime": 20340.2782,
+      "train_tokens_per_second": 10261.152
+    },
+    {
+      "epoch": 0.8341003221352968,
+      "grad_norm": 0.747913662247983,
+      "learning_rate": 6.290021706588161e-06,
+      "loss": 0.4638,
+      "num_input_tokens_seen": 208984992,
+      "step": 3625,
+      "train_runtime": 20376.237,
+      "train_tokens_per_second": 10256.309
+    },
+    {
+      "epoch": 0.835250805338242,
+      "grad_norm": 0.872880930133351,
+      "learning_rate": 6.281289651606826e-06,
+      "loss": 0.4661,
+      "num_input_tokens_seen": 209282392,
+      "step": 3630,
+      "train_runtime": 20399.95,
+      "train_tokens_per_second": 10258.966
+    },
+    {
+      "epoch": 0.8364012885411873,
+      "grad_norm": 0.8500666763199355,
+      "learning_rate": 6.272553412087462e-06,
+      "loss": 0.4957,
+      "num_input_tokens_seen": 209572256,
+      "step": 3635,
+      "train_runtime": 20424.2547,
+      "train_tokens_per_second": 10260.95
+    },
+    {
+      "epoch": 0.8375517717441325,
+      "grad_norm": 0.7600132230417115,
+      "learning_rate": 6.263813016561578e-06,
+      "loss": 0.472,
+      "num_input_tokens_seen": 209860168,
+      "step": 3640,
+      "train_runtime": 20447.2664,
+      "train_tokens_per_second": 10263.483
+    },
+    {
+      "epoch": 0.8387022549470777,
+      "grad_norm": 0.7115474608663648,
+      "learning_rate": 6.255068493574252e-06,
+      "loss": 0.4672,
+      "num_input_tokens_seen": 210151968,
+      "step": 3645,
+      "train_runtime": 20471.8478,
+      "train_tokens_per_second": 10265.413
+    },
+    {
+      "epoch": 0.839852738150023,
+      "grad_norm": 0.7283542087289824,
+      "learning_rate": 6.246319871684048e-06,
+      "loss": 0.4704,
+      "num_input_tokens_seen": 210468176,
+      "step": 3650,
+      "train_runtime": 20509.149,
+      "train_tokens_per_second": 10262.16
+    },
+    {
+      "epoch": 0.8410032213529682,
+      "grad_norm": 0.7428812243992796,
+      "learning_rate": 6.2375671794629135e-06,
+      "loss": 0.4628,
+      "num_input_tokens_seen": 210753824,
+      "step": 3655,
+      "train_runtime": 20537.878,
+      "train_tokens_per_second": 10261.714
+    },
+    {
+      "epoch": 0.8421537045559134,
+      "grad_norm": 0.9159700903486078,
+      "learning_rate": 6.22881044549609e-06,
+      "loss": 0.4932,
+      "num_input_tokens_seen": 211057624,
+      "step": 3660,
+      "train_runtime": 20564.3438,
+      "train_tokens_per_second": 10263.28
+    },
+    {
+      "epoch": 0.8433041877588587,
+      "grad_norm": 0.8194569028320462,
+      "learning_rate": 6.220049698382018e-06,
+      "loss": 0.4767,
+      "num_input_tokens_seen": 211348072,
+      "step": 3665,
+      "train_runtime": 20592.2154,
+      "train_tokens_per_second": 10263.494
+    },
+    {
+      "epoch": 0.8444546709618039,
+      "grad_norm": 0.7771311544600832,
+      "learning_rate": 6.211284966732243e-06,
+      "loss": 0.4841,
+      "num_input_tokens_seen": 211645744,
+      "step": 3670,
+      "train_runtime": 20621.9276,
+      "train_tokens_per_second": 10263.141
+    },
+    {
+      "epoch": 0.8456051541647492,
+      "grad_norm": 0.7976531207079911,
+      "learning_rate": 6.2025162791713274e-06,
+      "loss": 0.4507,
+      "num_input_tokens_seen": 211923192,
+      "step": 3675,
+      "train_runtime": 20654.7497,
+      "train_tokens_per_second": 10260.264
+    },
+    {
+      "epoch": 0.8467556373676944,
+      "grad_norm": 0.8969413553458617,
+      "learning_rate": 6.193743664336751e-06,
+      "loss": 0.4689,
+      "num_input_tokens_seen": 212196432,
+      "step": 3680,
+      "train_runtime": 20681.942,
+      "train_tokens_per_second": 10259.986
+    },
+    {
+      "epoch": 0.8479061205706396,
+      "grad_norm": 0.7860412287886049,
+      "learning_rate": 6.184967150878819e-06,
+      "loss": 0.4507,
+      "num_input_tokens_seen": 212487376,
+      "step": 3685,
+      "train_runtime": 20709.4689,
+      "train_tokens_per_second": 10260.397
+    },
+    {
+      "epoch": 0.8490566037735849,
+      "grad_norm": 0.6933136365518415,
+      "learning_rate": 6.176186767460568e-06,
+      "loss": 0.4545,
+      "num_input_tokens_seen": 212761984,
+      "step": 3690,
+      "train_runtime": 20740.6264,
+      "train_tokens_per_second": 10258.224
+    },
+    {
+      "epoch": 0.8502070869765301,
+      "grad_norm": 0.7640013057412888,
+      "learning_rate": 6.167402542757679e-06,
+      "loss": 0.4473,
+      "num_input_tokens_seen": 213029344,
+      "step": 3695,
+      "train_runtime": 20761.6388,
+      "train_tokens_per_second": 10260.719
+    },
+    {
+      "epoch": 0.8513575701794753,
+      "grad_norm": 0.8306793785918809,
+      "learning_rate": 6.158614505458371e-06,
+      "loss": 0.4597,
+      "num_input_tokens_seen": 213324968,
+      "step": 3700,
+      "train_runtime": 20791.3377,
+      "train_tokens_per_second": 10260.281
+    },
+    {
+      "epoch": 0.8525080533824206,
+      "grad_norm": 0.8854726350970317,
+      "learning_rate": 6.14982268426332e-06,
+      "loss": 0.4496,
+      "num_input_tokens_seen": 213611312,
+      "step": 3705,
+      "train_runtime": 20820.7056,
+      "train_tokens_per_second": 10259.562
+    },
+    {
+      "epoch": 0.8536585365853658,
+      "grad_norm": 0.8152464601745089,
+      "learning_rate": 6.141027107885556e-06,
+      "loss": 0.4494,
+      "num_input_tokens_seen": 213890600,
+      "step": 3710,
+      "train_runtime": 20847.646,
+      "train_tokens_per_second": 10259.7
+    },
+    {
+      "epoch": 0.8548090197883111,
+      "grad_norm": 0.8927317248751763,
+      "learning_rate": 6.132227805050375e-06,
+      "loss": 0.461,
+      "num_input_tokens_seen": 214157312,
+      "step": 3715,
+      "train_runtime": 20874.1988,
+      "train_tokens_per_second": 10259.427
+    },
+    {
+      "epoch": 0.8559595029912563,
+      "grad_norm": 0.8664592136069698,
+      "learning_rate": 6.123424804495242e-06,
+      "loss": 0.4787,
+      "num_input_tokens_seen": 214466968,
+      "step": 3720,
+      "train_runtime": 20902.0694,
+      "train_tokens_per_second": 10260.561
+    },
+    {
+      "epoch": 0.8571099861942015,
+      "grad_norm": 0.7577010296084473,
+      "learning_rate": 6.114618134969698e-06,
+      "loss": 0.471,
+      "num_input_tokens_seen": 214751648,
+      "step": 3725,
+      "train_runtime": 20932.22,
+      "train_tokens_per_second": 10259.382
+    },
+    {
+      "epoch": 0.8582604693971468,
+      "grad_norm": 0.7887188457032649,
+      "learning_rate": 6.105807825235272e-06,
+      "loss": 0.4489,
+      "num_input_tokens_seen": 215033248,
+      "step": 3730,
+      "train_runtime": 20952.6908,
+      "train_tokens_per_second": 10262.799
+    },
+    {
+      "epoch": 0.859410952600092,
+      "grad_norm": 0.7067079785945384,
+      "learning_rate": 6.0969939040653725e-06,
+      "loss": 0.4555,
+      "num_input_tokens_seen": 215316912,
+      "step": 3735,
+      "train_runtime": 20977.6272,
+      "train_tokens_per_second": 10264.121
+    },
+    {
+      "epoch": 0.8605614358030372,
+      "grad_norm": 0.9114882170218999,
+      "learning_rate": 6.088176400245208e-06,
+      "loss": 0.4832,
+      "num_input_tokens_seen": 215588504,
+      "step": 3740,
+      "train_runtime": 21006.0532,
+      "train_tokens_per_second": 10263.161
+    },
+    {
+      "epoch": 0.8617119190059825,
+      "grad_norm": 0.7591813069921569,
+      "learning_rate": 6.079355342571685e-06,
+      "loss": 0.4552,
+      "num_input_tokens_seen": 215873136,
+      "step": 3745,
+      "train_runtime": 21036.0349,
+      "train_tokens_per_second": 10262.064
+    },
+    {
+      "epoch": 0.8628624022089277,
+      "grad_norm": 0.7774251262868298,
+      "learning_rate": 6.070530759853321e-06,
+      "loss": 0.4727,
+      "num_input_tokens_seen": 216151960,
+      "step": 3750,
+      "train_runtime": 21069.4716,
+      "train_tokens_per_second": 10259.012
+    },
+    {
+      "epoch": 0.8640128854118729,
+      "grad_norm": 0.7624754952357142,
+      "learning_rate": 6.061702680910142e-06,
+      "loss": 0.4821,
+      "num_input_tokens_seen": 216439840,
+      "step": 3755,
+      "train_runtime": 21103.7188,
+      "train_tokens_per_second": 10256.005
+    },
+    {
+      "epoch": 0.8651633686148182,
+      "grad_norm": 0.8151147635624715,
+      "learning_rate": 6.05287113457359e-06,
+      "loss": 0.4814,
+      "num_input_tokens_seen": 216730600,
+      "step": 3760,
+      "train_runtime": 21122.401,
+      "train_tokens_per_second": 10260.699
+    },
+    {
+      "epoch": 0.8663138518177634,
+      "grad_norm": 0.8512324130464384,
+      "learning_rate": 6.044036149686436e-06,
+      "loss": 0.4515,
+      "num_input_tokens_seen": 217022248,
+      "step": 3765,
+      "train_runtime": 21146.6778,
+      "train_tokens_per_second": 10262.711
+    },
+    {
+      "epoch": 0.8674643350207087,
+      "grad_norm": 0.7253129184005315,
+      "learning_rate": 6.03519775510268e-06,
+      "loss": 0.4949,
+      "num_input_tokens_seen": 217299440,
+      "step": 3770,
+      "train_runtime": 21178.4958,
+      "train_tokens_per_second": 10260.381
+    },
+    {
+      "epoch": 0.8686148182236539,
+      "grad_norm": 0.9019536853186185,
+      "learning_rate": 6.026355979687455e-06,
+      "loss": 0.4699,
+      "num_input_tokens_seen": 217582728,
+      "step": 3775,
+      "train_runtime": 21203.2557,
+      "train_tokens_per_second": 10261.76
+    },
+    {
+      "epoch": 0.8697653014265991,
+      "grad_norm": 0.7417044967518114,
+      "learning_rate": 6.017510852316939e-06,
+      "loss": 0.4506,
+      "num_input_tokens_seen": 217867080,
+      "step": 3780,
+      "train_runtime": 21224.0519,
+      "train_tokens_per_second": 10265.103
+    },
+    {
+      "epoch": 0.8709157846295444,
+      "grad_norm": 0.687792045397184,
+      "learning_rate": 6.008662401878251e-06,
+      "loss": 0.4726,
+      "num_input_tokens_seen": 218174520,
+      "step": 3785,
+      "train_runtime": 21257.323,
+      "train_tokens_per_second": 10263.499
+    },
+    {
+      "epoch": 0.8720662678324896,
+      "grad_norm": 0.8075538059049464,
+      "learning_rate": 5.9998106572693714e-06,
+      "loss": 0.4701,
+      "num_input_tokens_seen": 218482200,
+      "step": 3790,
+      "train_runtime": 21281.2609,
+      "train_tokens_per_second": 10266.412
+    },
+    {
+      "epoch": 0.8732167510354348,
+      "grad_norm": 0.7351227151093981,
+      "learning_rate": 5.990955647399032e-06,
+      "loss": 0.4904,
+      "num_input_tokens_seen": 218793200,
+      "step": 3795,
+      "train_runtime": 21314.7398,
+      "train_tokens_per_second": 10264.878
+    },
+    {
+      "epoch": 0.8743672342383801,
+      "grad_norm": 0.8842670416599772,
+      "learning_rate": 5.98209740118663e-06,
+      "loss": 0.44,
+      "num_input_tokens_seen": 219097936,
+      "step": 3800,
+      "train_runtime": 21340.3767,
+      "train_tokens_per_second": 10266.826
+    },
+    {
+      "epoch": 0.8755177174413253,
+      "grad_norm": 0.830883427410278,
+      "learning_rate": 5.973235947562137e-06,
+      "loss": 0.4503,
+      "num_input_tokens_seen": 219387808,
+      "step": 3805,
+      "train_runtime": 21366.9109,
+      "train_tokens_per_second": 10267.643
+    },
+    {
+      "epoch": 0.8766682006442706,
+      "grad_norm": 1.826895850317921,
+      "learning_rate": 5.964371315465992e-06,
+      "loss": 0.482,
+      "num_input_tokens_seen": 219679184,
+      "step": 3810,
+      "train_runtime": 21396.4942,
+      "train_tokens_per_second": 10267.064
+    },
+    {
+      "epoch": 0.8778186838472158,
+      "grad_norm": 0.8035599415132906,
+      "learning_rate": 5.955503533849018e-06,
+      "loss": 0.4662,
+      "num_input_tokens_seen": 219990280,
+      "step": 3815,
+      "train_runtime": 21434.2491,
+      "train_tokens_per_second": 10263.494
+    },
+    {
+      "epoch": 0.878969167050161,
+      "grad_norm": 0.9701839215980341,
+      "learning_rate": 5.946632631672327e-06,
+      "loss": 0.4672,
+      "num_input_tokens_seen": 220272416,
+      "step": 3820,
+      "train_runtime": 21460.5859,
+      "train_tokens_per_second": 10264.045
+    },
+    {
+      "epoch": 0.8801196502531063,
+      "grad_norm": 0.7380079493207466,
+      "learning_rate": 5.937758637907216e-06,
+      "loss": 0.4323,
+      "num_input_tokens_seen": 220560072,
+      "step": 3825,
+      "train_runtime": 21479.4969,
+      "train_tokens_per_second": 10268.4
+    },
+    {
+      "epoch": 0.8812701334560515,
+      "grad_norm": 0.7102951396271845,
+      "learning_rate": 5.928881581535086e-06,
+      "loss": 0.4491,
+      "num_input_tokens_seen": 220845616,
+      "step": 3830,
+      "train_runtime": 21507.8409,
+      "train_tokens_per_second": 10268.144
+    },
+    {
+      "epoch": 0.8824206166589967,
+      "grad_norm": 0.800373469843162,
+      "learning_rate": 5.920001491547334e-06,
+      "loss": 0.4798,
+      "num_input_tokens_seen": 221102032,
+      "step": 3835,
+      "train_runtime": 21539.4038,
+      "train_tokens_per_second": 10265.002
+    },
+    {
+      "epoch": 0.883571099861942,
+      "grad_norm": 0.9459883297025254,
+      "learning_rate": 5.911118396945265e-06,
+      "loss": 0.5094,
+      "num_input_tokens_seen": 221398576,
+      "step": 3840,
+      "train_runtime": 21566.4206,
+      "train_tokens_per_second": 10265.893
+    },
+    {
+      "epoch": 0.8847215830648872,
+      "grad_norm": 0.7210229777202655,
+      "learning_rate": 5.902232326740004e-06,
+      "loss": 0.4763,
+      "num_input_tokens_seen": 221685216,
+      "step": 3845,
+      "train_runtime": 21594.5425,
+      "train_tokens_per_second": 10265.798
+    },
+    {
+      "epoch": 0.8858720662678324,
+      "grad_norm": 0.851381412264532,
+      "learning_rate": 5.893343309952384e-06,
+      "loss": 0.4715,
+      "num_input_tokens_seen": 221988528,
+      "step": 3850,
+      "train_runtime": 21625.9858,
+      "train_tokens_per_second": 10264.898
+    },
+    {
+      "epoch": 0.8870225494707777,
+      "grad_norm": 2.2953659249331344,
+      "learning_rate": 5.884451375612865e-06,
+      "loss": 0.4751,
+      "num_input_tokens_seen": 222290688,
+      "step": 3855,
+      "train_runtime": 21655.74,
+      "train_tokens_per_second": 10264.747
+    },
+    {
+      "epoch": 0.8881730326737229,
+      "grad_norm": 0.7532702637461327,
+      "learning_rate": 5.87555655276144e-06,
+      "loss": 0.4714,
+      "num_input_tokens_seen": 222585664,
+      "step": 3860,
+      "train_runtime": 21688.5648,
+      "train_tokens_per_second": 10262.812
+    },
+    {
+      "epoch": 0.8893235158766682,
+      "grad_norm": 0.785608291407384,
+      "learning_rate": 5.866658870447528e-06,
+      "loss": 0.453,
+      "num_input_tokens_seen": 222878536,
+      "step": 3865,
+      "train_runtime": 21715.4814,
+      "train_tokens_per_second": 10263.578
+    },
+    {
+      "epoch": 0.8904739990796134,
+      "grad_norm": 0.7299235505222788,
+      "learning_rate": 5.857758357729892e-06,
+      "loss": 0.4624,
+      "num_input_tokens_seen": 223155656,
+      "step": 3870,
+      "train_runtime": 21747.2258,
+      "train_tokens_per_second": 10261.339
+    },
+    {
+      "epoch": 0.8916244822825586,
+      "grad_norm": 0.8485706310257409,
+      "learning_rate": 5.848855043676537e-06,
+      "loss": 0.476,
+      "num_input_tokens_seen": 223437112,
+      "step": 3875,
+      "train_runtime": 21778.8647,
+      "train_tokens_per_second": 10259.355
+    },
+    {
+      "epoch": 0.892774965485504,
+      "grad_norm": 0.745903894026012,
+      "learning_rate": 5.839948957364615e-06,
+      "loss": 0.4651,
+      "num_input_tokens_seen": 223742912,
+      "step": 3880,
+      "train_runtime": 21807.2686,
+      "train_tokens_per_second": 10260.015
+    },
+    {
+      "epoch": 0.8939254486884491,
+      "grad_norm": 0.9499252808364167,
+      "learning_rate": 5.831040127880337e-06,
+      "loss": 0.4818,
+      "num_input_tokens_seen": 224021248,
+      "step": 3885,
+      "train_runtime": 21831.9659,
+      "train_tokens_per_second": 10261.158
+    },
+    {
+      "epoch": 0.8950759318913943,
+      "grad_norm": 0.794100288053391,
+      "learning_rate": 5.822128584318865e-06,
+      "loss": 0.4755,
+      "num_input_tokens_seen": 224323072,
+      "step": 3890,
+      "train_runtime": 21856.3215,
+      "train_tokens_per_second": 10263.533
+    },
+    {
+      "epoch": 0.8962264150943396,
+      "grad_norm": 0.7496047013431127,
+      "learning_rate": 5.813214355784232e-06,
+      "loss": 0.4603,
+      "num_input_tokens_seen": 224625376,
+      "step": 3895,
+      "train_runtime": 21883.8021,
+      "train_tokens_per_second": 10264.458
+    },
+    {
+      "epoch": 0.8973768982972848,
+      "grad_norm": 0.8851270109649124,
+      "learning_rate": 5.80429747138924e-06,
+      "loss": 0.4565,
+      "num_input_tokens_seen": 224924856,
+      "step": 3900,
+      "train_runtime": 21913.9584,
+      "train_tokens_per_second": 10263.999
+    },
+    {
+      "epoch": 0.8985273815002301,
+      "grad_norm": 0.9086903972268473,
+      "learning_rate": 5.795377960255356e-06,
+      "loss": 0.4601,
+      "num_input_tokens_seen": 225166552,
+      "step": 3905,
+      "train_runtime": 21936.6501,
+      "train_tokens_per_second": 10264.4
+    },
+    {
+      "epoch": 0.8996778647031753,
+      "grad_norm": 0.8244545744198543,
+      "learning_rate": 5.786455851512636e-06,
+      "loss": 0.4626,
+      "num_input_tokens_seen": 225438112,
+      "step": 3910,
+      "train_runtime": 21965.1443,
+      "train_tokens_per_second": 10263.448
+    },
+    {
+      "epoch": 0.9008283479061205,
+      "grad_norm": 0.8562537707612213,
+      "learning_rate": 5.7775311742996125e-06,
+      "loss": 0.4459,
+      "num_input_tokens_seen": 225726208,
+      "step": 3915,
+      "train_runtime": 21991.2298,
+      "train_tokens_per_second": 10264.374
+    },
+    {
+      "epoch": 0.9019788311090658,
+      "grad_norm": 1.6167273613749633,
+      "learning_rate": 5.768603957763209e-06,
+      "loss": 0.4412,
+      "num_input_tokens_seen": 226009408,
+      "step": 3920,
+      "train_runtime": 22015.2368,
+      "train_tokens_per_second": 10266.045
+    },
+    {
+      "epoch": 0.903129314312011,
+      "grad_norm": 0.7703223797525869,
+      "learning_rate": 5.759674231058643e-06,
+      "loss": 0.4736,
+      "num_input_tokens_seen": 226274808,
+      "step": 3925,
+      "train_runtime": 22043.5059,
+      "train_tokens_per_second": 10264.919
+    },
+    {
+      "epoch": 0.9042797975149562,
+      "grad_norm": 0.723195366256002,
+      "learning_rate": 5.7507420233493294e-06,
+      "loss": 0.4857,
+      "num_input_tokens_seen": 226552536,
+      "step": 3930,
+      "train_runtime": 22080.7925,
+      "train_tokens_per_second": 10260.163
+    },
+    {
+      "epoch": 0.9054302807179015,
+      "grad_norm": 1.1126223414402407,
+      "learning_rate": 5.741807363806784e-06,
+      "loss": 0.4717,
+      "num_input_tokens_seen": 226851024,
+      "step": 3935,
+      "train_runtime": 22105.0525,
+      "train_tokens_per_second": 10262.406
+    },
+    {
+      "epoch": 0.9065807639208467,
+      "grad_norm": 0.8601599410371354,
+      "learning_rate": 5.732870281610529e-06,
+      "loss": 0.4674,
+      "num_input_tokens_seen": 227113296,
+      "step": 3940,
+      "train_runtime": 22124.8322,
+      "train_tokens_per_second": 10265.086
+    },
+    {
+      "epoch": 0.9077312471237919,
+      "grad_norm": 0.718042742084185,
+      "learning_rate": 5.723930805948008e-06,
+      "loss": 0.4465,
+      "num_input_tokens_seen": 227413288,
+      "step": 3945,
+      "train_runtime": 22153.5678,
+      "train_tokens_per_second": 10265.312
+    },
+    {
+      "epoch": 0.9088817303267372,
+      "grad_norm": 0.8379168746720947,
+      "learning_rate": 5.714988966014465e-06,
+      "loss": 0.4564,
+      "num_input_tokens_seen": 227685904,
+      "step": 3950,
+      "train_runtime": 22186.7618,
+      "train_tokens_per_second": 10262.241
+    },
+    {
+      "epoch": 0.9100322135296824,
+      "grad_norm": 0.7660656718758599,
+      "learning_rate": 5.706044791012882e-06,
+      "loss": 0.4624,
+      "num_input_tokens_seen": 227987184,
+      "step": 3955,
+      "train_runtime": 22208.3863,
+      "train_tokens_per_second": 10265.815
+    },
+    {
+      "epoch": 0.9111826967326278,
+      "grad_norm": 0.7133434607590677,
+      "learning_rate": 5.697098310153856e-06,
+      "loss": 0.4557,
+      "num_input_tokens_seen": 228273088,
+      "step": 3960,
+      "train_runtime": 22241.0002,
+      "train_tokens_per_second": 10263.616
+    },
+    {
+      "epoch": 0.912333179935573,
+      "grad_norm": 0.7841124284606561,
+      "learning_rate": 5.68814955265552e-06,
+      "loss": 0.4314,
+      "num_input_tokens_seen": 228537488,
+      "step": 3965,
+      "train_runtime": 22271.1592,
+      "train_tokens_per_second": 10261.589
+    },
+    {
+      "epoch": 0.9134836631385181,
+      "grad_norm": 0.7749870254515138,
+      "learning_rate": 5.67919854774344e-06,
+      "loss": 0.4907,
+      "num_input_tokens_seen": 228830768,
+      "step": 3970,
+      "train_runtime": 22294.4728,
+      "train_tokens_per_second": 10264.013
+    },
+    {
+      "epoch": 0.9146341463414634,
+      "grad_norm": 0.7393997488792055,
+      "learning_rate": 5.67024532465052e-06,
+      "loss": 0.4709,
+      "num_input_tokens_seen": 229130448,
+      "step": 3975,
+      "train_runtime": 22328.6982,
+      "train_tokens_per_second": 10261.702
+    },
+    {
+      "epoch": 0.9157846295444086,
+      "grad_norm": 0.7943893629622056,
+      "learning_rate": 5.6612899126169154e-06,
+      "loss": 0.4572,
+      "num_input_tokens_seen": 229398360,
+      "step": 3980,
+      "train_runtime": 22355.0044,
+      "train_tokens_per_second": 10261.611
+    },
+    {
+      "epoch": 0.9169351127473538,
+      "grad_norm": 0.8854024125812803,
+      "learning_rate": 5.652332340889923e-06,
+      "loss": 0.4496,
+      "num_input_tokens_seen": 229632160,
+      "step": 3985,
+      "train_runtime": 22382.7618,
+      "train_tokens_per_second": 10259.331
+    },
+    {
+      "epoch": 0.9180855959502991,
+      "grad_norm": 0.7617908638851778,
+      "learning_rate": 5.643372638723895e-06,
+      "loss": 0.4711,
+      "num_input_tokens_seen": 229904032,
+      "step": 3990,
+      "train_runtime": 22414.8571,
+      "train_tokens_per_second": 10256.77
+    },
+    {
+      "epoch": 0.9192360791532443,
+      "grad_norm": 0.7311577613755941,
+      "learning_rate": 5.634410835380144e-06,
+      "loss": 0.433,
+      "num_input_tokens_seen": 230174792,
+      "step": 3995,
+      "train_runtime": 22441.6005,
+      "train_tokens_per_second": 10256.612
+    },
+    {
+      "epoch": 0.9203865623561897,
+      "grad_norm": 0.7945277457210232,
+      "learning_rate": 5.625446960126844e-06,
+      "loss": 0.4675,
+      "num_input_tokens_seen": 230454488,
+      "step": 4000,
+      "train_runtime": 22472.4239,
+      "train_tokens_per_second": 10254.99
+    },
+    {
+      "epoch": 0.9215370455591348,
+      "grad_norm": 0.822763357273076,
+      "learning_rate": 5.6164810422389285e-06,
+      "loss": 0.4795,
+      "num_input_tokens_seen": 230713680,
+      "step": 4005,
+      "train_runtime": 22498.4171,
+      "train_tokens_per_second": 10254.663
+    },
+    {
+      "epoch": 0.92268752876208,
+      "grad_norm": 0.7253683411246508,
+      "learning_rate": 5.607513110998013e-06,
+      "loss": 0.4686,
+      "num_input_tokens_seen": 231013000,
+      "step": 4010,
+      "train_runtime": 22524.5929,
+      "train_tokens_per_second": 10256.034
+    },
+    {
+      "epoch": 0.9238380119650254,
+      "grad_norm": 0.951613925895999,
+      "learning_rate": 5.598543195692282e-06,
+      "loss": 0.4474,
+      "num_input_tokens_seen": 231290736,
+      "step": 4015,
+      "train_runtime": 22551.0408,
+      "train_tokens_per_second": 10256.322
+    },
+    {
+      "epoch": 0.9249884951679705,
+      "grad_norm": 0.7146448110555723,
+      "learning_rate": 5.589571325616399e-06,
+      "loss": 0.4587,
+      "num_input_tokens_seen": 231577496,
+      "step": 4020,
+      "train_runtime": 22575.1708,
+      "train_tokens_per_second": 10258.062
+    },
+    {
+      "epoch": 0.9261389783709157,
+      "grad_norm": 0.9007085422234455,
+      "learning_rate": 5.580597530071419e-06,
+      "loss": 0.4825,
+      "num_input_tokens_seen": 231861944,
+      "step": 4025,
+      "train_runtime": 22604.9214,
+      "train_tokens_per_second": 10257.144
+    },
+    {
+      "epoch": 0.927289461573861,
+      "grad_norm": 0.7637911539168034,
+      "learning_rate": 5.571621838364674e-06,
+      "loss": 0.4666,
+      "num_input_tokens_seen": 232156880,
+      "step": 4030,
+      "train_runtime": 22629.0304,
+      "train_tokens_per_second": 10259.25
+    },
+    {
+      "epoch": 0.9284399447768062,
+      "grad_norm": 0.8154531234225789,
+      "learning_rate": 5.562644279809697e-06,
+      "loss": 0.4557,
+      "num_input_tokens_seen": 232458056,
+      "step": 4035,
+      "train_runtime": 22650.9641,
+      "train_tokens_per_second": 10262.612
+    },
+    {
+      "epoch": 0.9295904279797514,
+      "grad_norm": 0.8097079944222262,
+      "learning_rate": 5.553664883726115e-06,
+      "loss": 0.4645,
+      "num_input_tokens_seen": 232719096,
+      "step": 4040,
+      "train_runtime": 22678.903,
+      "train_tokens_per_second": 10261.479
+    },
+    {
+      "epoch": 0.9307409111826968,
+      "grad_norm": 0.7258910614098031,
+      "learning_rate": 5.544683679439556e-06,
+      "loss": 0.4439,
+      "num_input_tokens_seen": 233002056,
+      "step": 4045,
+      "train_runtime": 22706.903,
+      "train_tokens_per_second": 10261.287
+    },
+    {
+      "epoch": 0.931891394385642,
+      "grad_norm": 0.724067044903452,
+      "learning_rate": 5.5357006962815565e-06,
+      "loss": 0.455,
+      "num_input_tokens_seen": 233301744,
+      "step": 4050,
+      "train_runtime": 22736.3332,
+      "train_tokens_per_second": 10261.186
+    },
+    {
+      "epoch": 0.9330418775885873,
+      "grad_norm": 0.7872715020860507,
+      "learning_rate": 5.5267159635894565e-06,
+      "loss": 0.4778,
+      "num_input_tokens_seen": 233588312,
+      "step": 4055,
+      "train_runtime": 22760.5197,
+      "train_tokens_per_second": 10262.872
+    },
+    {
+      "epoch": 0.9341923607915325,
+      "grad_norm": 0.7999347350595398,
+      "learning_rate": 5.517729510706316e-06,
+      "loss": 0.4232,
+      "num_input_tokens_seen": 233865440,
+      "step": 4060,
+      "train_runtime": 22794.3517,
+      "train_tokens_per_second": 10259.798
+    },
+    {
+      "epoch": 0.9353428439944776,
+      "grad_norm": 0.9154903263202742,
+      "learning_rate": 5.508741366980809e-06,
+      "loss": 0.4814,
+      "num_input_tokens_seen": 234158504,
+      "step": 4065,
+      "train_runtime": 22814.3672,
+      "train_tokens_per_second": 10263.642
+    },
+    {
+      "epoch": 0.936493327197423,
+      "grad_norm": 0.6526366744065775,
+      "learning_rate": 5.499751561767131e-06,
+      "loss": 0.4384,
+      "num_input_tokens_seen": 234444888,
+      "step": 4070,
+      "train_runtime": 22849.7018,
+      "train_tokens_per_second": 10260.304
+    },
+    {
+      "epoch": 0.9376438104003681,
+      "grad_norm": 0.6967262639605767,
+      "learning_rate": 5.490760124424909e-06,
+      "loss": 0.4563,
+      "num_input_tokens_seen": 234729368,
+      "step": 4075,
+      "train_runtime": 22876.5519,
+      "train_tokens_per_second": 10260.697
+    },
+    {
+      "epoch": 0.9387942936033133,
+      "grad_norm": 0.6917759776497254,
+      "learning_rate": 5.481767084319095e-06,
+      "loss": 0.4593,
+      "num_input_tokens_seen": 235033328,
+      "step": 4080,
+      "train_runtime": 22909.2631,
+      "train_tokens_per_second": 10259.314
+    },
+    {
+      "epoch": 0.9399447768062587,
+      "grad_norm": 0.7036537673687676,
+      "learning_rate": 5.472772470819877e-06,
+      "loss": 0.4698,
+      "num_input_tokens_seen": 235301064,
+      "step": 4085,
+      "train_runtime": 22942.5944,
+      "train_tokens_per_second": 10256.079
+    },
+    {
+      "epoch": 0.9410952600092038,
+      "grad_norm": 0.7789282099656489,
+      "learning_rate": 5.463776313302583e-06,
+      "loss": 0.4593,
+      "num_input_tokens_seen": 235604288,
+      "step": 4090,
+      "train_runtime": 22975.9587,
+      "train_tokens_per_second": 10254.383
+    },
+    {
+      "epoch": 0.9422457432121492,
+      "grad_norm": 0.7777857333147328,
+      "learning_rate": 5.4547786411475826e-06,
+      "loss": 0.4808,
+      "num_input_tokens_seen": 235894656,
+      "step": 4095,
+      "train_runtime": 23006.8904,
+      "train_tokens_per_second": 10253.218
+    },
+    {
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.8026466094151004,
+      "learning_rate": 5.44577948374019e-06,
+      "loss": 0.5118,
+      "num_input_tokens_seen": 236169976,
+      "step": 4100,
+      "train_runtime": 23036.8399,
+      "train_tokens_per_second": 10251.839
+    },
+    {
+      "epoch": 0.9445467096180395,
+      "grad_norm": 0.7353096345297931,
+      "learning_rate": 5.4367788704705725e-06,
+      "loss": 0.4772,
+      "num_input_tokens_seen": 236448432,
+      "step": 4105,
+      "train_runtime": 23065.3723,
+      "train_tokens_per_second": 10251.23
+    },
+    {
+      "epoch": 0.9456971928209849,
+      "grad_norm": 0.828744047199074,
+      "learning_rate": 5.427776830733652e-06,
+      "loss": 0.4425,
+      "num_input_tokens_seen": 236728352,
+      "step": 4110,
+      "train_runtime": 23095.0925,
+      "train_tokens_per_second": 10250.158
+    },
+    {
+      "epoch": 0.94684767602393,
+      "grad_norm": 0.7797148768873371,
+      "learning_rate": 5.418773393929009e-06,
+      "loss": 0.4883,
+      "num_input_tokens_seen": 237034496,
+      "step": 4115,
+      "train_runtime": 23126.4272,
+      "train_tokens_per_second": 10249.508
+    },
+    {
+      "epoch": 0.9479981592268752,
+      "grad_norm": 0.7806438937611688,
+      "learning_rate": 5.409768589460786e-06,
+      "loss": 0.4619,
+      "num_input_tokens_seen": 237305240,
+      "step": 4120,
+      "train_runtime": 23150.2359,
+      "train_tokens_per_second": 10250.662
+    },
+    {
+      "epoch": 0.9491486424298206,
+      "grad_norm": 0.9357037202614625,
+      "learning_rate": 5.40076244673759e-06,
+      "loss": 0.4467,
+      "num_input_tokens_seen": 237568480,
+      "step": 4125,
+      "train_runtime": 23177.2136,
+      "train_tokens_per_second": 10250.088
+    },
+    {
+      "epoch": 0.9502991256327658,
+      "grad_norm": 0.7164601407417234,
+      "learning_rate": 5.391754995172403e-06,
+      "loss": 0.4612,
+      "num_input_tokens_seen": 237861424,
+      "step": 4130,
+      "train_runtime": 23202.0432,
+      "train_tokens_per_second": 10251.745
+    },
+    {
+      "epoch": 0.951449608835711,
+      "grad_norm": 0.6526853702491786,
+      "learning_rate": 5.38274626418248e-06,
+      "loss": 0.4425,
+      "num_input_tokens_seen": 238147304,
+      "step": 4135,
+      "train_runtime": 23223.9936,
+      "train_tokens_per_second": 10254.365
+    },
+    {
+      "epoch": 0.9526000920386563,
+      "grad_norm": 0.786329687537114,
+      "learning_rate": 5.37373628318925e-06,
+      "loss": 0.4945,
+      "num_input_tokens_seen": 238452072,
+      "step": 4140,
+      "train_runtime": 23252.1082,
+      "train_tokens_per_second": 10255.073
+    },
+    {
+      "epoch": 0.9537505752416015,
+      "grad_norm": 0.7426612879633065,
+      "learning_rate": 5.3647250816182336e-06,
+      "loss": 0.4814,
+      "num_input_tokens_seen": 238719624,
+      "step": 4145,
+      "train_runtime": 23282.5567,
+      "train_tokens_per_second": 10253.153
+    },
+    {
+      "epoch": 0.9549010584445468,
+      "grad_norm": 0.7854874726049166,
+      "learning_rate": 5.3557126888989295e-06,
+      "loss": 0.4746,
+      "num_input_tokens_seen": 239000600,
+      "step": 4150,
+      "train_runtime": 23311.8413,
+      "train_tokens_per_second": 10252.326
+    },
+    {
+      "epoch": 0.956051541647492,
+      "grad_norm": 0.7589502956859765,
+      "learning_rate": 5.346699134464729e-06,
+      "loss": 0.4495,
+      "num_input_tokens_seen": 239306304,
+      "step": 4155,
+      "train_runtime": 23346.1508,
+      "train_tokens_per_second": 10250.354
+    },
+    {
+      "epoch": 0.9572020248504371,
+      "grad_norm": 0.707090468283349,
+      "learning_rate": 5.337684447752818e-06,
+      "loss": 0.4572,
+      "num_input_tokens_seen": 239594736,
+      "step": 4160,
+      "train_runtime": 23369.1862,
+      "train_tokens_per_second": 10252.592
+    },
+    {
+      "epoch": 0.9583525080533825,
+      "grad_norm": 0.8799939543227425,
+      "learning_rate": 5.328668658204081e-06,
+      "loss": 0.461,
+      "num_input_tokens_seen": 239862168,
+      "step": 4165,
+      "train_runtime": 23393.6591,
+      "train_tokens_per_second": 10253.298
+    },
+    {
+      "epoch": 0.9595029912563277,
+      "grad_norm": 1.0913370998409717,
+      "learning_rate": 5.319651795263002e-06,
+      "loss": 0.4445,
+      "num_input_tokens_seen": 240124896,
+      "step": 4170,
+      "train_runtime": 23418.5786,
+      "train_tokens_per_second": 10253.607
+    },
+    {
+      "epoch": 0.9606534744592728,
+      "grad_norm": 0.8029530066129562,
+      "learning_rate": 5.310633888377572e-06,
+      "loss": 0.4788,
+      "num_input_tokens_seen": 240407288,
+      "step": 4175,
+      "train_runtime": 23447.9721,
+      "train_tokens_per_second": 10252.797
+    },
+    {
+      "epoch": 0.9618039576622182,
+      "grad_norm": 0.8484570084726811,
+      "learning_rate": 5.301614966999191e-06,
+      "loss": 0.4463,
+      "num_input_tokens_seen": 240683024,
+      "step": 4180,
+      "train_runtime": 23474.7394,
+      "train_tokens_per_second": 10252.852
+    },
+    {
+      "epoch": 0.9629544408651634,
+      "grad_norm": 0.6941322982636642,
+      "learning_rate": 5.2925950605825726e-06,
+      "loss": 0.464,
+      "num_input_tokens_seen": 240992640,
+      "step": 4185,
+      "train_runtime": 23499.4764,
+      "train_tokens_per_second": 10255.234
+    },
+    {
+      "epoch": 0.9641049240681087,
+      "grad_norm": 0.7437515787487572,
+      "learning_rate": 5.283574198585649e-06,
+      "loss": 0.4481,
+      "num_input_tokens_seen": 241304328,
+      "step": 4190,
+      "train_runtime": 23524.1833,
+      "train_tokens_per_second": 10257.713
+    },
+    {
+      "epoch": 0.9652554072710539,
+      "grad_norm": 0.8615207288149016,
+      "learning_rate": 5.274552410469469e-06,
+      "loss": 0.4572,
+      "num_input_tokens_seen": 241573952,
+      "step": 4195,
+      "train_runtime": 23554.3326,
+      "train_tokens_per_second": 10256.03
+    },
+    {
+      "epoch": 0.966405890473999,
+      "grad_norm": 0.8262730111003602,
+      "learning_rate": 5.26552972569811e-06,
+      "loss": 0.4649,
+      "num_input_tokens_seen": 241851976,
+      "step": 4200,
+      "train_runtime": 23577.356,
+      "train_tokens_per_second": 10257.807
+    },
+    {
+      "epoch": 0.9675563736769444,
+      "grad_norm": 0.7954758835877728,
+      "learning_rate": 5.2565061737385744e-06,
+      "loss": 0.4649,
+      "num_input_tokens_seen": 242144416,
+      "step": 4205,
+      "train_runtime": 23601.4898,
+      "train_tokens_per_second": 10259.709
+    },
+    {
+      "epoch": 0.9687068568798896,
+      "grad_norm": 1.2750102723070045,
+      "learning_rate": 5.2474817840607e-06,
+      "loss": 0.4783,
+      "num_input_tokens_seen": 242444696,
+      "step": 4210,
+      "train_runtime": 23635.0866,
+      "train_tokens_per_second": 10257.83
+    },
+    {
+      "epoch": 0.9698573400828348,
+      "grad_norm": 0.7666188823844923,
+      "learning_rate": 5.238456586137061e-06,
+      "loss": 0.443,
+      "num_input_tokens_seen": 242735736,
+      "step": 4215,
+      "train_runtime": 23662.1712,
+      "train_tokens_per_second": 10258.388
+    },
+    {
+      "epoch": 0.9710078232857801,
+      "grad_norm": 0.8600345409396983,
+      "learning_rate": 5.229430609442863e-06,
+      "loss": 0.4662,
+      "num_input_tokens_seen": 243025016,
+      "step": 4220,
+      "train_runtime": 23694.1745,
+      "train_tokens_per_second": 10256.741
+    },
+    {
+      "epoch": 0.9721583064887253,
+      "grad_norm": 0.7615713483191652,
+      "learning_rate": 5.220403883455869e-06,
+      "loss": 0.4892,
+      "num_input_tokens_seen": 243305056,
+      "step": 4225,
+      "train_runtime": 23721.5658,
+      "train_tokens_per_second": 10256.703
+    },
+    {
+      "epoch": 0.9733087896916705,
+      "grad_norm": 0.6628111093294257,
+      "learning_rate": 5.211376437656276e-06,
+      "loss": 0.437,
+      "num_input_tokens_seen": 243606568,
+      "step": 4230,
+      "train_runtime": 23752.964,
+      "train_tokens_per_second": 10255.839
+    },
+    {
+      "epoch": 0.9744592728946158,
+      "grad_norm": 0.9360452015301092,
+      "learning_rate": 5.202348301526638e-06,
+      "loss": 0.4737,
+      "num_input_tokens_seen": 243905952,
+      "step": 4235,
+      "train_runtime": 23784.375,
+      "train_tokens_per_second": 10254.882
+    },
+    {
+      "epoch": 0.975609756097561,
+      "grad_norm": 0.8415599479856687,
+      "learning_rate": 5.193319504551765e-06,
+      "loss": 0.4862,
+      "num_input_tokens_seen": 244175552,
+      "step": 4240,
+      "train_runtime": 23811.9042,
+      "train_tokens_per_second": 10254.348
+    },
+    {
+      "epoch": 0.9767602393005063,
+      "grad_norm": 0.7246526060577209,
+      "learning_rate": 5.18429007621862e-06,
+      "loss": 0.427,
+      "num_input_tokens_seen": 244466296,
+      "step": 4245,
+      "train_runtime": 23840.1253,
+      "train_tokens_per_second": 10254.405
+    },
+    {
+      "epoch": 0.9779107225034515,
+      "grad_norm": 1.5314570588861347,
+      "learning_rate": 5.175260046016233e-06,
+      "loss": 0.4573,
+      "num_input_tokens_seen": 244787616,
+      "step": 4250,
+      "train_runtime": 23866.7932,
+      "train_tokens_per_second": 10256.41
+    },
+    {
+      "epoch": 0.9790612057063967,
+      "grad_norm": 0.7523012896233956,
+      "learning_rate": 5.166229443435597e-06,
+      "loss": 0.4601,
+      "num_input_tokens_seen": 245066488,
+      "step": 4255,
+      "train_runtime": 23895.5559,
+      "train_tokens_per_second": 10255.735
+    },
+    {
+      "epoch": 0.980211688909342,
+      "grad_norm": 0.7518189729940569,
+      "learning_rate": 5.157198297969572e-06,
+      "loss": 0.4676,
+      "num_input_tokens_seen": 245338832,
+      "step": 4260,
+      "train_runtime": 23930.3402,
+      "train_tokens_per_second": 10252.208
+    },
+    {
+      "epoch": 0.9813621721122872,
+      "grad_norm": 0.7156827103170686,
+      "learning_rate": 5.148166639112799e-06,
+      "loss": 0.4759,
+      "num_input_tokens_seen": 245627464,
+      "step": 4265,
+      "train_runtime": 23952.2735,
+      "train_tokens_per_second": 10254.871
+    },
+    {
+      "epoch": 0.9825126553152324,
+      "grad_norm": 0.7484533196854116,
+      "learning_rate": 5.139134496361588e-06,
+      "loss": 0.4587,
+      "num_input_tokens_seen": 245904296,
+      "step": 4270,
+      "train_runtime": 23985.5908,
+      "train_tokens_per_second": 10252.168
+    },
+    {
+      "epoch": 0.9836631385181777,
+      "grad_norm": 0.8332925906323673,
+      "learning_rate": 5.1301018992138305e-06,
+      "loss": 0.4552,
+      "num_input_tokens_seen": 246167088,
+      "step": 4275,
+      "train_runtime": 24015.0274,
+      "train_tokens_per_second": 10250.544
+    },
+    {
+      "epoch": 0.9848136217211229,
+      "grad_norm": 0.8637612234045138,
+      "learning_rate": 5.121068877168904e-06,
+      "loss": 0.4724,
+      "num_input_tokens_seen": 246435688,
+      "step": 4280,
+      "train_runtime": 24045.1825,
+      "train_tokens_per_second": 10248.859
+    },
+    {
+      "epoch": 0.9859641049240682,
+      "grad_norm": 1.0370668399902798,
+      "learning_rate": 5.112035459727575e-06,
+      "loss": 0.4631,
+      "num_input_tokens_seen": 246703744,
+      "step": 4285,
+      "train_runtime": 24073.5439,
+      "train_tokens_per_second": 10247.92
+    },
+    {
+      "epoch": 0.9871145881270134,
+      "grad_norm": 0.7497822856914327,
+      "learning_rate": 5.1030016763918966e-06,
+      "loss": 0.4762,
+      "num_input_tokens_seen": 246990408,
+      "step": 4290,
+      "train_runtime": 24100.0484,
+      "train_tokens_per_second": 10248.544
+    },
+    {
+      "epoch": 0.9882650713299586,
+      "grad_norm": 0.6975906684954454,
+      "learning_rate": 5.093967556665123e-06,
+      "loss": 0.4931,
+      "num_input_tokens_seen": 247297976,
+      "step": 4295,
+      "train_runtime": 24128.3376,
+      "train_tokens_per_second": 10249.275
+    },
+    {
+      "epoch": 0.9894155545329039,
+      "grad_norm": 0.8384451116322716,
+      "learning_rate": 5.0849331300515995e-06,
+      "loss": 0.4561,
+      "num_input_tokens_seen": 247603272,
+      "step": 4300,
+      "train_runtime": 24154.846,
+      "train_tokens_per_second": 10250.667
+    },
+    {
+      "epoch": 0.9905660377358491,
+      "grad_norm": 0.7810393354954812,
+      "learning_rate": 5.075898426056679e-06,
+      "loss": 0.4486,
+      "num_input_tokens_seen": 247833904,
+      "step": 4305,
+      "train_runtime": 24184.8652,
+      "train_tokens_per_second": 10247.479
+    },
+    {
+      "epoch": 0.9917165209387943,
+      "grad_norm": 0.7271731400638357,
+      "learning_rate": 5.066863474186622e-06,
+      "loss": 0.457,
+      "num_input_tokens_seen": 248114072,
+      "step": 4310,
+      "train_runtime": 24210.8787,
+      "train_tokens_per_second": 10248.041
+    },
+    {
+      "epoch": 0.9928670041417396,
+      "grad_norm": 0.8060132432122766,
+      "learning_rate": 5.057828303948491e-06,
+      "loss": 0.4541,
+      "num_input_tokens_seen": 248398576,
+      "step": 4315,
+      "train_runtime": 24238.7558,
+      "train_tokens_per_second": 10247.992
+    },
+    {
+      "epoch": 0.9940174873446848,
+      "grad_norm": 0.8283092501142488,
+      "learning_rate": 5.048792944850069e-06,
+      "loss": 0.4543,
+      "num_input_tokens_seen": 248704384,
+      "step": 4320,
+      "train_runtime": 24265.285,
+      "train_tokens_per_second": 10249.391
+    },
+    {
+      "epoch": 0.99516797054763,
+      "grad_norm": 0.7259948230634278,
+      "learning_rate": 5.039757426399754e-06,
+      "loss": 0.4397,
+      "num_input_tokens_seen": 248999448,
+      "step": 4325,
+      "train_runtime": 24299.1728,
+      "train_tokens_per_second": 10247.24
+    },
+    {
+      "epoch": 0.9963184537505753,
+      "grad_norm": 0.696448224866946,
+      "learning_rate": 5.030721778106461e-06,
+      "loss": 0.4461,
+      "num_input_tokens_seen": 249293272,
+      "step": 4330,
+      "train_runtime": 24327.6111,
+      "train_tokens_per_second": 10247.339
+    },
+    {
+      "epoch": 0.9974689369535205,
+      "grad_norm": 0.6957114874301872,
+      "learning_rate": 5.021686029479533e-06,
+      "loss": 0.4501,
+      "num_input_tokens_seen": 249589712,
+      "step": 4335,
+      "train_runtime": 24360.6543,
+      "train_tokens_per_second": 10245.608
+    },
+    {
+      "epoch": 0.9986194201564658,
+      "grad_norm": 0.7051099077430307,
+      "learning_rate": 5.012650210028638e-06,
+      "loss": 0.4527,
+      "num_input_tokens_seen": 249896440,
+      "step": 4340,
+      "train_runtime": 24389.5593,
+      "train_tokens_per_second": 10246.042
+    },
+    {
+      "epoch": 0.999769903359411,
+      "grad_norm": 0.7450916210677371,
+      "learning_rate": 5.003614349263678e-06,
+      "loss": 0.4418,
+      "num_input_tokens_seen": 250188152,
+      "step": 4345,
+      "train_runtime": 24421.6513,
+      "train_tokens_per_second": 10244.522
+    },
+    {
+      "epoch": 1.0009203865623562,
+      "grad_norm": 0.8055127975544757,
+      "learning_rate": 4.994578476694687e-06,
+      "loss": 0.4541,
+      "num_input_tokens_seen": 250464168,
+      "step": 4350,
+      "train_runtime": 24457.3998,
+      "train_tokens_per_second": 10240.834
+    },
+    {
+      "epoch": 1.0020708697653014,
+      "grad_norm": 0.8360638972337195,
+      "learning_rate": 4.98554262183174e-06,
+      "loss": 0.42,
+      "num_input_tokens_seen": 250757592,
+      "step": 4355,
+      "train_runtime": 24485.4807,
+      "train_tokens_per_second": 10241.073
+    },
+    {
+      "epoch": 1.0032213529682468,
+      "grad_norm": 0.8069609313015079,
+      "learning_rate": 4.976506814184852e-06,
+      "loss": 0.4127,
+      "num_input_tokens_seen": 251077976,
+      "step": 4360,
+      "train_runtime": 24515.853,
+      "train_tokens_per_second": 10241.454
+    },
+    {
+      "epoch": 1.004371836171192,
+      "grad_norm": 0.7677466303759067,
+      "learning_rate": 4.967471083263886e-06,
+      "loss": 0.4148,
+      "num_input_tokens_seen": 251382480,
+      "step": 4365,
+      "train_runtime": 24545.3054,
+      "train_tokens_per_second": 10241.571
+    },
+    {
+      "epoch": 1.0055223193741372,
+      "grad_norm": 0.7512602113247858,
+      "learning_rate": 4.958435458578454e-06,
+      "loss": 0.4013,
+      "num_input_tokens_seen": 251678568,
+      "step": 4370,
+      "train_runtime": 24575.3081,
+      "train_tokens_per_second": 10241.115
+    },
+    {
+      "epoch": 1.0066728025770824,
+      "grad_norm": 0.8048200245675851,
+      "learning_rate": 4.949399969637819e-06,
+      "loss": 0.4071,
+      "num_input_tokens_seen": 251985672,
+      "step": 4375,
+      "train_runtime": 24610.5056,
+      "train_tokens_per_second": 10238.947
+    },
+    {
+      "epoch": 1.0078232857800276,
+      "grad_norm": 0.7786327819366277,
+      "learning_rate": 4.9403646459508035e-06,
+      "loss": 0.3809,
+      "num_input_tokens_seen": 252273864,
+      "step": 4380,
+      "train_runtime": 24641.5247,
+      "train_tokens_per_second": 10237.754
+    },
+    {
+      "epoch": 1.0089737689829728,
+      "grad_norm": 0.7330610760592257,
+      "learning_rate": 4.931329517025686e-06,
+      "loss": 0.393,
+      "num_input_tokens_seen": 252542976,
+      "step": 4385,
+      "train_runtime": 24665.4255,
+      "train_tokens_per_second": 10238.744
+    },
+    {
+      "epoch": 1.0101242521859182,
+      "grad_norm": 0.8266264718133753,
+      "learning_rate": 4.922294612370113e-06,
+      "loss": 0.3862,
+      "num_input_tokens_seen": 252831296,
+      "step": 4390,
+      "train_runtime": 24693.0405,
+      "train_tokens_per_second": 10238.97
+    },
+    {
+      "epoch": 1.0112747353888634,
+      "grad_norm": 1.2800956096608413,
+      "learning_rate": 4.913259961491001e-06,
+      "loss": 0.4244,
+      "num_input_tokens_seen": 253143064,
+      "step": 4395,
+      "train_runtime": 24719.855,
+      "train_tokens_per_second": 10240.475
+    },
+    {
+      "epoch": 1.0124252185918086,
+      "grad_norm": 0.8801271982814937,
+      "learning_rate": 4.90422559389443e-06,
+      "loss": 0.4154,
+      "num_input_tokens_seen": 253393976,
+      "step": 4400,
+      "train_runtime": 24746.8498,
+      "train_tokens_per_second": 10239.444
+    },
+    {
+      "epoch": 1.0135757017947538,
+      "grad_norm": 0.9381556073890547,
+      "learning_rate": 4.8951915390855606e-06,
+      "loss": 0.414,
+      "num_input_tokens_seen": 253692944,
+      "step": 4405,
+      "train_runtime": 24771.5979,
+      "train_tokens_per_second": 10241.283
+    },
+    {
+      "epoch": 1.014726184997699,
+      "grad_norm": 0.997695188517519,
+      "learning_rate": 4.886157826568529e-06,
+      "loss": 0.3887,
+      "num_input_tokens_seen": 253981184,
+      "step": 4410,
+      "train_runtime": 24801.537,
+      "train_tokens_per_second": 10240.542
+    },
+    {
+      "epoch": 1.0158766682006444,
+      "grad_norm": 0.8061412076274901,
+      "learning_rate": 4.8771244858463545e-06,
+      "loss": 0.3975,
+      "num_input_tokens_seen": 254278912,
+      "step": 4415,
+      "train_runtime": 24824.2385,
+      "train_tokens_per_second": 10243.171
+    },
+    {
+      "epoch": 1.0170271514035896,
+      "grad_norm": 0.8417682229396781,
+      "learning_rate": 4.868091546420845e-06,
+      "loss": 0.4016,
+      "num_input_tokens_seen": 254585296,
+      "step": 4420,
+      "train_runtime": 24849.0752,
+      "train_tokens_per_second": 10245.262
+    },
+    {
+      "epoch": 1.0181776346065348,
+      "grad_norm": 0.821804508066651,
+      "learning_rate": 4.859059037792493e-06,
+      "loss": 0.4174,
+      "num_input_tokens_seen": 254902536,
+      "step": 4425,
+      "train_runtime": 24879.348,
+      "train_tokens_per_second": 10245.547
+    },
+    {
+      "epoch": 1.01932811780948,
+      "grad_norm": 0.8286348594694012,
+      "learning_rate": 4.8500269894603875e-06,
+      "loss": 0.3899,
+      "num_input_tokens_seen": 255181280,
+      "step": 4430,
+      "train_runtime": 24909.6065,
+      "train_tokens_per_second": 10244.292
+    },
+    {
+      "epoch": 1.0204786010124252,
+      "grad_norm": 0.851250427461543,
+      "learning_rate": 4.840995430922111e-06,
+      "loss": 0.3744,
+      "num_input_tokens_seen": 255462176,
+      "step": 4435,
+      "train_runtime": 24935.2501,
+      "train_tokens_per_second": 10245.022
+    },
+    {
+      "epoch": 1.0216290842153704,
+      "grad_norm": 0.9087772659646176,
+      "learning_rate": 4.83196439167365e-06,
+      "loss": 0.3838,
+      "num_input_tokens_seen": 255759336,
+      "step": 4440,
+      "train_runtime": 24959.6567,
+      "train_tokens_per_second": 10246.909
+    },
+    {
+      "epoch": 1.0227795674183158,
+      "grad_norm": 0.8804030110755663,
+      "learning_rate": 4.822933901209291e-06,
+      "loss": 0.3798,
+      "num_input_tokens_seen": 256056312,
+      "step": 4445,
+      "train_runtime": 24988.7102,
+      "train_tokens_per_second": 10246.88
+    },
+    {
+      "epoch": 1.023930050621261,
+      "grad_norm": 0.7568123232516367,
+      "learning_rate": 4.813903989021535e-06,
+      "loss": 0.402,
+      "num_input_tokens_seen": 256354960,
+      "step": 4450,
+      "train_runtime": 25021.1381,
+      "train_tokens_per_second": 10245.536
+    },
+    {
+      "epoch": 1.0250805338242062,
+      "grad_norm": 0.7938955066156531,
+      "learning_rate": 4.804874684600987e-06,
+      "loss": 0.4162,
+      "num_input_tokens_seen": 256629968,
+      "step": 4455,
+      "train_runtime": 25047.1218,
+      "train_tokens_per_second": 10245.887
+    },
+    {
+      "epoch": 1.0262310170271514,
+      "grad_norm": 0.8016985907487973,
+      "learning_rate": 4.795846017436271e-06,
+      "loss": 0.3915,
+      "num_input_tokens_seen": 256930392,
+      "step": 4460,
+      "train_runtime": 25071.4769,
+      "train_tokens_per_second": 10247.916
+    },
+    {
+      "epoch": 1.0273815002300966,
+      "grad_norm": 0.8348867690896439,
+      "learning_rate": 4.786818017013929e-06,
+      "loss": 0.3993,
+      "num_input_tokens_seen": 257212944,
+      "step": 4465,
+      "train_runtime": 25102.334,
+      "train_tokens_per_second": 10246.575
+    },
+    {
+      "epoch": 1.028531983433042,
+      "grad_norm": 0.7347282609656529,
+      "learning_rate": 4.777790712818324e-06,
+      "loss": 0.3961,
+      "num_input_tokens_seen": 257496976,
+      "step": 4470,
+      "train_runtime": 25135.2023,
+      "train_tokens_per_second": 10244.476
+    },
+    {
+      "epoch": 1.0296824666359872,
+      "grad_norm": 0.7533222701246517,
+      "learning_rate": 4.7687641343315496e-06,
+      "loss": 0.4087,
+      "num_input_tokens_seen": 257773536,
+      "step": 4475,
+      "train_runtime": 25161.624,
+      "train_tokens_per_second": 10244.71
+    },
+    {
+      "epoch": 1.0308329498389324,
+      "grad_norm": 0.8931163496314972,
+      "learning_rate": 4.759738311033325e-06,
+      "loss": 0.3977,
+      "num_input_tokens_seen": 258040520,
+      "step": 4480,
+      "train_runtime": 25193.7495,
+      "train_tokens_per_second": 10242.244
+    },
+    {
+      "epoch": 1.0319834330418776,
+      "grad_norm": 0.8198073924045631,
+      "learning_rate": 4.750713272400904e-06,
+      "loss": 0.4236,
+      "num_input_tokens_seen": 258322592,
+      "step": 4485,
+      "train_runtime": 25218.3035,
+      "train_tokens_per_second": 10243.456
+    },
+    {
+      "epoch": 1.0331339162448228,
+      "grad_norm": 0.7397952244814002,
+      "learning_rate": 4.741689047908978e-06,
+      "loss": 0.3762,
+      "num_input_tokens_seen": 258611664,
+      "step": 4490,
+      "train_runtime": 25243.2642,
+      "train_tokens_per_second": 10244.779
+    },
+    {
+      "epoch": 1.0342843994477682,
+      "grad_norm": 0.7232127556508265,
+      "learning_rate": 4.7326656670295815e-06,
+      "loss": 0.3973,
+      "num_input_tokens_seen": 258917144,
+      "step": 4495,
+      "train_runtime": 25274.8383,
+      "train_tokens_per_second": 10244.067
+    },
+    {
+      "epoch": 1.0354348826507134,
+      "grad_norm": 0.7391450819790961,
+      "learning_rate": 4.723643159231988e-06,
+      "loss": 0.4015,
+      "num_input_tokens_seen": 259233016,
+      "step": 4500,
+      "train_runtime": 25309.1597,
+      "train_tokens_per_second": 10242.656
+    },
+    {
+      "epoch": 1.0365853658536586,
+      "grad_norm": 0.8871898782270556,
+      "learning_rate": 4.714621553982629e-06,
+      "loss": 0.3775,
+      "num_input_tokens_seen": 259504904,
+      "step": 4505,
+      "train_runtime": 25331.4949,
+      "train_tokens_per_second": 10244.358
+    },
+    {
+      "epoch": 1.0377358490566038,
+      "grad_norm": 0.849664846993735,
+      "learning_rate": 4.705600880744981e-06,
+      "loss": 0.3856,
+      "num_input_tokens_seen": 259774136,
+      "step": 4510,
+      "train_runtime": 25363.2799,
+      "train_tokens_per_second": 10242.135
+    },
+    {
+      "epoch": 1.038886332259549,
+      "grad_norm": 0.8549631111041464,
+      "learning_rate": 4.696581168979478e-06,
+      "loss": 0.3781,
+      "num_input_tokens_seen": 260046888,
+      "step": 4515,
+      "train_runtime": 25389.3877,
+      "train_tokens_per_second": 10242.346
+    },
+    {
+      "epoch": 1.0400368154624942,
+      "grad_norm": 0.7207327406694509,
+      "learning_rate": 4.687562448143415e-06,
+      "loss": 0.381,
+      "num_input_tokens_seen": 260373720,
+      "step": 4520,
+      "train_runtime": 25415.7044,
+      "train_tokens_per_second": 10244.6
+    },
+    {
+      "epoch": 1.0411872986654396,
+      "grad_norm": 0.6902726653018164,
+      "learning_rate": 4.67854474769085e-06,
+      "loss": 0.3842,
+      "num_input_tokens_seen": 260653912,
+      "step": 4525,
+      "train_runtime": 25450.9883,
+      "train_tokens_per_second": 10241.406
+    },
+    {
+      "epoch": 1.0423377818683848,
+      "grad_norm": 0.7385657772279343,
+      "learning_rate": 4.669528097072511e-06,
+      "loss": 0.3931,
+      "num_input_tokens_seen": 260950792,
+      "step": 4530,
+      "train_runtime": 25489.3248,
+      "train_tokens_per_second": 10237.65
+    },
+    {
+      "epoch": 1.04348826507133,
+      "grad_norm": 0.7946238807267509,
+      "learning_rate": 4.660512525735694e-06,
+      "loss": 0.3949,
+      "num_input_tokens_seen": 261209528,
+      "step": 4535,
+      "train_runtime": 25517.1582,
+      "train_tokens_per_second": 10236.623
+    },
+    {
+      "epoch": 1.0446387482742752,
+      "grad_norm": 0.7358322955924065,
+      "learning_rate": 4.651498063124172e-06,
+      "loss": 0.4049,
+      "num_input_tokens_seen": 261507472,
+      "step": 4540,
+      "train_runtime": 25544.7336,
+      "train_tokens_per_second": 10237.236
+    },
+    {
+      "epoch": 1.0457892314772204,
+      "grad_norm": 0.8480766747587802,
+      "learning_rate": 4.642484738678096e-06,
+      "loss": 0.4178,
+      "num_input_tokens_seen": 261793504,
+      "step": 4545,
+      "train_runtime": 25573.3963,
+      "train_tokens_per_second": 10236.947
+    },
+    {
+      "epoch": 1.0469397146801658,
+      "grad_norm": 0.7467442235664481,
+      "learning_rate": 4.633472581833901e-06,
+      "loss": 0.3963,
+      "num_input_tokens_seen": 262079832,
+      "step": 4550,
+      "train_runtime": 25601.1983,
+      "train_tokens_per_second": 10237.014
+    },
+    {
+      "epoch": 1.048090197883111,
+      "grad_norm": 0.7957985362397747,
+      "learning_rate": 4.624461622024207e-06,
+      "loss": 0.4039,
+      "num_input_tokens_seen": 262366856,
+      "step": 4555,
+      "train_runtime": 25624.3959,
+      "train_tokens_per_second": 10238.948
+    },
+    {
+      "epoch": 1.0492406810860562,
+      "grad_norm": 0.7563492073086401,
+      "learning_rate": 4.615451888677728e-06,
+      "loss": 0.3937,
+      "num_input_tokens_seen": 262639624,
+      "step": 4560,
+      "train_runtime": 25648.9858,
+      "train_tokens_per_second": 10239.766
+    },
+    {
+      "epoch": 1.0503911642890014,
+      "grad_norm": 0.8475546003821913,
+      "learning_rate": 4.6064434112191694e-06,
+      "loss": 0.4136,
+      "num_input_tokens_seen": 262931048,
+      "step": 4565,
+      "train_runtime": 25676.8673,
+      "train_tokens_per_second": 10239.997
+    },
+    {
+      "epoch": 1.0515416474919466,
+      "grad_norm": 0.8676283068302014,
+      "learning_rate": 4.597436219069137e-06,
+      "loss": 0.3886,
+      "num_input_tokens_seen": 263204840,
+      "step": 4570,
+      "train_runtime": 25704.3836,
+      "train_tokens_per_second": 10239.687
+    },
+    {
+      "epoch": 1.0526921306948918,
+      "grad_norm": 0.886125162294371,
+      "learning_rate": 4.588430341644035e-06,
+      "loss": 0.3983,
+      "num_input_tokens_seen": 263475560,
+      "step": 4575,
+      "train_runtime": 25729.8595,
+      "train_tokens_per_second": 10240.07
+    },
+    {
+      "epoch": 1.0538426138978372,
+      "grad_norm": 0.8216160500198864,
+      "learning_rate": 4.579425808355979e-06,
+      "loss": 0.3927,
+      "num_input_tokens_seen": 263747968,
+      "step": 4580,
+      "train_runtime": 25752.7545,
+      "train_tokens_per_second": 10241.544
+    },
+    {
+      "epoch": 1.0549930971007824,
+      "grad_norm": 0.8189744308796768,
+      "learning_rate": 4.570422648612693e-06,
+      "loss": 0.3804,
+      "num_input_tokens_seen": 264031504,
+      "step": 4585,
+      "train_runtime": 25788.2058,
+      "train_tokens_per_second": 10238.46
+    },
+    {
+      "epoch": 1.0561435803037276,
+      "grad_norm": 0.7748374537081644,
+      "learning_rate": 4.561420891817416e-06,
+      "loss": 0.3956,
+      "num_input_tokens_seen": 264322600,
+      "step": 4590,
+      "train_runtime": 25819.279,
+      "train_tokens_per_second": 10237.412
+    },
+    {
+      "epoch": 1.0572940635066728,
+      "grad_norm": 0.8906066722225721,
+      "learning_rate": 4.5524205673688e-06,
+      "loss": 0.3968,
+      "num_input_tokens_seen": 264598680,
+      "step": 4595,
+      "train_runtime": 25849.3274,
+      "train_tokens_per_second": 10236.192
+    },
+    {
+      "epoch": 1.058444546709618,
+      "grad_norm": 0.8025218941904273,
+      "learning_rate": 4.5434217046608275e-06,
+      "loss": 0.385,
+      "num_input_tokens_seen": 264887624,
+      "step": 4600,
+      "train_runtime": 25881.1908,
+      "train_tokens_per_second": 10234.754
+    },
+    {
+      "epoch": 1.0595950299125634,
+      "grad_norm": 0.6938084154351666,
+      "learning_rate": 4.5344243330826994e-06,
+      "loss": 0.395,
+      "num_input_tokens_seen": 265195376,
+      "step": 4605,
+      "train_runtime": 25910.448,
+      "train_tokens_per_second": 10235.075
+    },
+    {
+      "epoch": 1.0607455131155086,
+      "grad_norm": 0.7819006040443388,
+      "learning_rate": 4.52542848201875e-06,
+      "loss": 0.4057,
+      "num_input_tokens_seen": 265487712,
+      "step": 4610,
+      "train_runtime": 25939.4126,
+      "train_tokens_per_second": 10234.916
+    },
+    {
+      "epoch": 1.0618959963184538,
+      "grad_norm": 0.8612574326477116,
+      "learning_rate": 4.5164341808483505e-06,
+      "loss": 0.4023,
+      "num_input_tokens_seen": 265789256,
+      "step": 4615,
+      "train_runtime": 25967.1718,
+      "train_tokens_per_second": 10235.587
+    },
+    {
+      "epoch": 1.063046479521399,
+      "grad_norm": 0.6919141908625002,
+      "learning_rate": 4.507441458945806e-06,
+      "loss": 0.3777,
+      "num_input_tokens_seen": 266088896,
+      "step": 4620,
+      "train_runtime": 25993.5582,
+      "train_tokens_per_second": 10236.725
+    },
+    {
+      "epoch": 1.0641969627243442,
+      "grad_norm": 0.7445554282115643,
+      "learning_rate": 4.4984503456802665e-06,
+      "loss": 0.3914,
+      "num_input_tokens_seen": 266373896,
+      "step": 4625,
+      "train_runtime": 26024.1839,
+      "train_tokens_per_second": 10235.629
+    },
+    {
+      "epoch": 1.0653474459272894,
+      "grad_norm": 0.8312744385437393,
+      "learning_rate": 4.489460870415625e-06,
+      "loss": 0.4093,
+      "num_input_tokens_seen": 266655264,
+      "step": 4630,
+      "train_runtime": 26051.1988,
+      "train_tokens_per_second": 10235.815
+    },
+    {
+      "epoch": 1.0664979291302348,
+      "grad_norm": 0.8610754759668592,
+      "learning_rate": 4.48047306251043e-06,
+      "loss": 0.4192,
+      "num_input_tokens_seen": 266927816,
+      "step": 4635,
+      "train_runtime": 26079.2338,
+      "train_tokens_per_second": 10235.263
+    },
+    {
+      "epoch": 1.06764841233318,
+      "grad_norm": 0.8037235231830722,
+      "learning_rate": 4.471486951317782e-06,
+      "loss": 0.3993,
+      "num_input_tokens_seen": 267230176,
+      "step": 4640,
+      "train_runtime": 26109.8502,
+      "train_tokens_per_second": 10234.841
+    },
+    {
+      "epoch": 1.0687988955361252,
+      "grad_norm": 0.8077749819141142,
+      "learning_rate": 4.462502566185239e-06,
+      "loss": 0.4178,
+      "num_input_tokens_seen": 267473360,
+      "step": 4645,
+      "train_runtime": 26135.1287,
+      "train_tokens_per_second": 10234.247
+    },
+    {
+      "epoch": 1.0699493787390704,
+      "grad_norm": 0.7895292830182086,
+      "learning_rate": 4.453519936454725e-06,
+      "loss": 0.3812,
+      "num_input_tokens_seen": 267735056,
+      "step": 4650,
+      "train_runtime": 26162.3105,
+      "train_tokens_per_second": 10233.617
+    },
+    {
+      "epoch": 1.0710998619420156,
+      "grad_norm": 1.0376875660262543,
+      "learning_rate": 4.444539091462427e-06,
+      "loss": 0.3839,
+      "num_input_tokens_seen": 268026072,
+      "step": 4655,
+      "train_runtime": 26188.5173,
+      "train_tokens_per_second": 10234.488
+    },
+    {
+      "epoch": 1.072250345144961,
+      "grad_norm": 0.8163138338949273,
+      "learning_rate": 4.435560060538704e-06,
+      "loss": 0.393,
+      "num_input_tokens_seen": 268316680,
+      "step": 4660,
+      "train_runtime": 26219.1581,
+      "train_tokens_per_second": 10233.612
+    },
+    {
+      "epoch": 1.0734008283479062,
+      "grad_norm": 0.894305907765372,
+      "learning_rate": 4.426582873007999e-06,
+      "loss": 0.3975,
+      "num_input_tokens_seen": 268591160,
+      "step": 4665,
+      "train_runtime": 26252.4476,
+      "train_tokens_per_second": 10231.09
+    },
+    {
+      "epoch": 1.0745513115508514,
+      "grad_norm": 0.7560110537755685,
+      "learning_rate": 4.41760755818872e-06,
+      "loss": 0.4078,
+      "num_input_tokens_seen": 268891176,
+      "step": 4670,
+      "train_runtime": 26284.2596,
+      "train_tokens_per_second": 10230.122
+    },
+    {
+      "epoch": 1.0757017947537966,
+      "grad_norm": 0.7858358679932324,
+      "learning_rate": 4.408634145393172e-06,
+      "loss": 0.3912,
+      "num_input_tokens_seen": 269171056,
+      "step": 4675,
+      "train_runtime": 26309.3801,
+      "train_tokens_per_second": 10230.992
+    },
+    {
+      "epoch": 1.0768522779567418,
+      "grad_norm": 0.7313484433252957,
+      "learning_rate": 4.3996626639274394e-06,
+      "loss": 0.3935,
+      "num_input_tokens_seen": 269458584,
+      "step": 4680,
+      "train_runtime": 26336.6426,
+      "train_tokens_per_second": 10231.319
+    },
+    {
+      "epoch": 1.078002761159687,
+      "grad_norm": 0.7675656801059764,
+      "learning_rate": 4.3906931430913055e-06,
+      "loss": 0.4002,
+      "num_input_tokens_seen": 269755936,
+      "step": 4685,
+      "train_runtime": 26366.8002,
+      "train_tokens_per_second": 10230.894
+    },
+    {
+      "epoch": 1.0791532443626324,
+      "grad_norm": 0.7365741703952282,
+      "learning_rate": 4.381725612178143e-06,
+      "loss": 0.3961,
+      "num_input_tokens_seen": 270045016,
+      "step": 4690,
+      "train_runtime": 26384.405,
+      "train_tokens_per_second": 10235.024
+    },
+    {
+      "epoch": 1.0803037275655776,
+      "grad_norm": 0.8703104742070946,
+      "learning_rate": 4.372760100474834e-06,
+      "loss": 0.3797,
+      "num_input_tokens_seen": 270283000,
+      "step": 4695,
+      "train_runtime": 26407.6038,
+      "train_tokens_per_second": 10235.044
+    },
+    {
+      "epoch": 1.0814542107685228,
+      "grad_norm": 0.9924205176857956,
+      "learning_rate": 4.363796637261663e-06,
+      "loss": 0.3922,
+      "num_input_tokens_seen": 270581008,
+      "step": 4700,
+      "train_runtime": 26432.6732,
+      "train_tokens_per_second": 10236.612
+    },
+    {
+      "epoch": 1.082604693971468,
+      "grad_norm": 0.8053820309157613,
+      "learning_rate": 4.354835251812221e-06,
+      "loss": 0.3925,
+      "num_input_tokens_seen": 270891304,
+      "step": 4705,
+      "train_runtime": 26456.0478,
+      "train_tokens_per_second": 10239.296
+    },
+    {
+      "epoch": 1.0837551771744132,
+      "grad_norm": 0.7988909829282372,
+      "learning_rate": 4.345875973393318e-06,
+      "loss": 0.4061,
+      "num_input_tokens_seen": 271182216,
+      "step": 4710,
+      "train_runtime": 26480.3774,
+      "train_tokens_per_second": 10240.874
+    },
+    {
+      "epoch": 1.0849056603773586,
+      "grad_norm": 0.7787072605862511,
+      "learning_rate": 4.336918831264878e-06,
+      "loss": 0.3776,
+      "num_input_tokens_seen": 271490744,
+      "step": 4715,
+      "train_runtime": 26509.3806,
+      "train_tokens_per_second": 10241.308
+    },
+    {
+      "epoch": 1.0860561435803038,
+      "grad_norm": 0.7044758615270815,
+      "learning_rate": 4.3279638546798535e-06,
+      "loss": 0.3674,
+      "num_input_tokens_seen": 271775504,
+      "step": 4720,
+      "train_runtime": 26543.997,
+      "train_tokens_per_second": 10238.68
+    },
+    {
+      "epoch": 1.087206626783249,
+      "grad_norm": 0.7984886088111115,
+      "learning_rate": 4.319011072884122e-06,
+      "loss": 0.3801,
+      "num_input_tokens_seen": 272058824,
+      "step": 4725,
+      "train_runtime": 26569.0059,
+      "train_tokens_per_second": 10239.707
+    },
+    {
+      "epoch": 1.0883571099861942,
+      "grad_norm": 0.752472878965637,
+      "learning_rate": 4.310060515116391e-06,
+      "loss": 0.3936,
+      "num_input_tokens_seen": 272324704,
+      "step": 4730,
+      "train_runtime": 26597.7429,
+      "train_tokens_per_second": 10238.64
+    },
+    {
+      "epoch": 1.0895075931891394,
+      "grad_norm": 0.9813717958863895,
+      "learning_rate": 4.301112210608109e-06,
+      "loss": 0.4081,
+      "num_input_tokens_seen": 272610984,
+      "step": 4735,
+      "train_runtime": 26624.9115,
+      "train_tokens_per_second": 10238.944
+    },
+    {
+      "epoch": 1.0906580763920846,
+      "grad_norm": 0.8339285304432086,
+      "learning_rate": 4.29216618858336e-06,
+      "loss": 0.3893,
+      "num_input_tokens_seen": 272905240,
+      "step": 4740,
+      "train_runtime": 26655.4338,
+      "train_tokens_per_second": 10238.259
+    },
+    {
+      "epoch": 1.09180855959503,
+      "grad_norm": 1.0448658937347346,
+      "learning_rate": 4.283222478258777e-06,
+      "loss": 0.4133,
+      "num_input_tokens_seen": 273191968,
+      "step": 4745,
+      "train_runtime": 26685.9071,
+      "train_tokens_per_second": 10237.312
+    },
+    {
+      "epoch": 1.0929590427979752,
+      "grad_norm": 0.8108857640360897,
+      "learning_rate": 4.274281108843448e-06,
+      "loss": 0.3954,
+      "num_input_tokens_seen": 273452936,
+      "step": 4750,
+      "train_runtime": 26716.3409,
+      "train_tokens_per_second": 10235.419
+    },
+    {
+      "epoch": 1.0941095260009204,
+      "grad_norm": 0.7354138879534817,
+      "learning_rate": 4.265342109538808e-06,
+      "loss": 0.3799,
+      "num_input_tokens_seen": 273722672,
+      "step": 4755,
+      "train_runtime": 26750.1977,
+      "train_tokens_per_second": 10232.548
+    },
+    {
+      "epoch": 1.0952600092038656,
+      "grad_norm": 0.8270455630855801,
+      "learning_rate": 4.256405509538553e-06,
+      "loss": 0.3925,
+      "num_input_tokens_seen": 274019720,
+      "step": 4760,
+      "train_runtime": 26776.8505,
+      "train_tokens_per_second": 10233.456
+    },
+    {
+      "epoch": 1.0964104924068108,
+      "grad_norm": 0.7518630368340412,
+      "learning_rate": 4.247471338028547e-06,
+      "loss": 0.3682,
+      "num_input_tokens_seen": 274304824,
+      "step": 4765,
+      "train_runtime": 26802.9117,
+      "train_tokens_per_second": 10234.143
+    },
+    {
+      "epoch": 1.0975609756097562,
+      "grad_norm": 0.7283457862193298,
+      "learning_rate": 4.238539624186717e-06,
+      "loss": 0.4049,
+      "num_input_tokens_seen": 274593232,
+      "step": 4770,
+      "train_runtime": 26824.6085,
+      "train_tokens_per_second": 10236.617
+    },
+    {
+      "epoch": 1.0987114588127014,
+      "grad_norm": 0.8658363685204213,
+      "learning_rate": 4.229610397182973e-06,
+      "loss": 0.3734,
+      "num_input_tokens_seen": 274882608,
+      "step": 4775,
+      "train_runtime": 26852.7231,
+      "train_tokens_per_second": 10236.675
+    },
+    {
+      "epoch": 1.0998619420156466,
+      "grad_norm": 0.7239978180740947,
+      "learning_rate": 4.220683686179094e-06,
+      "loss": 0.3809,
+      "num_input_tokens_seen": 275186728,
+      "step": 4780,
+      "train_runtime": 26877.697,
+      "train_tokens_per_second": 10238.479
+    },
+    {
+      "epoch": 1.1010124252185918,
+      "grad_norm": 0.7731391268336381,
+      "learning_rate": 4.2117595203286435e-06,
+      "loss": 0.396,
+      "num_input_tokens_seen": 275465680,
+      "step": 4785,
+      "train_runtime": 26907.9079,
+      "train_tokens_per_second": 10237.35
+    },
+    {
+      "epoch": 1.102162908421537,
+      "grad_norm": 0.8581659612009893,
+      "learning_rate": 4.2028379287768785e-06,
+      "loss": 0.3969,
+      "num_input_tokens_seen": 275745896,
+      "step": 4790,
+      "train_runtime": 26933.7245,
+      "train_tokens_per_second": 10237.942
+    },
+    {
+      "epoch": 1.1033133916244822,
+      "grad_norm": 0.8641089063339618,
+      "learning_rate": 4.193918940660642e-06,
+      "loss": 0.3906,
+      "num_input_tokens_seen": 276017496,
+      "step": 4795,
+      "train_runtime": 26955.48,
+      "train_tokens_per_second": 10239.754
+    },
+    {
+      "epoch": 1.1044638748274276,
+      "grad_norm": 0.7740870533129622,
+      "learning_rate": 4.185002585108277e-06,
+      "loss": 0.3923,
+      "num_input_tokens_seen": 276299528,
+      "step": 4800,
+      "train_runtime": 26982.3584,
+      "train_tokens_per_second": 10240.007
+    },
+    {
+      "epoch": 1.1056143580303728,
+      "grad_norm": 0.71162626374314,
+      "learning_rate": 4.1760888912395335e-06,
+      "loss": 0.3985,
+      "num_input_tokens_seen": 276587504,
+      "step": 4805,
+      "train_runtime": 27007.5524,
+      "train_tokens_per_second": 10241.117
+    },
+    {
+      "epoch": 1.106764841233318,
+      "grad_norm": 0.7208187775593198,
+      "learning_rate": 4.16717788816546e-06,
+      "loss": 0.4109,
+      "num_input_tokens_seen": 276890256,
+      "step": 4810,
+      "train_runtime": 27033.1843,
+      "train_tokens_per_second": 10242.606
+    },
+    {
+      "epoch": 1.1079153244362632,
+      "grad_norm": 0.843109727472828,
+      "learning_rate": 4.158269604988326e-06,
+      "loss": 0.374,
+      "num_input_tokens_seen": 277173960,
+      "step": 4815,
+      "train_runtime": 27063.1927,
+      "train_tokens_per_second": 10241.732
+    },
+    {
+      "epoch": 1.1090658076392084,
+      "grad_norm": 0.9870944975730508,
+      "learning_rate": 4.149364070801509e-06,
+      "loss": 0.3871,
+      "num_input_tokens_seen": 277460408,
+      "step": 4820,
+      "train_runtime": 27093.159,
+      "train_tokens_per_second": 10240.977
+    },
+    {
+      "epoch": 1.1102162908421538,
+      "grad_norm": 0.8934851335460822,
+      "learning_rate": 4.140461314689415e-06,
+      "loss": 0.3825,
+      "num_input_tokens_seen": 277739016,
+      "step": 4825,
+      "train_runtime": 27116.5339,
+      "train_tokens_per_second": 10242.423
+    },
+    {
+      "epoch": 1.111366774045099,
+      "grad_norm": 0.8347095019102736,
+      "learning_rate": 4.131561365727379e-06,
+      "loss": 0.3821,
+      "num_input_tokens_seen": 277999080,
+      "step": 4830,
+      "train_runtime": 27144.5795,
+      "train_tokens_per_second": 10241.421
+    },
+    {
+      "epoch": 1.1125172572480442,
+      "grad_norm": 0.727926651750452,
+      "learning_rate": 4.122664252981561e-06,
+      "loss": 0.375,
+      "num_input_tokens_seen": 278292608,
+      "step": 4835,
+      "train_runtime": 27181.4647,
+      "train_tokens_per_second": 10238.323
+    },
+    {
+      "epoch": 1.1136677404509894,
+      "grad_norm": 0.7729482124149348,
+      "learning_rate": 4.113770005508864e-06,
+      "loss": 0.4011,
+      "num_input_tokens_seen": 278611152,
+      "step": 4840,
+      "train_runtime": 27206.5055,
+      "train_tokens_per_second": 10240.608
+    },
+    {
+      "epoch": 1.1148182236539346,
+      "grad_norm": 0.7951179787139396,
+      "learning_rate": 4.10487865235683e-06,
+      "loss": 0.3792,
+      "num_input_tokens_seen": 278901144,
+      "step": 4845,
+      "train_runtime": 27237.6328,
+      "train_tokens_per_second": 10239.552
+    },
+    {
+      "epoch": 1.11596870685688,
+      "grad_norm": 0.8043174643236001,
+      "learning_rate": 4.095990222563549e-06,
+      "loss": 0.4087,
+      "num_input_tokens_seen": 279197064,
+      "step": 4850,
+      "train_runtime": 27268.9332,
+      "train_tokens_per_second": 10238.65
+    },
+    {
+      "epoch": 1.1171191900598252,
+      "grad_norm": 0.8804410241168195,
+      "learning_rate": 4.087104745157569e-06,
+      "loss": 0.3765,
+      "num_input_tokens_seen": 279455232,
+      "step": 4855,
+      "train_runtime": 27298.4187,
+      "train_tokens_per_second": 10237.048
+    },
+    {
+      "epoch": 1.1182696732627704,
+      "grad_norm": 0.8519129971987875,
+      "learning_rate": 4.078222249157789e-06,
+      "loss": 0.4133,
+      "num_input_tokens_seen": 279713104,
+      "step": 4860,
+      "train_runtime": 27324.6399,
+      "train_tokens_per_second": 10236.662
+    },
+    {
+      "epoch": 1.1194201564657156,
+      "grad_norm": 0.9663699324062075,
+      "learning_rate": 4.069342763573371e-06,
+      "loss": 0.387,
+      "num_input_tokens_seen": 280002856,
+      "step": 4865,
+      "train_runtime": 27358.6696,
+      "train_tokens_per_second": 10234.52
+    },
+    {
+      "epoch": 1.1205706396686608,
+      "grad_norm": 0.7813297201978426,
+      "learning_rate": 4.060466317403652e-06,
+      "loss": 0.3947,
+      "num_input_tokens_seen": 280303648,
+      "step": 4870,
+      "train_runtime": 27389.2169,
+      "train_tokens_per_second": 10234.088
+    },
+    {
+      "epoch": 1.1217211228716062,
+      "grad_norm": 0.81213676704039,
+      "learning_rate": 4.051592939638036e-06,
+      "loss": 0.4095,
+      "num_input_tokens_seen": 280594488,
+      "step": 4875,
+      "train_runtime": 27419.0453,
+      "train_tokens_per_second": 10233.562
+    },
+    {
+      "epoch": 1.1228716060745514,
+      "grad_norm": 0.7768259799425594,
+      "learning_rate": 4.042722659255907e-06,
+      "loss": 0.4115,
+      "num_input_tokens_seen": 280911808,
+      "step": 4880,
+      "train_runtime": 27443.0552,
+      "train_tokens_per_second": 10236.171
+    },
+    {
+      "epoch": 1.1240220892774966,
+      "grad_norm": 0.7942968659348584,
+      "learning_rate": 4.033855505226539e-06,
+      "loss": 0.391,
+      "num_input_tokens_seen": 281210768,
+      "step": 4885,
+      "train_runtime": 27471.9468,
+      "train_tokens_per_second": 10236.288
+    },
+    {
+      "epoch": 1.1251725724804418,
+      "grad_norm": 0.8259906678557996,
+      "learning_rate": 4.024991506508989e-06,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 281507248,
+      "step": 4890,
+      "train_runtime": 27502.8751,
+      "train_tokens_per_second": 10235.557
+    },
+    {
+      "epoch": 1.126323055683387,
+      "grad_norm": 0.8939134889708473,
+      "learning_rate": 4.016130692052012e-06,
+      "loss": 0.3971,
+      "num_input_tokens_seen": 281757176,
+      "step": 4895,
+      "train_runtime": 27535.3033,
+      "train_tokens_per_second": 10232.579
+    },
+    {
+      "epoch": 1.1274735388863322,
+      "grad_norm": 0.8739857668646632,
+      "learning_rate": 4.007273090793963e-06,
+      "loss": 0.4075,
+      "num_input_tokens_seen": 282036840,
+      "step": 4900,
+      "train_runtime": 27560.6639,
+      "train_tokens_per_second": 10233.311
+    },
+    {
+      "epoch": 1.1286240220892776,
+      "grad_norm": 0.8708490839964679,
+      "learning_rate": 3.9984187316627025e-06,
+      "loss": 0.3853,
+      "num_input_tokens_seen": 282311048,
+      "step": 4905,
+      "train_runtime": 27585.9588,
+      "train_tokens_per_second": 10233.868
+    },
+    {
+      "epoch": 1.1297745052922228,
+      "grad_norm": 1.0751864009990957,
+      "learning_rate": 3.989567643575506e-06,
+      "loss": 0.3912,
+      "num_input_tokens_seen": 282613264,
+      "step": 4910,
+      "train_runtime": 27612.275,
+      "train_tokens_per_second": 10235.059
+    },
+    {
+      "epoch": 1.130924988495168,
+      "grad_norm": 0.8375121058001744,
+      "learning_rate": 3.98071985543896e-06,
+      "loss": 0.4117,
+      "num_input_tokens_seen": 282905992,
+      "step": 4915,
+      "train_runtime": 27642.6416,
+      "train_tokens_per_second": 10234.405
+    },
+    {
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.9910229720187773,
+      "learning_rate": 3.971875396148882e-06,
+      "loss": 0.408,
+      "num_input_tokens_seen": 283164728,
+      "step": 4920,
+      "train_runtime": 27663.5237,
+      "train_tokens_per_second": 10236.033
+    },
+    {
+      "epoch": 1.1332259549010584,
+      "grad_norm": 0.8237758132381411,
+      "learning_rate": 3.9630342945902084e-06,
+      "loss": 0.4146,
+      "num_input_tokens_seen": 283468144,
+      "step": 4925,
+      "train_runtime": 27685.8892,
+      "train_tokens_per_second": 10238.723
+    },
+    {
+      "epoch": 1.1343764381040038,
+      "grad_norm": 0.7646666154561143,
+      "learning_rate": 3.954196579636918e-06,
+      "loss": 0.3909,
+      "num_input_tokens_seen": 283758176,
+      "step": 4930,
+      "train_runtime": 27713.9652,
+      "train_tokens_per_second": 10238.815
+    },
+    {
+      "epoch": 1.135526921306949,
+      "grad_norm": 0.8381063850440155,
+      "learning_rate": 3.945362280151922e-06,
+      "loss": 0.3985,
+      "num_input_tokens_seen": 284028360,
+      "step": 4935,
+      "train_runtime": 27745.4623,
+      "train_tokens_per_second": 10236.93
+    },
+    {
+      "epoch": 1.1366774045098942,
+      "grad_norm": 0.680356894367383,
+      "learning_rate": 3.936531424986984e-06,
+      "loss": 0.3947,
+      "num_input_tokens_seen": 284335112,
+      "step": 4940,
+      "train_runtime": 27769.6801,
+      "train_tokens_per_second": 10239.049
+    },
+    {
+      "epoch": 1.1378278877128394,
+      "grad_norm": 0.7962485701227486,
+      "learning_rate": 3.927704042982616e-06,
+      "loss": 0.417,
+      "num_input_tokens_seen": 284630192,
+      "step": 4945,
+      "train_runtime": 27796.1942,
+      "train_tokens_per_second": 10239.898
+    },
+    {
+      "epoch": 1.1389783709157846,
+      "grad_norm": 0.7048255521513512,
+      "learning_rate": 3.918880162967985e-06,
+      "loss": 0.3873,
+      "num_input_tokens_seen": 284923496,
+      "step": 4950,
+      "train_runtime": 27826.4243,
+      "train_tokens_per_second": 10239.314
+    },
+    {
+      "epoch": 1.1401288541187298,
+      "grad_norm": 0.8119616571046254,
+      "learning_rate": 3.910059813760825e-06,
+      "loss": 0.4015,
+      "num_input_tokens_seen": 285212160,
+      "step": 4955,
+      "train_runtime": 27849.2243,
+      "train_tokens_per_second": 10241.296
+    },
+    {
+      "epoch": 1.1412793373216752,
+      "grad_norm": 0.7400842358997406,
+      "learning_rate": 3.9012430241673324e-06,
+      "loss": 0.4013,
+      "num_input_tokens_seen": 285517528,
+      "step": 4960,
+      "train_runtime": 27873.513,
+      "train_tokens_per_second": 10243.328
+    },
+    {
+      "epoch": 1.1424298205246204,
+      "grad_norm": 0.8050097936390155,
+      "learning_rate": 3.892429822982086e-06,
+      "loss": 0.3662,
+      "num_input_tokens_seen": 285802072,
+      "step": 4965,
+      "train_runtime": 27901.4084,
+      "train_tokens_per_second": 10243.285
+    },
+    {
+      "epoch": 1.1435803037275656,
+      "grad_norm": 0.7341591841110063,
+      "learning_rate": 3.883620238987942e-06,
+      "loss": 0.4052,
+      "num_input_tokens_seen": 286089984,
+      "step": 4970,
+      "train_runtime": 27927.8645,
+      "train_tokens_per_second": 10243.89
+    },
+    {
+      "epoch": 1.1447307869305108,
+      "grad_norm": 0.7321617663202142,
+      "learning_rate": 3.874814300955942e-06,
+      "loss": 0.4045,
+      "num_input_tokens_seen": 286397184,
+      "step": 4975,
+      "train_runtime": 27957.6106,
+      "train_tokens_per_second": 10243.979
+    },
+    {
+      "epoch": 1.145881270133456,
+      "grad_norm": 0.7409584935904456,
+      "learning_rate": 3.866012037645222e-06,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 286673088,
+      "step": 4980,
+      "train_runtime": 27988.7889,
+      "train_tokens_per_second": 10242.426
+    },
+    {
+      "epoch": 1.1470317533364014,
+      "grad_norm": 0.917159703938577,
+      "learning_rate": 3.857213477802915e-06,
+      "loss": 0.3746,
+      "num_input_tokens_seen": 286947440,
+      "step": 4985,
+      "train_runtime": 28016.6857,
+      "train_tokens_per_second": 10242.019
+    },
+    {
+      "epoch": 1.1481822365393466,
+      "grad_norm": 0.7919422807273204,
+      "learning_rate": 3.848418650164059e-06,
+      "loss": 0.4058,
+      "num_input_tokens_seen": 287251096,
+      "step": 4990,
+      "train_runtime": 28043.3533,
+      "train_tokens_per_second": 10243.108
+    },
+    {
+      "epoch": 1.1493327197422918,
+      "grad_norm": 0.6929598783193462,
+      "learning_rate": 3.839627583451509e-06,
+      "loss": 0.3829,
+      "num_input_tokens_seen": 287532168,
+      "step": 4995,
+      "train_runtime": 28074.6917,
+      "train_tokens_per_second": 10241.686
+    },
+    {
+      "epoch": 1.150483202945237,
+      "grad_norm": 0.814785949219345,
+      "learning_rate": 3.830840306375828e-06,
+      "loss": 0.3912,
+      "num_input_tokens_seen": 287799552,
+      "step": 5000,
+      "train_runtime": 28104.9246,
+      "train_tokens_per_second": 10240.182
+    },
+    {
+      "epoch": 1.1516336861481822,
+      "grad_norm": 0.7256587435508985,
+      "learning_rate": 3.8220568476352096e-06,
+      "loss": 0.4177,
+      "num_input_tokens_seen": 288102200,
+      "step": 5005,
+      "train_runtime": 28136.1442,
+      "train_tokens_per_second": 10239.576
+    },
+    {
+      "epoch": 1.1527841693511274,
+      "grad_norm": 0.756522381079062,
+      "learning_rate": 3.813277235915372e-06,
+      "loss": 0.3874,
+      "num_input_tokens_seen": 288373216,
+      "step": 5010,
+      "train_runtime": 28160.6763,
+      "train_tokens_per_second": 10240.28
+    },
+    {
+      "epoch": 1.1539346525540728,
+      "grad_norm": 0.8244595515540546,
+      "learning_rate": 3.8045014998894725e-06,
+      "loss": 0.3936,
+      "num_input_tokens_seen": 288684096,
+      "step": 5015,
+      "train_runtime": 28188.1636,
+      "train_tokens_per_second": 10241.323
+    },
+    {
+      "epoch": 1.155085135757018,
+      "grad_norm": 0.7424530519613302,
+      "learning_rate": 3.7957296682180136e-06,
+      "loss": 0.3769,
+      "num_input_tokens_seen": 288965032,
+      "step": 5020,
+      "train_runtime": 28216.2684,
+      "train_tokens_per_second": 10241.079
+    },
+    {
+      "epoch": 1.1562356189599632,
+      "grad_norm": 0.7325580868284697,
+      "learning_rate": 3.7869617695487393e-06,
+      "loss": 0.3771,
+      "num_input_tokens_seen": 289257344,
+      "step": 5025,
+      "train_runtime": 28244.4661,
+      "train_tokens_per_second": 10241.204
+    },
+    {
+      "epoch": 1.1573861021629084,
+      "grad_norm": 0.8851470092199445,
+      "learning_rate": 3.778197832516557e-06,
+      "loss": 0.3904,
+      "num_input_tokens_seen": 289518240,
+      "step": 5030,
+      "train_runtime": 28264.3099,
+      "train_tokens_per_second": 10243.245
+    },
+    {
+      "epoch": 1.1585365853658536,
+      "grad_norm": 0.7194549766325347,
+      "learning_rate": 3.7694378857434276e-06,
+      "loss": 0.3865,
+      "num_input_tokens_seen": 289809136,
+      "step": 5035,
+      "train_runtime": 28297.4389,
+      "train_tokens_per_second": 10241.532
+    },
+    {
+      "epoch": 1.159687068568799,
+      "grad_norm": 0.6941467202843258,
+      "learning_rate": 3.7606819578382864e-06,
+      "loss": 0.3878,
+      "num_input_tokens_seen": 290081720,
+      "step": 5040,
+      "train_runtime": 28319.8645,
+      "train_tokens_per_second": 10243.048
+    },
+    {
+      "epoch": 1.1608375517717442,
+      "grad_norm": 0.740018325919588,
+      "learning_rate": 3.751930077396943e-06,
+      "loss": 0.3935,
+      "num_input_tokens_seen": 290374544,
+      "step": 5045,
+      "train_runtime": 28347.4469,
+      "train_tokens_per_second": 10243.411
+    },
+    {
+      "epoch": 1.1619880349746894,
+      "grad_norm": 0.7585687529094086,
+      "learning_rate": 3.7431822730019875e-06,
+      "loss": 0.4015,
+      "num_input_tokens_seen": 290662504,
+      "step": 5050,
+      "train_runtime": 28372.5969,
+      "train_tokens_per_second": 10244.48
+    },
+    {
+      "epoch": 1.1631385181776346,
+      "grad_norm": 0.7630828619964979,
+      "learning_rate": 3.734438573222695e-06,
+      "loss": 0.3941,
+      "num_input_tokens_seen": 290947424,
+      "step": 5055,
+      "train_runtime": 28400.1189,
+      "train_tokens_per_second": 10244.585
+    },
+    {
+      "epoch": 1.1642890013805798,
+      "grad_norm": 0.8676732107406873,
+      "learning_rate": 3.725699006614941e-06,
+      "loss": 0.421,
+      "num_input_tokens_seen": 291228384,
+      "step": 5060,
+      "train_runtime": 28424.7737,
+      "train_tokens_per_second": 10245.583
+    },
+    {
+      "epoch": 1.165439484583525,
+      "grad_norm": 0.6825541563812574,
+      "learning_rate": 3.7169636017210975e-06,
+      "loss": 0.4086,
+      "num_input_tokens_seen": 291536824,
+      "step": 5065,
+      "train_runtime": 28452.5323,
+      "train_tokens_per_second": 10246.428
+    },
+    {
+      "epoch": 1.1665899677864704,
+      "grad_norm": 0.8986331937646597,
+      "learning_rate": 3.7082323870699473e-06,
+      "loss": 0.4101,
+      "num_input_tokens_seen": 291836104,
+      "step": 5070,
+      "train_runtime": 28479.4234,
+      "train_tokens_per_second": 10247.262
+    },
+    {
+      "epoch": 1.1677404509894156,
+      "grad_norm": 0.7657401595790261,
+      "learning_rate": 3.699505391176591e-06,
+      "loss": 0.378,
+      "num_input_tokens_seen": 292134200,
+      "step": 5075,
+      "train_runtime": 28514.1213,
+      "train_tokens_per_second": 10245.246
+    },
+    {
+      "epoch": 1.1688909341923608,
+      "grad_norm": 0.799182640935118,
+      "learning_rate": 3.6907826425423465e-06,
+      "loss": 0.4149,
+      "num_input_tokens_seen": 292428880,
+      "step": 5080,
+      "train_runtime": 28547.6556,
+      "train_tokens_per_second": 10243.534
+    },
+    {
+      "epoch": 1.170041417395306,
+      "grad_norm": 0.745442329560717,
+      "learning_rate": 3.682064169654663e-06,
+      "loss": 0.4037,
+      "num_input_tokens_seen": 292736640,
+      "step": 5085,
+      "train_runtime": 28574.5413,
+      "train_tokens_per_second": 10244.666
+    },
+    {
+      "epoch": 1.1711919005982512,
+      "grad_norm": 0.7706662280029165,
+      "learning_rate": 3.673350000987024e-06,
+      "loss": 0.3999,
+      "num_input_tokens_seen": 293012632,
+      "step": 5090,
+      "train_runtime": 28605.2305,
+      "train_tokens_per_second": 10243.324
+    },
+    {
+      "epoch": 1.1723423838011966,
+      "grad_norm": 0.7508231867795107,
+      "learning_rate": 3.6646401649988573e-06,
+      "loss": 0.3842,
+      "num_input_tokens_seen": 293311912,
+      "step": 5095,
+      "train_runtime": 28634.8222,
+      "train_tokens_per_second": 10243.19
+    },
+    {
+      "epoch": 1.1734928670041418,
+      "grad_norm": 0.7737491080166808,
+      "learning_rate": 3.6559346901354442e-06,
+      "loss": 0.4018,
+      "num_input_tokens_seen": 293600008,
+      "step": 5100,
+      "train_runtime": 28664.5293,
+      "train_tokens_per_second": 10242.624
+    },
+    {
+      "epoch": 1.174643350207087,
+      "grad_norm": 0.6860305896787925,
+      "learning_rate": 3.6472336048278145e-06,
+      "loss": 0.37,
+      "num_input_tokens_seen": 293875248,
+      "step": 5105,
+      "train_runtime": 28698.5436,
+      "train_tokens_per_second": 10240.075
+    },
+    {
+      "epoch": 1.1757938334100322,
+      "grad_norm": 0.7430334188966102,
+      "learning_rate": 3.6385369374926717e-06,
+      "loss": 0.3946,
+      "num_input_tokens_seen": 294178016,
+      "step": 5110,
+      "train_runtime": 28728.2934,
+      "train_tokens_per_second": 10240.01
+    },
+    {
+      "epoch": 1.1769443166129774,
+      "grad_norm": 0.8330702115511835,
+      "learning_rate": 3.6298447165322827e-06,
+      "loss": 0.415,
+      "num_input_tokens_seen": 294449008,
+      "step": 5115,
+      "train_runtime": 28754.6243,
+      "train_tokens_per_second": 10240.058
+    },
+    {
+      "epoch": 1.1780947998159226,
+      "grad_norm": 0.8234297390231555,
+      "learning_rate": 3.6211569703343986e-06,
+      "loss": 0.4,
+      "num_input_tokens_seen": 294716320,
+      "step": 5120,
+      "train_runtime": 28783.8012,
+      "train_tokens_per_second": 10238.965
+    },
+    {
+      "epoch": 1.179245283018868,
+      "grad_norm": 0.8253445003430321,
+      "learning_rate": 3.6124737272721515e-06,
+      "loss": 0.3725,
+      "num_input_tokens_seen": 294979528,
+      "step": 5125,
+      "train_runtime": 28818.0968,
+      "train_tokens_per_second": 10235.913
+    },
+    {
+      "epoch": 1.1803957662218132,
+      "grad_norm": 0.8104783230609601,
+      "learning_rate": 3.6037950157039737e-06,
+      "loss": 0.3865,
+      "num_input_tokens_seen": 295263832,
+      "step": 5130,
+      "train_runtime": 28846.4732,
+      "train_tokens_per_second": 10235.7
+    },
+    {
+      "epoch": 1.1815462494247584,
+      "grad_norm": 0.8201722059697558,
+      "learning_rate": 3.5951208639734938e-06,
+      "loss": 0.3797,
+      "num_input_tokens_seen": 295537896,
+      "step": 5135,
+      "train_runtime": 28878.1075,
+      "train_tokens_per_second": 10233.977
+    },
+    {
+      "epoch": 1.1826967326277036,
+      "grad_norm": 0.8306811173379203,
+      "learning_rate": 3.586451300409446e-06,
+      "loss": 0.4089,
+      "num_input_tokens_seen": 295829344,
+      "step": 5140,
+      "train_runtime": 28908.0688,
+      "train_tokens_per_second": 10233.452
+    },
+    {
+      "epoch": 1.1838472158306488,
+      "grad_norm": 0.8365689252455771,
+      "learning_rate": 3.577786353325586e-06,
+      "loss": 0.4091,
+      "num_input_tokens_seen": 296113840,
+      "step": 5145,
+      "train_runtime": 28936.1453,
+      "train_tokens_per_second": 10233.355
+    },
+    {
+      "epoch": 1.1849976990335942,
+      "grad_norm": 0.809733932127399,
+      "learning_rate": 3.569126051020587e-06,
+      "loss": 0.4123,
+      "num_input_tokens_seen": 296414696,
+      "step": 5150,
+      "train_runtime": 28966.5523,
+      "train_tokens_per_second": 10232.999
+    },
+    {
+      "epoch": 1.1861481822365394,
+      "grad_norm": 0.7692030069854735,
+      "learning_rate": 3.5604704217779585e-06,
+      "loss": 0.3896,
+      "num_input_tokens_seen": 296700816,
+      "step": 5155,
+      "train_runtime": 28986.16,
+      "train_tokens_per_second": 10235.948
+    },
+    {
+      "epoch": 1.1872986654394846,
+      "grad_norm": 0.8072562530145048,
+      "learning_rate": 3.5518194938659446e-06,
+      "loss": 0.3825,
+      "num_input_tokens_seen": 296983720,
+      "step": 5160,
+      "train_runtime": 29011.3649,
+      "train_tokens_per_second": 10236.806
+    },
+    {
+      "epoch": 1.1884491486424298,
+      "grad_norm": 0.766229463281634,
+      "learning_rate": 3.5431732955374352e-06,
+      "loss": 0.3993,
+      "num_input_tokens_seen": 297293560,
+      "step": 5165,
+      "train_runtime": 29042.3512,
+      "train_tokens_per_second": 10236.553
+    },
+    {
+      "epoch": 1.189599631845375,
+      "grad_norm": 0.7554767445688814,
+      "learning_rate": 3.534531855029877e-06,
+      "loss": 0.3766,
+      "num_input_tokens_seen": 297543744,
+      "step": 5170,
+      "train_runtime": 29068.9327,
+      "train_tokens_per_second": 10235.799
+    },
+    {
+      "epoch": 1.1907501150483202,
+      "grad_norm": 0.7970734721925233,
+      "learning_rate": 3.525895200565173e-06,
+      "loss": 0.3901,
+      "num_input_tokens_seen": 297815640,
+      "step": 5175,
+      "train_runtime": 29095.3609,
+      "train_tokens_per_second": 10235.846
+    },
+    {
+      "epoch": 1.1919005982512656,
+      "grad_norm": 0.813572380108505,
+      "learning_rate": 3.5172633603496004e-06,
+      "loss": 0.3785,
+      "num_input_tokens_seen": 298125712,
+      "step": 5180,
+      "train_runtime": 29129.5392,
+      "train_tokens_per_second": 10234.481
+    },
+    {
+      "epoch": 1.1930510814542108,
+      "grad_norm": 0.7478498833957793,
+      "learning_rate": 3.508636362573713e-06,
+      "loss": 0.3731,
+      "num_input_tokens_seen": 298414712,
+      "step": 5185,
+      "train_runtime": 29155.8287,
+      "train_tokens_per_second": 10235.165
+    },
+    {
+      "epoch": 1.194201564657156,
+      "grad_norm": 0.7822328792284898,
+      "learning_rate": 3.5000142354122497e-06,
+      "loss": 0.3901,
+      "num_input_tokens_seen": 298695296,
+      "step": 5190,
+      "train_runtime": 29184.3456,
+      "train_tokens_per_second": 10234.778
+    },
+    {
+      "epoch": 1.1953520478601012,
+      "grad_norm": 0.7647627760227742,
+      "learning_rate": 3.4913970070240388e-06,
+      "loss": 0.3778,
+      "num_input_tokens_seen": 298990656,
+      "step": 5195,
+      "train_runtime": 29217.0901,
+      "train_tokens_per_second": 10233.417
+    },
+    {
+      "epoch": 1.1965025310630466,
+      "grad_norm": 0.858569092273545,
+      "learning_rate": 3.4827847055519155e-06,
+      "loss": 0.3699,
+      "num_input_tokens_seen": 299245600,
+      "step": 5200,
+      "train_runtime": 29246.0617,
+      "train_tokens_per_second": 10231.996
+    },
+    {
+      "epoch": 1.1976530142659918,
+      "grad_norm": 0.866864391199339,
+      "learning_rate": 3.474177359122618e-06,
+      "loss": 0.3858,
+      "num_input_tokens_seen": 299519088,
+      "step": 5205,
+      "train_runtime": 29272.4924,
+      "train_tokens_per_second": 10232.101
+    },
+    {
+      "epoch": 1.198803497468937,
+      "grad_norm": 0.7732539013942521,
+      "learning_rate": 3.4655749958467085e-06,
+      "loss": 0.3974,
+      "num_input_tokens_seen": 299819640,
+      "step": 5210,
+      "train_runtime": 29293.1883,
+      "train_tokens_per_second": 10235.132
+    },
+    {
+      "epoch": 1.1999539806718822,
+      "grad_norm": 0.8044138306522474,
+      "learning_rate": 3.456977643818471e-06,
+      "loss": 0.393,
+      "num_input_tokens_seen": 300124536,
+      "step": 5215,
+      "train_runtime": 29323.5051,
+      "train_tokens_per_second": 10234.948
+    },
+    {
+      "epoch": 1.2011044638748274,
+      "grad_norm": 1.5564147439281406,
+      "learning_rate": 3.4483853311158234e-06,
+      "loss": 0.3954,
+      "num_input_tokens_seen": 300418848,
+      "step": 5220,
+      "train_runtime": 29352.044,
+      "train_tokens_per_second": 10235.023
+    },
+    {
+      "epoch": 1.2022549470777726,
+      "grad_norm": 0.6884871290161069,
+      "learning_rate": 3.439798085800228e-06,
+      "loss": 0.3978,
+      "num_input_tokens_seen": 300702680,
+      "step": 5225,
+      "train_runtime": 29376.1382,
+      "train_tokens_per_second": 10236.29
+    },
+    {
+      "epoch": 1.2034054302807178,
+      "grad_norm": 0.8436686526479331,
+      "learning_rate": 3.4312159359165923e-06,
+      "loss": 0.3941,
+      "num_input_tokens_seen": 300979872,
+      "step": 5230,
+      "train_runtime": 29402.7752,
+      "train_tokens_per_second": 10236.444
+    },
+    {
+      "epoch": 1.2045559134836632,
+      "grad_norm": 0.7897505599307836,
+      "learning_rate": 3.4226389094931905e-06,
+      "loss": 0.3892,
+      "num_input_tokens_seen": 301262224,
+      "step": 5235,
+      "train_runtime": 29438.8125,
+      "train_tokens_per_second": 10233.505
+    },
+    {
+      "epoch": 1.2057063966866084,
+      "grad_norm": 0.825576333719414,
+      "learning_rate": 3.4140670345415595e-06,
+      "loss": 0.4195,
+      "num_input_tokens_seen": 301567840,
+      "step": 5240,
+      "train_runtime": 29469.8325,
+      "train_tokens_per_second": 10233.103
+    },
+    {
+      "epoch": 1.2068568798895536,
+      "grad_norm": 0.8134862308980768,
+      "learning_rate": 3.405500339056411e-06,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 301825496,
+      "step": 5245,
+      "train_runtime": 29496.2405,
+      "train_tokens_per_second": 10232.677
+    },
+    {
+      "epoch": 1.2080073630924988,
+      "grad_norm": 0.7794146983439229,
+      "learning_rate": 3.396938851015546e-06,
+      "loss": 0.3895,
+      "num_input_tokens_seen": 302110360,
+      "step": 5250,
+      "train_runtime": 29522.6637,
+      "train_tokens_per_second": 10233.167
+    },
+    {
+      "epoch": 1.2091578462954442,
+      "grad_norm": 0.8584854730325425,
+      "learning_rate": 3.3883825983797518e-06,
+      "loss": 0.3869,
+      "num_input_tokens_seen": 302394320,
+      "step": 5255,
+      "train_runtime": 29556.6821,
+      "train_tokens_per_second": 10230.997
+    },
+    {
+      "epoch": 1.2103083294983894,
+      "grad_norm": 0.7195431986397431,
+      "learning_rate": 3.3798316090927226e-06,
+      "loss": 0.3668,
+      "num_input_tokens_seen": 302707320,
+      "step": 5260,
+      "train_runtime": 29586.8412,
+      "train_tokens_per_second": 10231.147
+    },
+    {
+      "epoch": 1.2114588127013346,
+      "grad_norm": 0.952377460801908,
+      "learning_rate": 3.371285911080964e-06,
+      "loss": 0.4103,
+      "num_input_tokens_seen": 302977904,
+      "step": 5265,
+      "train_runtime": 29612.3293,
+      "train_tokens_per_second": 10231.478
+    },
+    {
+      "epoch": 1.2126092959042798,
+      "grad_norm": 1.0392046702202777,
+      "learning_rate": 3.3627455322536965e-06,
+      "loss": 0.399,
+      "num_input_tokens_seen": 303271112,
+      "step": 5270,
+      "train_runtime": 29632.5652,
+      "train_tokens_per_second": 10234.386
+    },
+    {
+      "epoch": 1.213759779107225,
+      "grad_norm": 0.764688447004531,
+      "learning_rate": 3.354210500502773e-06,
+      "loss": 0.388,
+      "num_input_tokens_seen": 303565048,
+      "step": 5275,
+      "train_runtime": 29662.977,
+      "train_tokens_per_second": 10233.802
+    },
+    {
+      "epoch": 1.2149102623101702,
+      "grad_norm": 0.7657998155014459,
+      "learning_rate": 3.3456808437025788e-06,
+      "loss": 0.4184,
+      "num_input_tokens_seen": 303865184,
+      "step": 5280,
+      "train_runtime": 29692.8745,
+      "train_tokens_per_second": 10233.606
+    },
+    {
+      "epoch": 1.2160607455131156,
+      "grad_norm": 0.856476784055678,
+      "learning_rate": 3.337156589709949e-06,
+      "loss": 0.4071,
+      "num_input_tokens_seen": 304169592,
+      "step": 5285,
+      "train_runtime": 29720.0135,
+      "train_tokens_per_second": 10234.504
+    },
+    {
+      "epoch": 1.2172112287160608,
+      "grad_norm": 0.7618347198853279,
+      "learning_rate": 3.3286377663640753e-06,
+      "loss": 0.4094,
+      "num_input_tokens_seen": 304478832,
+      "step": 5290,
+      "train_runtime": 29755.3709,
+      "train_tokens_per_second": 10232.735
+    },
+    {
+      "epoch": 1.218361711919006,
+      "grad_norm": 0.8068711270434028,
+      "learning_rate": 3.3201244014864075e-06,
+      "loss": 0.3801,
+      "num_input_tokens_seen": 304740448,
+      "step": 5295,
+      "train_runtime": 29781.1491,
+      "train_tokens_per_second": 10232.663
+    },
+    {
+      "epoch": 1.2195121951219512,
+      "grad_norm": 1.7160505057139055,
+      "learning_rate": 3.3116165228805752e-06,
+      "loss": 0.3875,
+      "num_input_tokens_seen": 305022360,
+      "step": 5300,
+      "train_runtime": 29809.2685,
+      "train_tokens_per_second": 10232.467
+    },
+    {
+      "epoch": 1.2206626783248964,
+      "grad_norm": 0.8584789049087513,
+      "learning_rate": 3.3031141583322843e-06,
+      "loss": 0.3908,
+      "num_input_tokens_seen": 305302112,
+      "step": 5305,
+      "train_runtime": 29838.5764,
+      "train_tokens_per_second": 10231.792
+    },
+    {
+      "epoch": 1.2218131615278418,
+      "grad_norm": 0.802014172685688,
+      "learning_rate": 3.29461733560924e-06,
+      "loss": 0.384,
+      "num_input_tokens_seen": 305587152,
+      "step": 5310,
+      "train_runtime": 29867.2873,
+      "train_tokens_per_second": 10231.5
+    },
+    {
+      "epoch": 1.222963644730787,
+      "grad_norm": 0.7888621317409564,
+      "learning_rate": 3.2861260824610385e-06,
+      "loss": 0.3766,
+      "num_input_tokens_seen": 305831464,
+      "step": 5315,
+      "train_runtime": 29898.0895,
+      "train_tokens_per_second": 10229.131
+    },
+    {
+      "epoch": 1.2241141279337322,
+      "grad_norm": 0.7239561709004839,
+      "learning_rate": 3.2776404266190965e-06,
+      "loss": 0.3766,
+      "num_input_tokens_seen": 306127752,
+      "step": 5320,
+      "train_runtime": 29925.9014,
+      "train_tokens_per_second": 10229.525
+    },
+    {
+      "epoch": 1.2252646111366774,
+      "grad_norm": 0.7686873400753765,
+      "learning_rate": 3.269160395796547e-06,
+      "loss": 0.4148,
+      "num_input_tokens_seen": 306416848,
+      "step": 5325,
+      "train_runtime": 29955.8047,
+      "train_tokens_per_second": 10228.964
+    },
+    {
+      "epoch": 1.2264150943396226,
+      "grad_norm": 0.7528342931599038,
+      "learning_rate": 3.260686017688148e-06,
+      "loss": 0.3771,
+      "num_input_tokens_seen": 306708632,
+      "step": 5330,
+      "train_runtime": 29983.8555,
+      "train_tokens_per_second": 10229.126
+    },
+    {
+      "epoch": 1.2275655775425678,
+      "grad_norm": 0.7060655115794888,
+      "learning_rate": 3.2522173199702036e-06,
+      "loss": 0.4029,
+      "num_input_tokens_seen": 306999968,
+      "step": 5335,
+      "train_runtime": 30015.871,
+      "train_tokens_per_second": 10227.921
+    },
+    {
+      "epoch": 1.2287160607455132,
+      "grad_norm": 0.7421050705001674,
+      "learning_rate": 3.243754330300457e-06,
+      "loss": 0.4218,
+      "num_input_tokens_seen": 307294344,
+      "step": 5340,
+      "train_runtime": 30040.2621,
+      "train_tokens_per_second": 10229.416
+    },
+    {
+      "epoch": 1.2298665439484584,
+      "grad_norm": 0.8304175682543862,
+      "learning_rate": 3.2352970763180238e-06,
+      "loss": 0.3775,
+      "num_input_tokens_seen": 307603032,
+      "step": 5345,
+      "train_runtime": 30062.2684,
+      "train_tokens_per_second": 10232.196
+    },
+    {
+      "epoch": 1.2310170271514036,
+      "grad_norm": 0.6830084552195947,
+      "learning_rate": 3.226845585643273e-06,
+      "loss": 0.3979,
+      "num_input_tokens_seen": 307887784,
+      "step": 5350,
+      "train_runtime": 30094.0732,
+      "train_tokens_per_second": 10230.845
+    },
+    {
+      "epoch": 1.2321675103543488,
+      "grad_norm": 0.7831335407218567,
+      "learning_rate": 3.218399885877761e-06,
+      "loss": 0.4056,
+      "num_input_tokens_seen": 308147912,
+      "step": 5355,
+      "train_runtime": 30114.4299,
+      "train_tokens_per_second": 10232.567
+    },
+    {
+      "epoch": 1.233317993557294,
+      "grad_norm": 0.745306520934252,
+      "learning_rate": 3.209960004604125e-06,
+      "loss": 0.3775,
+      "num_input_tokens_seen": 308418288,
+      "step": 5360,
+      "train_runtime": 30146.1651,
+      "train_tokens_per_second": 10230.764
+    },
+    {
+      "epoch": 1.2344684767602394,
+      "grad_norm": 0.7289995863347345,
+      "learning_rate": 3.2015259693860067e-06,
+      "loss": 0.3819,
+      "num_input_tokens_seen": 308719584,
+      "step": 5365,
+      "train_runtime": 30179.7772,
+      "train_tokens_per_second": 10229.353
+    },
+    {
+      "epoch": 1.2356189599631846,
+      "grad_norm": 0.9119552038693038,
+      "learning_rate": 3.1930978077679473e-06,
+      "loss": 0.4061,
+      "num_input_tokens_seen": 309021960,
+      "step": 5370,
+      "train_runtime": 30204.9543,
+      "train_tokens_per_second": 10230.837
+    },
+    {
+      "epoch": 1.2367694431661298,
+      "grad_norm": 0.8169793851481623,
+      "learning_rate": 3.1846755472753133e-06,
+      "loss": 0.4043,
+      "num_input_tokens_seen": 309314568,
+      "step": 5375,
+      "train_runtime": 30223.5168,
+      "train_tokens_per_second": 10234.235
+    },
+    {
+      "epoch": 1.237919926369075,
+      "grad_norm": 1.1380183561154287,
+      "learning_rate": 3.1762592154141957e-06,
+      "loss": 0.384,
+      "num_input_tokens_seen": 309587696,
+      "step": 5380,
+      "train_runtime": 30252.5062,
+      "train_tokens_per_second": 10233.456
+    },
+    {
+      "epoch": 1.2390704095720202,
+      "grad_norm": 0.7325818908927773,
+      "learning_rate": 3.1678488396713204e-06,
+      "loss": 0.4072,
+      "num_input_tokens_seen": 309869672,
+      "step": 5385,
+      "train_runtime": 30286.35,
+      "train_tokens_per_second": 10231.331
+    },
+    {
+      "epoch": 1.2402208927749654,
+      "grad_norm": 0.7306024302680725,
+      "learning_rate": 3.159444447513966e-06,
+      "loss": 0.4012,
+      "num_input_tokens_seen": 310170992,
+      "step": 5390,
+      "train_runtime": 30317.9728,
+      "train_tokens_per_second": 10230.598
+    },
+    {
+      "epoch": 1.2413713759779108,
+      "grad_norm": 0.8074947967546335,
+      "learning_rate": 3.1510460663898647e-06,
+      "loss": 0.3814,
+      "num_input_tokens_seen": 310449320,
+      "step": 5395,
+      "train_runtime": 30341.7825,
+      "train_tokens_per_second": 10231.743
+    },
+    {
+      "epoch": 1.242521859180856,
+      "grad_norm": 0.8995200267357046,
+      "learning_rate": 3.1426537237271223e-06,
+      "loss": 0.3905,
+      "num_input_tokens_seen": 310734856,
+      "step": 5400,
+      "train_runtime": 30372.3594,
+      "train_tokens_per_second": 10230.844
+    },
+    {
+      "epoch": 1.2436723423838012,
+      "grad_norm": 0.7348808350561553,
+      "learning_rate": 3.1342674469341216e-06,
+      "loss": 0.4004,
+      "num_input_tokens_seen": 310992904,
+      "step": 5405,
+      "train_runtime": 30405.1408,
+      "train_tokens_per_second": 10228.3
+    },
+    {
+      "epoch": 1.2448228255867464,
+      "grad_norm": 0.7026860879496805,
+      "learning_rate": 3.125887263399434e-06,
+      "loss": 0.4027,
+      "num_input_tokens_seen": 311292904,
+      "step": 5410,
+      "train_runtime": 30431.8609,
+      "train_tokens_per_second": 10229.177
+    },
+    {
+      "epoch": 1.2459733087896916,
+      "grad_norm": 0.8017436798702655,
+      "learning_rate": 3.1175132004917325e-06,
+      "loss": 0.397,
+      "num_input_tokens_seen": 311571464,
+      "step": 5415,
+      "train_runtime": 30461.5467,
+      "train_tokens_per_second": 10228.353
+    },
+    {
+      "epoch": 1.247123791992637,
+      "grad_norm": 0.7944379549806092,
+      "learning_rate": 3.1091452855596976e-06,
+      "loss": 0.3911,
+      "num_input_tokens_seen": 311868560,
+      "step": 5420,
+      "train_runtime": 30487.9763,
+      "train_tokens_per_second": 10229.231
+    },
+    {
+      "epoch": 1.2482742751955822,
+      "grad_norm": 0.7747803400177635,
+      "learning_rate": 3.100783545931937e-06,
+      "loss": 0.3909,
+      "num_input_tokens_seen": 312133240,
+      "step": 5425,
+      "train_runtime": 30521.8512,
+      "train_tokens_per_second": 10226.55
+    },
+    {
+      "epoch": 1.2494247583985274,
+      "grad_norm": 0.7473621702388492,
+      "learning_rate": 3.092428008916888e-06,
+      "loss": 0.3737,
+      "num_input_tokens_seen": 312409912,
+      "step": 5430,
+      "train_runtime": 30555.7425,
+      "train_tokens_per_second": 10224.262
+    },
+    {
+      "epoch": 1.2505752416014726,
+      "grad_norm": 0.8155057059003216,
+      "learning_rate": 3.0840787018027276e-06,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 312697904,
+      "step": 5435,
+      "train_runtime": 30584.3949,
+      "train_tokens_per_second": 10224.1
+    },
+    {
+      "epoch": 1.2517257248044178,
+      "grad_norm": 0.8622012102503992,
+      "learning_rate": 3.0757356518572917e-06,
+      "loss": 0.3964,
+      "num_input_tokens_seen": 312961952,
+      "step": 5440,
+      "train_runtime": 30609.6688,
+      "train_tokens_per_second": 10224.284
+    },
+    {
+      "epoch": 1.252876208007363,
+      "grad_norm": 0.9451261206575333,
+      "learning_rate": 3.0673988863279767e-06,
+      "loss": 0.3971,
+      "num_input_tokens_seen": 313267888,
+      "step": 5445,
+      "train_runtime": 30637.0128,
+      "train_tokens_per_second": 10225.145
+    },
+    {
+      "epoch": 1.2540266912103084,
+      "grad_norm": 0.8940986470669272,
+      "learning_rate": 3.059068432441658e-06,
+      "loss": 0.3656,
+      "num_input_tokens_seen": 313550360,
+      "step": 5450,
+      "train_runtime": 30662.6087,
+      "train_tokens_per_second": 10225.821
+    },
+    {
+      "epoch": 1.2551771744132536,
+      "grad_norm": 0.8186207363277986,
+      "learning_rate": 3.0507443174045993e-06,
+      "loss": 0.4035,
+      "num_input_tokens_seen": 313837872,
+      "step": 5455,
+      "train_runtime": 30688.0922,
+      "train_tokens_per_second": 10226.699
+    },
+    {
+      "epoch": 1.2563276576161988,
+      "grad_norm": 0.7492257827708334,
+      "learning_rate": 3.0424265684023556e-06,
+      "loss": 0.3858,
+      "num_input_tokens_seen": 314100432,
+      "step": 5460,
+      "train_runtime": 30721.1276,
+      "train_tokens_per_second": 10224.248
+    },
+    {
+      "epoch": 1.257478140819144,
+      "grad_norm": 0.8770480992381332,
+      "learning_rate": 3.0341152125996998e-06,
+      "loss": 0.3993,
+      "num_input_tokens_seen": 314386664,
+      "step": 5465,
+      "train_runtime": 30751.3599,
+      "train_tokens_per_second": 10223.504
+    },
+    {
+      "epoch": 1.2586286240220894,
+      "grad_norm": 0.77095642154202,
+      "learning_rate": 3.0258102771405183e-06,
+      "loss": 0.3821,
+      "num_input_tokens_seen": 314692400,
+      "step": 5470,
+      "train_runtime": 30783.363,
+      "train_tokens_per_second": 10222.808
+    },
+    {
+      "epoch": 1.2597791072250346,
+      "grad_norm": 0.8488277493049597,
+      "learning_rate": 3.0175117891477317e-06,
+      "loss": 0.417,
+      "num_input_tokens_seen": 315004576,
+      "step": 5475,
+      "train_runtime": 30809.0241,
+      "train_tokens_per_second": 10224.426
+    },
+    {
+      "epoch": 1.2609295904279798,
+      "grad_norm": 0.7462936156825254,
+      "learning_rate": 3.009219775723208e-06,
+      "loss": 0.3935,
+      "num_input_tokens_seen": 315303896,
+      "step": 5480,
+      "train_runtime": 30839.2601,
+      "train_tokens_per_second": 10224.107
+    },
+    {
+      "epoch": 1.262080073630925,
+      "grad_norm": 0.8093901969273054,
+      "learning_rate": 3.0009342639476645e-06,
+      "loss": 0.395,
+      "num_input_tokens_seen": 315581168,
+      "step": 5485,
+      "train_runtime": 30870.1516,
+      "train_tokens_per_second": 10222.858
+    },
+    {
+      "epoch": 1.2632305568338702,
+      "grad_norm": 0.94080963065678,
+      "learning_rate": 2.9926552808805866e-06,
+      "loss": 0.3759,
+      "num_input_tokens_seen": 315838352,
+      "step": 5490,
+      "train_runtime": 30895.863,
+      "train_tokens_per_second": 10222.675
+    },
+    {
+      "epoch": 1.2643810400368154,
+      "grad_norm": 0.8025832136775756,
+      "learning_rate": 2.98438285356014e-06,
+      "loss": 0.3713,
+      "num_input_tokens_seen": 316114784,
+      "step": 5495,
+      "train_runtime": 30921.4185,
+      "train_tokens_per_second": 10223.166
+    },
+    {
+      "epoch": 1.2655315232397606,
+      "grad_norm": 0.8509075884558909,
+      "learning_rate": 2.9761170090030755e-06,
+      "loss": 0.3797,
+      "num_input_tokens_seen": 316419192,
+      "step": 5500,
+      "train_runtime": 30956.1757,
+      "train_tokens_per_second": 10221.521
+    },
+    {
+      "epoch": 1.266682006442706,
+      "grad_norm": 0.8672748582544808,
+      "learning_rate": 2.967857774204649e-06,
+      "loss": 0.4045,
+      "num_input_tokens_seen": 316719056,
+      "step": 5505,
+      "train_runtime": 30980.9881,
+      "train_tokens_per_second": 10223.013
+    },
+    {
+      "epoch": 1.2678324896456512,
+      "grad_norm": 0.841438561512811,
+      "learning_rate": 2.9596051761385315e-06,
+      "loss": 0.3998,
+      "num_input_tokens_seen": 317008976,
+      "step": 5510,
+      "train_runtime": 31013.6049,
+      "train_tokens_per_second": 10221.61
+    },
+    {
+      "epoch": 1.2689829728485964,
+      "grad_norm": 0.8783238589632737,
+      "learning_rate": 2.9513592417567126e-06,
+      "loss": 0.3786,
+      "num_input_tokens_seen": 317265728,
+      "step": 5515,
+      "train_runtime": 31035.6879,
+      "train_tokens_per_second": 10222.61
+    },
+    {
+      "epoch": 1.2701334560515416,
+      "grad_norm": 0.777277667215578,
+      "learning_rate": 2.9431199979894266e-06,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 317568200,
+      "step": 5520,
+      "train_runtime": 31056.5987,
+      "train_tokens_per_second": 10225.466
+    },
+    {
+      "epoch": 1.271283939254487,
+      "grad_norm": 0.7614235551456291,
+      "learning_rate": 2.93488747174505e-06,
+      "loss": 0.383,
+      "num_input_tokens_seen": 317845352,
+      "step": 5525,
+      "train_runtime": 31087.9722,
+      "train_tokens_per_second": 10224.062
+    },
+    {
+      "epoch": 1.2724344224574322,
+      "grad_norm": 0.7870606424503414,
+      "learning_rate": 2.9266616899100265e-06,
+      "loss": 0.3925,
+      "num_input_tokens_seen": 318146296,
+      "step": 5530,
+      "train_runtime": 31109.9859,
+      "train_tokens_per_second": 10226.501
+    },
+    {
+      "epoch": 1.2735849056603774,
+      "grad_norm": 0.7518992324511721,
+      "learning_rate": 2.9184426793487707e-06,
+      "loss": 0.3815,
+      "num_input_tokens_seen": 318437984,
+      "step": 5535,
+      "train_runtime": 31136.0358,
+      "train_tokens_per_second": 10227.313
+    },
+    {
+      "epoch": 1.2747353888633226,
+      "grad_norm": 0.8619277883111832,
+      "learning_rate": 2.910230466903586e-06,
+      "loss": 0.4037,
+      "num_input_tokens_seen": 318711024,
+      "step": 5540,
+      "train_runtime": 31157.452,
+      "train_tokens_per_second": 10229.046
+    },
+    {
+      "epoch": 1.2758858720662678,
+      "grad_norm": 0.9187440964387367,
+      "learning_rate": 2.902025079394566e-06,
+      "loss": 0.4088,
+      "num_input_tokens_seen": 319005112,
+      "step": 5545,
+      "train_runtime": 31190.3707,
+      "train_tokens_per_second": 10227.679
+    },
+    {
+      "epoch": 1.277036355269213,
+      "grad_norm": 0.7335448146801731,
+      "learning_rate": 2.8938265436195233e-06,
+      "loss": 0.3843,
+      "num_input_tokens_seen": 319300848,
+      "step": 5550,
+      "train_runtime": 31213.5996,
+      "train_tokens_per_second": 10229.543
+    },
+    {
+      "epoch": 1.2781868384721582,
+      "grad_norm": 0.7143002237031876,
+      "learning_rate": 2.885634886353891e-06,
+      "loss": 0.393,
+      "num_input_tokens_seen": 319590856,
+      "step": 5555,
+      "train_runtime": 31240.3014,
+      "train_tokens_per_second": 10230.082
+    },
+    {
+      "epoch": 1.2793373216751036,
+      "grad_norm": 0.6736600972963379,
+      "learning_rate": 2.8774501343506358e-06,
+      "loss": 0.3719,
+      "num_input_tokens_seen": 319896968,
+      "step": 5560,
+      "train_runtime": 31271.787,
+      "train_tokens_per_second": 10229.571
+    },
+    {
+      "epoch": 1.2804878048780488,
+      "grad_norm": 0.7924585941195597,
+      "learning_rate": 2.8692723143401747e-06,
+      "loss": 0.3937,
+      "num_input_tokens_seen": 320175472,
+      "step": 5565,
+      "train_runtime": 31294.2819,
+      "train_tokens_per_second": 10231.117
+    },
+    {
+      "epoch": 1.281638288080994,
+      "grad_norm": 0.7456698695468156,
+      "learning_rate": 2.8611014530302853e-06,
+      "loss": 0.4294,
+      "num_input_tokens_seen": 320472792,
+      "step": 5570,
+      "train_runtime": 31327.6434,
+      "train_tokens_per_second": 10229.713
+    },
+    {
+      "epoch": 1.2827887712839392,
+      "grad_norm": 0.8578783232371641,
+      "learning_rate": 2.852937577106022e-06,
+      "loss": 0.3755,
+      "num_input_tokens_seen": 320761064,
+      "step": 5575,
+      "train_runtime": 31355.4835,
+      "train_tokens_per_second": 10229.824
+    },
+    {
+      "epoch": 1.2839392544868846,
+      "grad_norm": 0.7815953140072133,
+      "learning_rate": 2.8447807132296163e-06,
+      "loss": 0.3891,
+      "num_input_tokens_seen": 321032536,
+      "step": 5580,
+      "train_runtime": 31375.9369,
+      "train_tokens_per_second": 10231.807
+    },
+    {
+      "epoch": 1.2850897376898298,
+      "grad_norm": 0.7695633336892503,
+      "learning_rate": 2.8366308880404063e-06,
+      "loss": 0.3988,
+      "num_input_tokens_seen": 321302912,
+      "step": 5585,
+      "train_runtime": 31403.9965,
+      "train_tokens_per_second": 10231.275
+    },
+    {
+      "epoch": 1.286240220892775,
+      "grad_norm": 0.7928137131162329,
+      "learning_rate": 2.828488128154748e-06,
+      "loss": 0.4041,
+      "num_input_tokens_seen": 321597888,
+      "step": 5590,
+      "train_runtime": 31429.5304,
+      "train_tokens_per_second": 10232.348
+    },
+    {
+      "epoch": 1.2873907040957202,
+      "grad_norm": 0.9142888728606737,
+      "learning_rate": 2.8203524601659118e-06,
+      "loss": 0.381,
+      "num_input_tokens_seen": 321898040,
+      "step": 5595,
+      "train_runtime": 31459.3515,
+      "train_tokens_per_second": 10232.189
+    },
+    {
+      "epoch": 1.2885411872986654,
+      "grad_norm": 0.8309789073376924,
+      "learning_rate": 2.8122239106440107e-06,
+      "loss": 0.3723,
+      "num_input_tokens_seen": 322183840,
+      "step": 5600,
+      "train_runtime": 31481.9017,
+      "train_tokens_per_second": 10233.938
+    },
+    {
+      "epoch": 1.2896916705016106,
+      "grad_norm": 0.7761585396647726,
+      "learning_rate": 2.8041025061359127e-06,
+      "loss": 0.3978,
+      "num_input_tokens_seen": 322485616,
+      "step": 5605,
+      "train_runtime": 31509.2181,
+      "train_tokens_per_second": 10234.644
+    },
+    {
+      "epoch": 1.2908421537045558,
+      "grad_norm": 0.9977012852625776,
+      "learning_rate": 2.7959882731651516e-06,
+      "loss": 0.3895,
+      "num_input_tokens_seen": 322742048,
+      "step": 5610,
+      "train_runtime": 31538.8177,
+      "train_tokens_per_second": 10233.169
+    },
+    {
+      "epoch": 1.2919926369075012,
+      "grad_norm": 0.751524644967736,
+      "learning_rate": 2.78788123823183e-06,
+      "loss": 0.3904,
+      "num_input_tokens_seen": 323032016,
+      "step": 5615,
+      "train_runtime": 31564.7786,
+      "train_tokens_per_second": 10233.939
+    },
+    {
+      "epoch": 1.2931431201104464,
+      "grad_norm": 0.8092093205233648,
+      "learning_rate": 2.779781427812555e-06,
+      "loss": 0.3911,
+      "num_input_tokens_seen": 323322584,
+      "step": 5620,
+      "train_runtime": 31591.8862,
+      "train_tokens_per_second": 10234.355
+    },
+    {
+      "epoch": 1.2942936033133916,
+      "grad_norm": 0.7215006161983699,
+      "learning_rate": 2.7716888683603362e-06,
+      "loss": 0.3921,
+      "num_input_tokens_seen": 323594864,
+      "step": 5625,
+      "train_runtime": 31609.6816,
+      "train_tokens_per_second": 10237.207
+    },
+    {
+      "epoch": 1.2954440865163368,
+      "grad_norm": 0.85960135314235,
+      "learning_rate": 2.763603586304494e-06,
+      "loss": 0.4018,
+      "num_input_tokens_seen": 323904960,
+      "step": 5630,
+      "train_runtime": 31641.9744,
+      "train_tokens_per_second": 10236.56
+    },
+    {
+      "epoch": 1.2965945697192822,
+      "grad_norm": 0.7612742530133434,
+      "learning_rate": 2.7555256080505923e-06,
+      "loss": 0.3838,
+      "num_input_tokens_seen": 324173672,
+      "step": 5635,
+      "train_runtime": 31668.4196,
+      "train_tokens_per_second": 10236.497
+    },
+    {
+      "epoch": 1.2977450529222274,
+      "grad_norm": 0.7985549387133005,
+      "learning_rate": 2.747454959980333e-06,
+      "loss": 0.4097,
+      "num_input_tokens_seen": 324456464,
+      "step": 5640,
+      "train_runtime": 31701.2763,
+      "train_tokens_per_second": 10234.808
+    },
+    {
+      "epoch": 1.2988955361251726,
+      "grad_norm": 0.7767694905375168,
+      "learning_rate": 2.739391668451491e-06,
+      "loss": 0.4035,
+      "num_input_tokens_seen": 324722184,
+      "step": 5645,
+      "train_runtime": 31730.2442,
+      "train_tokens_per_second": 10233.838
+    },
+    {
+      "epoch": 1.3000460193281178,
+      "grad_norm": 0.8862728538560583,
+      "learning_rate": 2.7313357597978e-06,
+      "loss": 0.3717,
+      "num_input_tokens_seen": 324996936,
+      "step": 5650,
+      "train_runtime": 31758.9592,
+      "train_tokens_per_second": 10233.236
+    },
+    {
+      "epoch": 1.301196502531063,
+      "grad_norm": 0.790923019240714,
+      "learning_rate": 2.7232872603288917e-06,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 325275344,
+      "step": 5655,
+      "train_runtime": 31784.5506,
+      "train_tokens_per_second": 10233.756
+    },
+    {
+      "epoch": 1.3023469857340082,
+      "grad_norm": 0.8115293566043257,
+      "learning_rate": 2.715246196330199e-06,
+      "loss": 0.3854,
+      "num_input_tokens_seen": 325574488,
+      "step": 5660,
+      "train_runtime": 31810.3364,
+      "train_tokens_per_second": 10234.865
+    },
+    {
+      "epoch": 1.3034974689369534,
+      "grad_norm": 0.8602595005239099,
+      "learning_rate": 2.70721259406287e-06,
+      "loss": 0.4043,
+      "num_input_tokens_seen": 325876928,
+      "step": 5665,
+      "train_runtime": 31839.6846,
+      "train_tokens_per_second": 10234.93
+    },
+    {
+      "epoch": 1.3046479521398988,
+      "grad_norm": 0.988253436940235,
+      "learning_rate": 2.6991864797636846e-06,
+      "loss": 0.3711,
+      "num_input_tokens_seen": 326145360,
+      "step": 5670,
+      "train_runtime": 31867.4885,
+      "train_tokens_per_second": 10234.423
+    },
+    {
+      "epoch": 1.305798435342844,
+      "grad_norm": 0.8257781968705976,
+      "learning_rate": 2.691167879644968e-06,
+      "loss": 0.3756,
+      "num_input_tokens_seen": 326451320,
+      "step": 5675,
+      "train_runtime": 31899.3264,
+      "train_tokens_per_second": 10233.8
+    },
+    {
+      "epoch": 1.3069489185457892,
+      "grad_norm": 0.8250137579008184,
+      "learning_rate": 2.683156819894506e-06,
+      "loss": 0.408,
+      "num_input_tokens_seen": 326744528,
+      "step": 5680,
+      "train_runtime": 31924.3268,
+      "train_tokens_per_second": 10234.97
+    },
+    {
+      "epoch": 1.3080994017487344,
+      "grad_norm": 0.7090130053126247,
+      "learning_rate": 2.6751533266754526e-06,
+      "loss": 0.3722,
+      "num_input_tokens_seen": 327053576,
+      "step": 5685,
+      "train_runtime": 31955.9458,
+      "train_tokens_per_second": 10234.514
+    },
+    {
+      "epoch": 1.3092498849516798,
+      "grad_norm": 0.7843729397835633,
+      "learning_rate": 2.6671574261262575e-06,
+      "loss": 0.4004,
+      "num_input_tokens_seen": 327304672,
+      "step": 5690,
+      "train_runtime": 31986.9544,
+      "train_tokens_per_second": 10232.442
+    },
+    {
+      "epoch": 1.310400368154625,
+      "grad_norm": 0.7860613706061185,
+      "learning_rate": 2.6591691443605707e-06,
+      "loss": 0.3957,
+      "num_input_tokens_seen": 327586576,
+      "step": 5695,
+      "train_runtime": 32011.2334,
+      "train_tokens_per_second": 10233.488
+    },
+    {
+      "epoch": 1.3115508513575702,
+      "grad_norm": 0.8688595453223656,
+      "learning_rate": 2.651188507467161e-06,
+      "loss": 0.402,
+      "num_input_tokens_seen": 327880104,
+      "step": 5700,
+      "train_runtime": 32033.6417,
+      "train_tokens_per_second": 10235.493
+    },
+    {
+      "epoch": 1.3127013345605154,
+      "grad_norm": 0.8318397742865863,
+      "learning_rate": 2.6432155415098294e-06,
+      "loss": 0.394,
+      "num_input_tokens_seen": 328189480,
+      "step": 5705,
+      "train_runtime": 32059.3426,
+      "train_tokens_per_second": 10236.937
+    },
+    {
+      "epoch": 1.3138518177634606,
+      "grad_norm": 0.8229790352014549,
+      "learning_rate": 2.635250272527324e-06,
+      "loss": 0.3958,
+      "num_input_tokens_seen": 328469920,
+      "step": 5710,
+      "train_runtime": 32087.9232,
+      "train_tokens_per_second": 10236.559
+    },
+    {
+      "epoch": 1.3150023009664058,
+      "grad_norm": 0.7010360952613612,
+      "learning_rate": 2.6272927265332592e-06,
+      "loss": 0.4087,
+      "num_input_tokens_seen": 328753080,
+      "step": 5715,
+      "train_runtime": 32120.6333,
+      "train_tokens_per_second": 10234.95
+    },
+    {
+      "epoch": 1.316152784169351,
+      "grad_norm": 0.7990424803561591,
+      "learning_rate": 2.6193429295160165e-06,
+      "loss": 0.3928,
+      "num_input_tokens_seen": 329022768,
+      "step": 5720,
+      "train_runtime": 32141.5927,
+      "train_tokens_per_second": 10236.667
+    },
+    {
+      "epoch": 1.3173032673722964,
+      "grad_norm": 0.8658287881636831,
+      "learning_rate": 2.611400907438685e-06,
+      "loss": 0.3919,
+      "num_input_tokens_seen": 329315088,
+      "step": 5725,
+      "train_runtime": 32170.951,
+      "train_tokens_per_second": 10236.411
+    },
+    {
+      "epoch": 1.3184537505752416,
+      "grad_norm": 0.7446785045937426,
+      "learning_rate": 2.6034666862389533e-06,
+      "loss": 0.3966,
+      "num_input_tokens_seen": 329598312,
+      "step": 5730,
+      "train_runtime": 32196.0723,
+      "train_tokens_per_second": 10237.221
+    },
+    {
+      "epoch": 1.3196042337781868,
+      "grad_norm": 0.7469833189795478,
+      "learning_rate": 2.595540291829031e-06,
+      "loss": 0.3811,
+      "num_input_tokens_seen": 329894960,
+      "step": 5735,
+      "train_runtime": 32223.4819,
+      "train_tokens_per_second": 10237.719
+    },
+    {
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.7105671804802904,
+      "learning_rate": 2.587621750095572e-06,
+      "loss": 0.4087,
+      "num_input_tokens_seen": 330194056,
+      "step": 5740,
+      "train_runtime": 32253.8847,
+      "train_tokens_per_second": 10237.342
+    },
+    {
+      "epoch": 1.3219052001840774,
+      "grad_norm": 0.8092118481370377,
+      "learning_rate": 2.579711086899581e-06,
+      "loss": 0.4029,
+      "num_input_tokens_seen": 330468784,
+      "step": 5745,
+      "train_runtime": 32286.5264,
+      "train_tokens_per_second": 10235.501
+    },
+    {
+      "epoch": 1.3230556833870226,
+      "grad_norm": 0.7260635481000137,
+      "learning_rate": 2.5718083280763353e-06,
+      "loss": 0.3735,
+      "num_input_tokens_seen": 330760992,
+      "step": 5750,
+      "train_runtime": 32315.9356,
+      "train_tokens_per_second": 10235.229
+    },
+    {
+      "epoch": 1.3242061665899678,
+      "grad_norm": 0.6944418596464104,
+      "learning_rate": 2.563913499435293e-06,
+      "loss": 0.3823,
+      "num_input_tokens_seen": 331053952,
+      "step": 5755,
+      "train_runtime": 32343.1542,
+      "train_tokens_per_second": 10235.673
+    },
+    {
+      "epoch": 1.325356649792913,
+      "grad_norm": 0.7883824871001375,
+      "learning_rate": 2.556026626760017e-06,
+      "loss": 0.4017,
+      "num_input_tokens_seen": 331352768,
+      "step": 5760,
+      "train_runtime": 32365.1346,
+      "train_tokens_per_second": 10237.954
+    },
+    {
+      "epoch": 1.3265071329958582,
+      "grad_norm": 0.7652517979712139,
+      "learning_rate": 2.548147735808088e-06,
+      "loss": 0.3709,
+      "num_input_tokens_seen": 331649976,
+      "step": 5765,
+      "train_runtime": 32388.1982,
+      "train_tokens_per_second": 10239.84
+    },
+    {
+      "epoch": 1.3276576161988034,
+      "grad_norm": 0.9075200266076386,
+      "learning_rate": 2.5402768523110127e-06,
+      "loss": 0.398,
+      "num_input_tokens_seen": 331926336,
+      "step": 5770,
+      "train_runtime": 32414.8993,
+      "train_tokens_per_second": 10239.931
+    },
+    {
+      "epoch": 1.3288080994017486,
+      "grad_norm": 0.7760782613436916,
+      "learning_rate": 2.532414001974151e-06,
+      "loss": 0.4041,
+      "num_input_tokens_seen": 332201048,
+      "step": 5775,
+      "train_runtime": 32445.1855,
+      "train_tokens_per_second": 10238.84
+    },
+    {
+      "epoch": 1.329958582604694,
+      "grad_norm": 0.9796397491987876,
+      "learning_rate": 2.524559210476631e-06,
+      "loss": 0.3879,
+      "num_input_tokens_seen": 332494872,
+      "step": 5780,
+      "train_runtime": 32476.5108,
+      "train_tokens_per_second": 10238.011
+    },
+    {
+      "epoch": 1.3311090658076392,
+      "grad_norm": 0.7323292185557704,
+      "learning_rate": 2.5167125034712575e-06,
+      "loss": 0.3801,
+      "num_input_tokens_seen": 332803864,
+      "step": 5785,
+      "train_runtime": 32501.8752,
+      "train_tokens_per_second": 10239.528
+    },
+    {
+      "epoch": 1.3322595490105844,
+      "grad_norm": 0.8545909361337058,
+      "learning_rate": 2.5088739065844293e-06,
+      "loss": 0.3646,
+      "num_input_tokens_seen": 333073632,
+      "step": 5790,
+      "train_runtime": 32529.5864,
+      "train_tokens_per_second": 10239.098
+    },
+    {
+      "epoch": 1.3334100322135296,
+      "grad_norm": 0.8131149762309935,
+      "learning_rate": 2.5010434454160632e-06,
+      "loss": 0.3878,
+      "num_input_tokens_seen": 333354840,
+      "step": 5795,
+      "train_runtime": 32557.2755,
+      "train_tokens_per_second": 10239.028
+    },
+    {
+      "epoch": 1.334560515416475,
+      "grad_norm": 0.7519009218314464,
+      "learning_rate": 2.4932211455395046e-06,
+      "loss": 0.3901,
+      "num_input_tokens_seen": 333650152,
+      "step": 5800,
+      "train_runtime": 32581.1588,
+      "train_tokens_per_second": 10240.586
+    },
+    {
+      "epoch": 1.3357109986194202,
+      "grad_norm": 0.7476544419388538,
+      "learning_rate": 2.4854070325014453e-06,
+      "loss": 0.3906,
+      "num_input_tokens_seen": 333946608,
+      "step": 5805,
+      "train_runtime": 32616.8999,
+      "train_tokens_per_second": 10238.453
+    },
+    {
+      "epoch": 1.3368614818223654,
+      "grad_norm": 0.8233278677864204,
+      "learning_rate": 2.477601131821839e-06,
+      "loss": 0.4021,
+      "num_input_tokens_seen": 334233080,
+      "step": 5810,
+      "train_runtime": 32646.6828,
+      "train_tokens_per_second": 10237.888
+    },
+    {
+      "epoch": 1.3380119650253106,
+      "grad_norm": 0.8143206971932435,
+      "learning_rate": 2.46980346899382e-06,
+      "loss": 0.3902,
+      "num_input_tokens_seen": 334509104,
+      "step": 5815,
+      "train_runtime": 32674.7587,
+      "train_tokens_per_second": 10237.539
+    },
+    {
+      "epoch": 1.3391624482282558,
+      "grad_norm": 0.7064634679086612,
+      "learning_rate": 2.4620140694836185e-06,
+      "loss": 0.4082,
+      "num_input_tokens_seen": 334818840,
+      "step": 5820,
+      "train_runtime": 32699.8339,
+      "train_tokens_per_second": 10239.16
+    },
+    {
+      "epoch": 1.340312931431201,
+      "grad_norm": 0.7894397395342508,
+      "learning_rate": 2.454232958730475e-06,
+      "loss": 0.3679,
+      "num_input_tokens_seen": 335134968,
+      "step": 5825,
+      "train_runtime": 32729.4133,
+      "train_tokens_per_second": 10239.565
+    },
+    {
+      "epoch": 1.3414634146341464,
+      "grad_norm": 0.6817861189391651,
+      "learning_rate": 2.446460162146561e-06,
+      "loss": 0.3999,
+      "num_input_tokens_seen": 335439240,
+      "step": 5830,
+      "train_runtime": 32758.7496,
+      "train_tokens_per_second": 10239.684
+    },
+    {
+      "epoch": 1.3426138978370916,
+      "grad_norm": 0.8270980666028354,
+      "learning_rate": 2.438695705116903e-06,
+      "loss": 0.3591,
+      "num_input_tokens_seen": 335726408,
+      "step": 5835,
+      "train_runtime": 32785.5151,
+      "train_tokens_per_second": 10240.083
+    },
+    {
+      "epoch": 1.3437643810400368,
+      "grad_norm": 0.7944137244618905,
+      "learning_rate": 2.430939612999278e-06,
+      "loss": 0.3698,
+      "num_input_tokens_seen": 336017584,
+      "step": 5840,
+      "train_runtime": 32811.52,
+      "train_tokens_per_second": 10240.842
+    },
+    {
+      "epoch": 1.344914864242982,
+      "grad_norm": 0.7359485112859462,
+      "learning_rate": 2.4231919111241525e-06,
+      "loss": 0.3861,
+      "num_input_tokens_seen": 336293968,
+      "step": 5845,
+      "train_runtime": 32839.3582,
+      "train_tokens_per_second": 10240.577
+    },
+    {
+      "epoch": 1.3460653474459272,
+      "grad_norm": 0.7930201600451178,
+      "learning_rate": 2.4154526247945903e-06,
+      "loss": 0.3926,
+      "num_input_tokens_seen": 336576408,
+      "step": 5850,
+      "train_runtime": 32861.2179,
+      "train_tokens_per_second": 10242.36
+    },
+    {
+      "epoch": 1.3472158306488726,
+      "grad_norm": 1.9883228377700568,
+      "learning_rate": 2.4077217792861707e-06,
+      "loss": 0.4077,
+      "num_input_tokens_seen": 336851592,
+      "step": 5855,
+      "train_runtime": 32884.7422,
+      "train_tokens_per_second": 10243.401
+    },
+    {
+      "epoch": 1.3483663138518178,
+      "grad_norm": 0.8871838912843301,
+      "learning_rate": 2.399999399846906e-06,
+      "loss": 0.3798,
+      "num_input_tokens_seen": 337141704,
+      "step": 5860,
+      "train_runtime": 32911.2319,
+      "train_tokens_per_second": 10243.971
+    },
+    {
+      "epoch": 1.349516797054763,
+      "grad_norm": 0.7998418742340505,
+      "learning_rate": 2.392285511697159e-06,
+      "loss": 0.4007,
+      "num_input_tokens_seen": 337436704,
+      "step": 5865,
+      "train_runtime": 32938.8753,
+      "train_tokens_per_second": 10244.33
+    },
+    {
+      "epoch": 1.3506672802577082,
+      "grad_norm": 0.7091462332936241,
+      "learning_rate": 2.384580140029563e-06,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 337711872,
+      "step": 5870,
+      "train_runtime": 32968.9195,
+      "train_tokens_per_second": 10243.341
+    },
+    {
+      "epoch": 1.3518177634606534,
+      "grad_norm": 0.807900150823506,
+      "learning_rate": 2.3768833100089335e-06,
+      "loss": 0.3938,
+      "num_input_tokens_seen": 337993752,
+      "step": 5875,
+      "train_runtime": 32994.7188,
+      "train_tokens_per_second": 10243.874
+    },
+    {
+      "epoch": 1.3529682466635986,
+      "grad_norm": 0.7726534382836008,
+      "learning_rate": 2.369195046772193e-06,
+      "loss": 0.3743,
+      "num_input_tokens_seen": 338287936,
+      "step": 5880,
+      "train_runtime": 33019.855,
+      "train_tokens_per_second": 10244.985
+    },
+    {
+      "epoch": 1.354118729866544,
+      "grad_norm": 0.8282279649275321,
+      "learning_rate": 2.3615153754282867e-06,
+      "loss": 0.3654,
+      "num_input_tokens_seen": 338576144,
+      "step": 5885,
+      "train_runtime": 33043.1699,
+      "train_tokens_per_second": 10246.479
+    },
+    {
+      "epoch": 1.3552692130694892,
+      "grad_norm": 1.043239840668336,
+      "learning_rate": 2.353844321058097e-06,
+      "loss": 0.3954,
+      "num_input_tokens_seen": 338840800,
+      "step": 5890,
+      "train_runtime": 33065.7332,
+      "train_tokens_per_second": 10247.491
+    },
+    {
+      "epoch": 1.3564196962724344,
+      "grad_norm": 0.7577484521230726,
+      "learning_rate": 2.3461819087143657e-06,
+      "loss": 0.374,
+      "num_input_tokens_seen": 339134560,
+      "step": 5895,
+      "train_runtime": 33100.8514,
+      "train_tokens_per_second": 10245.494
+    },
+    {
+      "epoch": 1.3575701794753796,
+      "grad_norm": 0.6960489487135174,
+      "learning_rate": 2.338528163421612e-06,
+      "loss": 0.4039,
+      "num_input_tokens_seen": 339431656,
+      "step": 5900,
+      "train_runtime": 33129.8632,
+      "train_tokens_per_second": 10245.489
+    },
+    {
+      "epoch": 1.358720662678325,
+      "grad_norm": 0.8520551406247164,
+      "learning_rate": 2.330883110176049e-06,
+      "loss": 0.3989,
+      "num_input_tokens_seen": 339731616,
+      "step": 5905,
+      "train_runtime": 33159.4315,
+      "train_tokens_per_second": 10245.399
+    },
+    {
+      "epoch": 1.3598711458812702,
+      "grad_norm": 0.7335245819924312,
+      "learning_rate": 2.323246773945496e-06,
+      "loss": 0.4014,
+      "num_input_tokens_seen": 340022872,
+      "step": 5910,
+      "train_runtime": 33187.4111,
+      "train_tokens_per_second": 10245.538
+    },
+    {
+      "epoch": 1.3610216290842154,
+      "grad_norm": 0.9274331864728906,
+      "learning_rate": 2.3156191796693162e-06,
+      "loss": 0.3981,
+      "num_input_tokens_seen": 340300600,
+      "step": 5915,
+      "train_runtime": 33216.5056,
+      "train_tokens_per_second": 10244.925
+    },
+    {
+      "epoch": 1.3621721122871606,
+      "grad_norm": 0.841667461394614,
+      "learning_rate": 2.3080003522583156e-06,
+      "loss": 0.4012,
+      "num_input_tokens_seen": 340616464,
+      "step": 5920,
+      "train_runtime": 33239.8617,
+      "train_tokens_per_second": 10247.229
+    },
+    {
+      "epoch": 1.3633225954901058,
+      "grad_norm": 0.7989241799206739,
+      "learning_rate": 2.3003903165946646e-06,
+      "loss": 0.4009,
+      "num_input_tokens_seen": 340890352,
+      "step": 5925,
+      "train_runtime": 33265.3149,
+      "train_tokens_per_second": 10247.621
+    },
+    {
+      "epoch": 1.364473078693051,
+      "grad_norm": 0.7402148676737335,
+      "learning_rate": 2.2927890975318275e-06,
+      "loss": 0.402,
+      "num_input_tokens_seen": 341205728,
+      "step": 5930,
+      "train_runtime": 33292.5575,
+      "train_tokens_per_second": 10248.709
+    },
+    {
+      "epoch": 1.3656235618959962,
+      "grad_norm": 0.7783594218207359,
+      "learning_rate": 2.2851967198944725e-06,
+      "loss": 0.3868,
+      "num_input_tokens_seen": 341503904,
+      "step": 5935,
+      "train_runtime": 33319.8296,
+      "train_tokens_per_second": 10249.269
+    },
+    {
+      "epoch": 1.3667740450989416,
+      "grad_norm": 0.7653530323567577,
+      "learning_rate": 2.277613208478392e-06,
+      "loss": 0.3883,
+      "num_input_tokens_seen": 341811624,
+      "step": 5940,
+      "train_runtime": 33346.0879,
+      "train_tokens_per_second": 10250.427
+    },
+    {
+      "epoch": 1.3679245283018868,
+      "grad_norm": 0.9065286519051775,
+      "learning_rate": 2.270038588050424e-06,
+      "loss": 0.4192,
+      "num_input_tokens_seen": 342104848,
+      "step": 5945,
+      "train_runtime": 33379.2222,
+      "train_tokens_per_second": 10249.036
+    },
+    {
+      "epoch": 1.369075011504832,
+      "grad_norm": 0.8034220450436759,
+      "learning_rate": 2.2624728833483672e-06,
+      "loss": 0.3744,
+      "num_input_tokens_seen": 342371584,
+      "step": 5950,
+      "train_runtime": 33410.4133,
+      "train_tokens_per_second": 10247.451
+    },
+    {
+      "epoch": 1.3702254947077772,
+      "grad_norm": 0.914306912393805,
+      "learning_rate": 2.2549161190809047e-06,
+      "loss": 0.3905,
+      "num_input_tokens_seen": 342653376,
+      "step": 5955,
+      "train_runtime": 33436.5917,
+      "train_tokens_per_second": 10247.856
+    },
+    {
+      "epoch": 1.3713759779107226,
+      "grad_norm": 0.7709228547630085,
+      "learning_rate": 2.2473683199275233e-06,
+      "loss": 0.3914,
+      "num_input_tokens_seen": 342947352,
+      "step": 5960,
+      "train_runtime": 33466.9073,
+      "train_tokens_per_second": 10247.357
+    },
+    {
+      "epoch": 1.3725264611136678,
+      "grad_norm": 0.7293255428363595,
+      "learning_rate": 2.2398295105384206e-06,
+      "loss": 0.4061,
+      "num_input_tokens_seen": 343246784,
+      "step": 5965,
+      "train_runtime": 33484.5143,
+      "train_tokens_per_second": 10250.911
+    },
+    {
+      "epoch": 1.373676944316613,
+      "grad_norm": 0.7529080197128678,
+      "learning_rate": 2.232299715534449e-06,
+      "loss": 0.3977,
+      "num_input_tokens_seen": 343529536,
+      "step": 5970,
+      "train_runtime": 33511.5041,
+      "train_tokens_per_second": 10251.093
+    },
+    {
+      "epoch": 1.3748274275195582,
+      "grad_norm": 0.783056292052615,
+      "learning_rate": 2.2247789595070135e-06,
+      "loss": 0.3829,
+      "num_input_tokens_seen": 343815144,
+      "step": 5975,
+      "train_runtime": 33545.6072,
+      "train_tokens_per_second": 10249.185
+    },
+    {
+      "epoch": 1.3759779107225034,
+      "grad_norm": 0.8537761141314953,
+      "learning_rate": 2.2172672670179964e-06,
+      "loss": 0.3777,
+      "num_input_tokens_seen": 344082392,
+      "step": 5980,
+      "train_runtime": 33571.5102,
+      "train_tokens_per_second": 10249.238
+    },
+    {
+      "epoch": 1.3771283939254486,
+      "grad_norm": 0.769385877750302,
+      "learning_rate": 2.209764662599683e-06,
+      "loss": 0.3808,
+      "num_input_tokens_seen": 344367328,
+      "step": 5985,
+      "train_runtime": 33600.3787,
+      "train_tokens_per_second": 10248.912
+    },
+    {
+      "epoch": 1.3782788771283938,
+      "grad_norm": 0.8053820919460166,
+      "learning_rate": 2.20227117075468e-06,
+      "loss": 0.3835,
+      "num_input_tokens_seen": 344660224,
+      "step": 5990,
+      "train_runtime": 33622.7429,
+      "train_tokens_per_second": 10250.806
+    },
+    {
+      "epoch": 1.3794293603313392,
+      "grad_norm": 0.7666141359349935,
+      "learning_rate": 2.1947868159558304e-06,
+      "loss": 0.38,
+      "num_input_tokens_seen": 344960672,
+      "step": 5995,
+      "train_runtime": 33651.9786,
+      "train_tokens_per_second": 10250.829
+    },
+    {
+      "epoch": 1.3805798435342844,
+      "grad_norm": 0.7447495881444395,
+      "learning_rate": 2.1873116226461364e-06,
+      "loss": 0.3929,
+      "num_input_tokens_seen": 345265552,
+      "step": 6000,
+      "train_runtime": 33675.7529,
+      "train_tokens_per_second": 10252.645
+    },
+    {
+      "epoch": 1.3817303267372296,
+      "grad_norm": 1.140548373441048,
+      "learning_rate": 2.179845615238683e-06,
+      "loss": 0.4028,
+      "num_input_tokens_seen": 345540952,
+      "step": 6005,
+      "train_runtime": 33704.8672,
+      "train_tokens_per_second": 10251.96
+    },
+    {
+      "epoch": 1.3828808099401748,
+      "grad_norm": 0.6706618849241764,
+      "learning_rate": 2.1723888181165535e-06,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 345824920,
+      "step": 6010,
+      "train_runtime": 33727.9098,
+      "train_tokens_per_second": 10253.375
+    },
+    {
+      "epoch": 1.3840312931431202,
+      "grad_norm": 0.9240162955404578,
+      "learning_rate": 2.164941255632748e-06,
+      "loss": 0.4083,
+      "num_input_tokens_seen": 346085040,
+      "step": 6015,
+      "train_runtime": 33755.1758,
+      "train_tokens_per_second": 10252.799
+    },
+    {
+      "epoch": 1.3851817763460654,
+      "grad_norm": 0.738617844973373,
+      "learning_rate": 2.1575029521101115e-06,
+      "loss": 0.3954,
+      "num_input_tokens_seen": 346376216,
+      "step": 6020,
+      "train_runtime": 33783.7878,
+      "train_tokens_per_second": 10252.735
+    },
+    {
+      "epoch": 1.3863322595490106,
+      "grad_norm": 0.7472125553663894,
+      "learning_rate": 2.150073931841253e-06,
+      "loss": 0.422,
+      "num_input_tokens_seen": 346672768,
+      "step": 6025,
+      "train_runtime": 33811.5938,
+      "train_tokens_per_second": 10253.074
+    },
+    {
+      "epoch": 1.3874827427519558,
+      "grad_norm": 0.7594974939172832,
+      "learning_rate": 2.142654219088457e-06,
+      "loss": 0.3909,
+      "num_input_tokens_seen": 346977832,
+      "step": 6030,
+      "train_runtime": 33839.9457,
+      "train_tokens_per_second": 10253.498
+    },
+    {
+      "epoch": 1.388633225954901,
+      "grad_norm": 0.7608539509518897,
+      "learning_rate": 2.135243838083613e-06,
+      "loss": 0.3853,
+      "num_input_tokens_seen": 347280616,
+      "step": 6035,
+      "train_runtime": 33865.7112,
+      "train_tokens_per_second": 10254.638
+    },
+    {
+      "epoch": 1.3897837091578462,
+      "grad_norm": 0.904998551045219,
+      "learning_rate": 2.1278428130281363e-06,
+      "loss": 0.3842,
+      "num_input_tokens_seen": 347548848,
+      "step": 6040,
+      "train_runtime": 33890.7066,
+      "train_tokens_per_second": 10254.99
+    },
+    {
+      "epoch": 1.3909341923607914,
+      "grad_norm": 0.9157601663755023,
+      "learning_rate": 2.1204511680928854e-06,
+      "loss": 0.3921,
+      "num_input_tokens_seen": 347849144,
+      "step": 6045,
+      "train_runtime": 33923.914,
+      "train_tokens_per_second": 10253.803
+    },
+    {
+      "epoch": 1.3920846755637368,
+      "grad_norm": 0.7965733682072061,
+      "learning_rate": 2.113068927418084e-06,
+      "loss": 0.3742,
+      "num_input_tokens_seen": 348137128,
+      "step": 6050,
+      "train_runtime": 33952.7092,
+      "train_tokens_per_second": 10253.589
+    },
+    {
+      "epoch": 1.393235158766682,
+      "grad_norm": 0.7171609642597983,
+      "learning_rate": 2.105696115113243e-06,
+      "loss": 0.3823,
+      "num_input_tokens_seen": 348442384,
+      "step": 6055,
+      "train_runtime": 33979.6213,
+      "train_tokens_per_second": 10254.452
+    },
+    {
+      "epoch": 1.3943856419696272,
+      "grad_norm": 0.8135258565640745,
+      "learning_rate": 2.098332755257083e-06,
+      "loss": 0.3793,
+      "num_input_tokens_seen": 348751120,
+      "step": 6060,
+      "train_runtime": 34009.0318,
+      "train_tokens_per_second": 10254.662
+    },
+    {
+      "epoch": 1.3955361251725724,
+      "grad_norm": 0.7991326630211882,
+      "learning_rate": 2.0909788718974493e-06,
+      "loss": 0.3943,
+      "num_input_tokens_seen": 349046808,
+      "step": 6065,
+      "train_runtime": 34039.3682,
+      "train_tokens_per_second": 10254.209
+    },
+    {
+      "epoch": 1.3966866083755178,
+      "grad_norm": 0.7340963817593918,
+      "learning_rate": 2.083634489051243e-06,
+      "loss": 0.4057,
+      "num_input_tokens_seen": 349338528,
+      "step": 6070,
+      "train_runtime": 34067.538,
+      "train_tokens_per_second": 10254.293
+    },
+    {
+      "epoch": 1.397837091578463,
+      "grad_norm": 0.8427501663933088,
+      "learning_rate": 2.076299630704334e-06,
+      "loss": 0.3693,
+      "num_input_tokens_seen": 349632792,
+      "step": 6075,
+      "train_runtime": 34095.4084,
+      "train_tokens_per_second": 10254.542
+    },
+    {
+      "epoch": 1.3989875747814082,
+      "grad_norm": 0.815720586144207,
+      "learning_rate": 2.068974320811493e-06,
+      "loss": 0.3699,
+      "num_input_tokens_seen": 349930104,
+      "step": 6080,
+      "train_runtime": 34121.9607,
+      "train_tokens_per_second": 10255.275
+    },
+    {
+      "epoch": 1.4001380579843534,
+      "grad_norm": 0.7190280208197313,
+      "learning_rate": 2.0616585832962975e-06,
+      "loss": 0.3726,
+      "num_input_tokens_seen": 350242528,
+      "step": 6085,
+      "train_runtime": 34145.6121,
+      "train_tokens_per_second": 10257.322
+    },
+    {
+      "epoch": 1.4012885411872986,
+      "grad_norm": 0.7494330104296019,
+      "learning_rate": 2.0543524420510687e-06,
+      "loss": 0.364,
+      "num_input_tokens_seen": 350536256,
+      "step": 6090,
+      "train_runtime": 34172.1702,
+      "train_tokens_per_second": 10257.945
+    },
+    {
+      "epoch": 1.4024390243902438,
+      "grad_norm": 0.8693275250749927,
+      "learning_rate": 2.047055920936785e-06,
+      "loss": 0.3791,
+      "num_input_tokens_seen": 350812640,
+      "step": 6095,
+      "train_runtime": 34204.7696,
+      "train_tokens_per_second": 10256.249
+    },
+    {
+      "epoch": 1.403589507593189,
+      "grad_norm": 0.7126367686364434,
+      "learning_rate": 2.0397690437830087e-06,
+      "loss": 0.3895,
+      "num_input_tokens_seen": 351082192,
+      "step": 6100,
+      "train_runtime": 34233.2321,
+      "train_tokens_per_second": 10255.596
+    },
+    {
+      "epoch": 1.4047399907961344,
+      "grad_norm": 0.8713470557404911,
+      "learning_rate": 2.032491834387804e-06,
+      "loss": 0.375,
+      "num_input_tokens_seen": 351336112,
+      "step": 6105,
+      "train_runtime": 34261.7091,
+      "train_tokens_per_second": 10254.483
+    },
+    {
+      "epoch": 1.4058904739990796,
+      "grad_norm": 0.782205186100688,
+      "learning_rate": 2.025224316517663e-06,
+      "loss": 0.3834,
+      "num_input_tokens_seen": 351642400,
+      "step": 6110,
+      "train_runtime": 34291.5385,
+      "train_tokens_per_second": 10254.495
+    },
+    {
+      "epoch": 1.4070409572020248,
+      "grad_norm": 0.8045018538193445,
+      "learning_rate": 2.017966513907427e-06,
+      "loss": 0.4035,
+      "num_input_tokens_seen": 351938192,
+      "step": 6115,
+      "train_runtime": 34315.7137,
+      "train_tokens_per_second": 10255.89
+    },
+    {
+      "epoch": 1.40819144040497,
+      "grad_norm": 0.7440262814911213,
+      "learning_rate": 2.010718450260204e-06,
+      "loss": 0.373,
+      "num_input_tokens_seen": 352236720,
+      "step": 6120,
+      "train_runtime": 34345.1484,
+      "train_tokens_per_second": 10255.793
+    },
+    {
+      "epoch": 1.4093419236079154,
+      "grad_norm": 0.7378149718286886,
+      "learning_rate": 2.0034801492473004e-06,
+      "loss": 0.382,
+      "num_input_tokens_seen": 352521064,
+      "step": 6125,
+      "train_runtime": 34374.6312,
+      "train_tokens_per_second": 10255.268
+    },
+    {
+      "epoch": 1.4104924068108606,
+      "grad_norm": 0.7376780647312631,
+      "learning_rate": 1.996251634508137e-06,
+      "loss": 0.4004,
+      "num_input_tokens_seen": 352800360,
+      "step": 6130,
+      "train_runtime": 34403.0388,
+      "train_tokens_per_second": 10254.919
+    },
+    {
+      "epoch": 1.4116428900138058,
+      "grad_norm": 0.8198966106493434,
+      "learning_rate": 1.9890329296501756e-06,
+      "loss": 0.3993,
+      "num_input_tokens_seen": 353090096,
+      "step": 6135,
+      "train_runtime": 34437.3604,
+      "train_tokens_per_second": 10253.111
+    },
+    {
+      "epoch": 1.412793373216751,
+      "grad_norm": 0.7944926551572694,
+      "learning_rate": 1.9818240582488376e-06,
+      "loss": 0.3864,
+      "num_input_tokens_seen": 353394840,
+      "step": 6140,
+      "train_runtime": 34462.9716,
+      "train_tokens_per_second": 10254.335
+    },
+    {
+      "epoch": 1.4139438564196962,
+      "grad_norm": 0.7845367874026294,
+      "learning_rate": 1.9746250438474306e-06,
+      "loss": 0.4079,
+      "num_input_tokens_seen": 353698928,
+      "step": 6145,
+      "train_runtime": 34483.7994,
+      "train_tokens_per_second": 10256.959
+    },
+    {
+      "epoch": 1.4150943396226414,
+      "grad_norm": 0.6997792955100524,
+      "learning_rate": 1.9674359099570733e-06,
+      "loss": 0.3765,
+      "num_input_tokens_seen": 353998376,
+      "step": 6150,
+      "train_runtime": 34512.8569,
+      "train_tokens_per_second": 10257.0
+    },
+    {
+      "epoch": 1.4162448228255866,
+      "grad_norm": 0.8453029438803158,
+      "learning_rate": 1.9602566800566077e-06,
+      "loss": 0.3991,
+      "num_input_tokens_seen": 354287664,
+      "step": 6155,
+      "train_runtime": 34537.9115,
+      "train_tokens_per_second": 10257.935
+    },
+    {
+      "epoch": 1.417395306028532,
+      "grad_norm": 0.7592671397985029,
+      "learning_rate": 1.95308737759254e-06,
+      "loss": 0.3812,
+      "num_input_tokens_seen": 354559704,
+      "step": 6160,
+      "train_runtime": 34563.707,
+      "train_tokens_per_second": 10258.15
+    },
+    {
+      "epoch": 1.4185457892314772,
+      "grad_norm": 0.7598328851019657,
+      "learning_rate": 1.9459280259789513e-06,
+      "loss": 0.4004,
+      "num_input_tokens_seen": 354860776,
+      "step": 6165,
+      "train_runtime": 34593.9416,
+      "train_tokens_per_second": 10257.888
+    },
+    {
+      "epoch": 1.4196962724344224,
+      "grad_norm": 0.7986998788546134,
+      "learning_rate": 1.938778648597421e-06,
+      "loss": 0.3836,
+      "num_input_tokens_seen": 355167408,
+      "step": 6170,
+      "train_runtime": 34625.2201,
+      "train_tokens_per_second": 10257.477
+    },
+    {
+      "epoch": 1.4208467556373676,
+      "grad_norm": 0.7853439941239752,
+      "learning_rate": 1.9316392687969564e-06,
+      "loss": 0.3849,
+      "num_input_tokens_seen": 355452184,
+      "step": 6175,
+      "train_runtime": 34656.0818,
+      "train_tokens_per_second": 10256.56
+    },
+    {
+      "epoch": 1.421997238840313,
+      "grad_norm": 0.7208992407636923,
+      "learning_rate": 1.9245099098939156e-06,
+      "loss": 0.3649,
+      "num_input_tokens_seen": 355744760,
+      "step": 6180,
+      "train_runtime": 34686.489,
+      "train_tokens_per_second": 10256.004
+    },
+    {
+      "epoch": 1.4231477220432582,
+      "grad_norm": 0.7700254672059196,
+      "learning_rate": 1.917390595171927e-06,
+      "loss": 0.4051,
+      "num_input_tokens_seen": 356039992,
+      "step": 6185,
+      "train_runtime": 34711.5889,
+      "train_tokens_per_second": 10257.093
+    },
+    {
+      "epoch": 1.4242982052462034,
+      "grad_norm": 0.8071293482472153,
+      "learning_rate": 1.9102813478818176e-06,
+      "loss": 0.3746,
+      "num_input_tokens_seen": 356307384,
+      "step": 6190,
+      "train_runtime": 34745.8243,
+      "train_tokens_per_second": 10254.682
+    },
+    {
+      "epoch": 1.4254486884491486,
+      "grad_norm": 0.9614681493453296,
+      "learning_rate": 1.9031821912415343e-06,
+      "loss": 0.3926,
+      "num_input_tokens_seen": 356597432,
+      "step": 6195,
+      "train_runtime": 34775.0577,
+      "train_tokens_per_second": 10254.402
+    },
+    {
+      "epoch": 1.4265991716520938,
+      "grad_norm": 0.7705567725907875,
+      "learning_rate": 1.8960931484360717e-06,
+      "loss": 0.3971,
+      "num_input_tokens_seen": 356883488,
+      "step": 6200,
+      "train_runtime": 34803.0331,
+      "train_tokens_per_second": 10254.379
+    },
+    {
+      "epoch": 1.427749654855039,
+      "grad_norm": 0.8331019932585293,
+      "learning_rate": 1.8890142426173873e-06,
+      "loss": 0.3755,
+      "num_input_tokens_seen": 357158560,
+      "step": 6205,
+      "train_runtime": 34823.1301,
+      "train_tokens_per_second": 10256.36
+    },
+    {
+      "epoch": 1.4289001380579842,
+      "grad_norm": 0.7314990404638625,
+      "learning_rate": 1.8819454969043383e-06,
+      "loss": 0.3938,
+      "num_input_tokens_seen": 357430136,
+      "step": 6210,
+      "train_runtime": 34848.0719,
+      "train_tokens_per_second": 10256.812
+    },
+    {
+      "epoch": 1.4300506212609296,
+      "grad_norm": 0.7392401636641908,
+      "learning_rate": 1.8748869343826032e-06,
+      "loss": 0.3749,
+      "num_input_tokens_seen": 357715560,
+      "step": 6215,
+      "train_runtime": 34877.9248,
+      "train_tokens_per_second": 10256.217
+    },
+    {
+      "epoch": 1.4312011044638748,
+      "grad_norm": 0.6797628355936115,
+      "learning_rate": 1.8678385781045932e-06,
+      "loss": 0.3806,
+      "num_input_tokens_seen": 358017584,
+      "step": 6220,
+      "train_runtime": 34913.0274,
+      "train_tokens_per_second": 10254.556
+    },
+    {
+      "epoch": 1.43235158766682,
+      "grad_norm": 0.7256302378274757,
+      "learning_rate": 1.8608004510893957e-06,
+      "loss": 0.3895,
+      "num_input_tokens_seen": 358322880,
+      "step": 6225,
+      "train_runtime": 34943.4633,
+      "train_tokens_per_second": 10254.361
+    },
+    {
+      "epoch": 1.4335020708697652,
+      "grad_norm": 0.7849200223774631,
+      "learning_rate": 1.8537725763226871e-06,
+      "loss": 0.3752,
+      "num_input_tokens_seen": 358587352,
+      "step": 6230,
+      "train_runtime": 34971.9772,
+      "train_tokens_per_second": 10253.562
+    },
+    {
+      "epoch": 1.4346525540727106,
+      "grad_norm": 0.819999289337283,
+      "learning_rate": 1.8467549767566624e-06,
+      "loss": 0.3907,
+      "num_input_tokens_seen": 358872912,
+      "step": 6235,
+      "train_runtime": 34997.6396,
+      "train_tokens_per_second": 10254.203
+    },
+    {
+      "epoch": 1.4358030372756558,
+      "grad_norm": 0.7481020750904736,
+      "learning_rate": 1.8397476753099575e-06,
+      "loss": 0.3828,
+      "num_input_tokens_seen": 359166368,
+      "step": 6240,
+      "train_runtime": 35023.5987,
+      "train_tokens_per_second": 10254.982
+    },
+    {
+      "epoch": 1.436953520478601,
+      "grad_norm": 1.0699689097508824,
+      "learning_rate": 1.8327506948675777e-06,
+      "loss": 0.3882,
+      "num_input_tokens_seen": 359451304,
+      "step": 6245,
+      "train_runtime": 35054.9116,
+      "train_tokens_per_second": 10253.95
+    },
+    {
+      "epoch": 1.4381040036815462,
+      "grad_norm": 0.7701223386457023,
+      "learning_rate": 1.8257640582808207e-06,
+      "loss": 0.4093,
+      "num_input_tokens_seen": 359752640,
+      "step": 6250,
+      "train_runtime": 35090.1591,
+      "train_tokens_per_second": 10252.237
+    },
+    {
+      "epoch": 1.4392544868844914,
+      "grad_norm": 0.7766183109565218,
+      "learning_rate": 1.8187877883672024e-06,
+      "loss": 0.3778,
+      "num_input_tokens_seen": 360050352,
+      "step": 6255,
+      "train_runtime": 35116.7669,
+      "train_tokens_per_second": 10252.947
+    },
+    {
+      "epoch": 1.4404049700874366,
+      "grad_norm": 0.9022104646513082,
+      "learning_rate": 1.8118219079103789e-06,
+      "loss": 0.3866,
+      "num_input_tokens_seen": 360324208,
+      "step": 6260,
+      "train_runtime": 35138.1781,
+      "train_tokens_per_second": 10254.493
+    },
+    {
+      "epoch": 1.441555453290382,
+      "grad_norm": 0.886922140041944,
+      "learning_rate": 1.8048664396600796e-06,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 360608224,
+      "step": 6265,
+      "train_runtime": 35169.7483,
+      "train_tokens_per_second": 10253.364
+    },
+    {
+      "epoch": 1.4427059364933272,
+      "grad_norm": 0.8491076848468636,
+      "learning_rate": 1.7979214063320328e-06,
+      "loss": 0.3708,
+      "num_input_tokens_seen": 360896688,
+      "step": 6270,
+      "train_runtime": 35199.9733,
+      "train_tokens_per_second": 10252.755
+    },
+    {
+      "epoch": 1.4438564196962724,
+      "grad_norm": 0.9802079858936508,
+      "learning_rate": 1.7909868306078765e-06,
+      "loss": 0.3654,
+      "num_input_tokens_seen": 361153480,
+      "step": 6275,
+      "train_runtime": 35225.4276,
+      "train_tokens_per_second": 10252.636
+    },
+    {
+      "epoch": 1.4450069028992176,
+      "grad_norm": 0.7729545228210233,
+      "learning_rate": 1.7840627351351047e-06,
+      "loss": 0.3886,
+      "num_input_tokens_seen": 361416752,
+      "step": 6280,
+      "train_runtime": 35251.5381,
+      "train_tokens_per_second": 10252.51
+    },
+    {
+      "epoch": 1.446157386102163,
+      "grad_norm": 0.8569106225920576,
+      "learning_rate": 1.777149142526981e-06,
+      "loss": 0.4006,
+      "num_input_tokens_seen": 361730144,
+      "step": 6285,
+      "train_runtime": 35284.5664,
+      "train_tokens_per_second": 10251.795
+    },
+    {
+      "epoch": 1.4473078693051082,
+      "grad_norm": 0.7684137701377277,
+      "learning_rate": 1.7702460753624674e-06,
+      "loss": 0.3871,
+      "num_input_tokens_seen": 362008296,
+      "step": 6290,
+      "train_runtime": 35312.9186,
+      "train_tokens_per_second": 10251.441
+    },
+    {
+      "epoch": 1.4484583525080534,
+      "grad_norm": 0.7307655594815168,
+      "learning_rate": 1.763353556186152e-06,
+      "loss": 0.3867,
+      "num_input_tokens_seen": 362319904,
+      "step": 6295,
+      "train_runtime": 35340.3121,
+      "train_tokens_per_second": 10252.312
+    },
+    {
+      "epoch": 1.4496088357109986,
+      "grad_norm": 0.7668374898488654,
+      "learning_rate": 1.7564716075081745e-06,
+      "loss": 0.3908,
+      "num_input_tokens_seen": 362601912,
+      "step": 6300,
+      "train_runtime": 35363.9548,
+      "train_tokens_per_second": 10253.432
+    },
+    {
+      "epoch": 1.4507593189139438,
+      "grad_norm": 0.7664806308227466,
+      "learning_rate": 1.749600251804154e-06,
+      "loss": 0.3833,
+      "num_input_tokens_seen": 362885864,
+      "step": 6305,
+      "train_runtime": 35390.0153,
+      "train_tokens_per_second": 10253.905
+    },
+    {
+      "epoch": 1.451909802116889,
+      "grad_norm": 0.7920815197925978,
+      "learning_rate": 1.742739511515108e-06,
+      "loss": 0.3768,
+      "num_input_tokens_seen": 363174512,
+      "step": 6310,
+      "train_runtime": 35420.1883,
+      "train_tokens_per_second": 10253.32
+    },
+    {
+      "epoch": 1.4530602853198342,
+      "grad_norm": 0.7836116547702102,
+      "learning_rate": 1.7358894090473928e-06,
+      "loss": 0.3823,
+      "num_input_tokens_seen": 363430064,
+      "step": 6315,
+      "train_runtime": 35449.4676,
+      "train_tokens_per_second": 10252.06
+    },
+    {
+      "epoch": 1.4542107685227796,
+      "grad_norm": 0.7611430189062584,
+      "learning_rate": 1.7290499667726197e-06,
+      "loss": 0.3904,
+      "num_input_tokens_seen": 363716368,
+      "step": 6320,
+      "train_runtime": 35475.0167,
+      "train_tokens_per_second": 10252.747
+    },
+    {
+      "epoch": 1.4553612517257248,
+      "grad_norm": 0.7446729093248826,
+      "learning_rate": 1.7222212070275846e-06,
+      "loss": 0.3961,
+      "num_input_tokens_seen": 364002264,
+      "step": 6325,
+      "train_runtime": 35498.7093,
+      "train_tokens_per_second": 10253.958
+    },
+    {
+      "epoch": 1.45651173492867,
+      "grad_norm": 0.8595781249698223,
+      "learning_rate": 1.7154031521141963e-06,
+      "loss": 0.3769,
+      "num_input_tokens_seen": 364305488,
+      "step": 6330,
+      "train_runtime": 35521.5198,
+      "train_tokens_per_second": 10255.909
+    },
+    {
+      "epoch": 1.4576622181316152,
+      "grad_norm": 0.8154775083595163,
+      "learning_rate": 1.7085958242994038e-06,
+      "loss": 0.3923,
+      "num_input_tokens_seen": 364614880,
+      "step": 6335,
+      "train_runtime": 35546.4468,
+      "train_tokens_per_second": 10257.421
+    },
+    {
+      "epoch": 1.4588127013345606,
+      "grad_norm": 0.7883774722312771,
+      "learning_rate": 1.7017992458151223e-06,
+      "loss": 0.3886,
+      "num_input_tokens_seen": 364919448,
+      "step": 6340,
+      "train_runtime": 35573.9057,
+      "train_tokens_per_second": 10258.065
+    },
+    {
+      "epoch": 1.4599631845375058,
+      "grad_norm": 0.9371874068783669,
+      "learning_rate": 1.6950134388581557e-06,
+      "loss": 0.3926,
+      "num_input_tokens_seen": 365214944,
+      "step": 6345,
+      "train_runtime": 35597.4507,
+      "train_tokens_per_second": 10259.581
+    },
+    {
+      "epoch": 1.461113667740451,
+      "grad_norm": 0.7815990446976682,
+      "learning_rate": 1.6882384255901396e-06,
+      "loss": 0.3808,
+      "num_input_tokens_seen": 365494832,
+      "step": 6350,
+      "train_runtime": 35624.5283,
+      "train_tokens_per_second": 10259.64
+    },
+    {
+      "epoch": 1.4622641509433962,
+      "grad_norm": 0.735391307802894,
+      "learning_rate": 1.6814742281374525e-06,
+      "loss": 0.3636,
+      "num_input_tokens_seen": 365799296,
+      "step": 6355,
+      "train_runtime": 35654.7726,
+      "train_tokens_per_second": 10259.476
+    },
+    {
+      "epoch": 1.4634146341463414,
+      "grad_norm": 0.7997141821876287,
+      "learning_rate": 1.674720868591147e-06,
+      "loss": 0.3813,
+      "num_input_tokens_seen": 366109368,
+      "step": 6360,
+      "train_runtime": 35684.3906,
+      "train_tokens_per_second": 10259.65
+    },
+    {
+      "epoch": 1.4645651173492866,
+      "grad_norm": 1.058137135965536,
+      "learning_rate": 1.6679783690068863e-06,
+      "loss": 0.4046,
+      "num_input_tokens_seen": 366380992,
+      "step": 6365,
+      "train_runtime": 35706.6504,
+      "train_tokens_per_second": 10260.861
+    },
+    {
+      "epoch": 1.4657156005522318,
+      "grad_norm": 0.8347387710883338,
+      "learning_rate": 1.661246751404863e-06,
+      "loss": 0.4038,
+      "num_input_tokens_seen": 366686936,
+      "step": 6370,
+      "train_runtime": 35728.9685,
+      "train_tokens_per_second": 10263.015
+    },
+    {
+      "epoch": 1.4668660837551772,
+      "grad_norm": 0.769136857953041,
+      "learning_rate": 1.654526037769732e-06,
+      "loss": 0.3841,
+      "num_input_tokens_seen": 366951320,
+      "step": 6375,
+      "train_runtime": 35753.0435,
+      "train_tokens_per_second": 10263.499
+    },
+    {
+      "epoch": 1.4680165669581224,
+      "grad_norm": 0.7671961225853601,
+      "learning_rate": 1.6478162500505357e-06,
+      "loss": 0.384,
+      "num_input_tokens_seen": 367210280,
+      "step": 6380,
+      "train_runtime": 35776.9592,
+      "train_tokens_per_second": 10263.876
+    },
+    {
+      "epoch": 1.4691670501610676,
+      "grad_norm": 0.7446400970905629,
+      "learning_rate": 1.6411174101606348e-06,
+      "loss": 0.3997,
+      "num_input_tokens_seen": 367485544,
+      "step": 6385,
+      "train_runtime": 35806.1841,
+      "train_tokens_per_second": 10263.186
+    },
+    {
+      "epoch": 1.4703175333640128,
+      "grad_norm": 0.6946175312403658,
+      "learning_rate": 1.6344295399776356e-06,
+      "loss": 0.3857,
+      "num_input_tokens_seen": 367796688,
+      "step": 6390,
+      "train_runtime": 35834.9641,
+      "train_tokens_per_second": 10263.627
+    },
+    {
+      "epoch": 1.4714680165669582,
+      "grad_norm": 0.7804915182336017,
+      "learning_rate": 1.6277526613433192e-06,
+      "loss": 0.3667,
+      "num_input_tokens_seen": 368091960,
+      "step": 6395,
+      "train_runtime": 35863.7779,
+      "train_tokens_per_second": 10263.614
+    },
+    {
+      "epoch": 1.4726184997699034,
+      "grad_norm": 0.7410242267575071,
+      "learning_rate": 1.6210867960635652e-06,
+      "loss": 0.3804,
+      "num_input_tokens_seen": 368389248,
+      "step": 6400,
+      "train_runtime": 35889.5367,
+      "train_tokens_per_second": 10264.531
+    },
+    {
+      "epoch": 1.4737689829728486,
+      "grad_norm": 0.8143465588168266,
+      "learning_rate": 1.6144319659082925e-06,
+      "loss": 0.3968,
+      "num_input_tokens_seen": 368669056,
+      "step": 6405,
+      "train_runtime": 35916.9017,
+      "train_tokens_per_second": 10264.501
+    },
+    {
+      "epoch": 1.4749194661757938,
+      "grad_norm": 0.809465913231125,
+      "learning_rate": 1.607788192611377e-06,
+      "loss": 0.3926,
+      "num_input_tokens_seen": 368949816,
+      "step": 6410,
+      "train_runtime": 35944.0469,
+      "train_tokens_per_second": 10264.56
+    },
+    {
+      "epoch": 1.476069949378739,
+      "grad_norm": 0.8464027874005609,
+      "learning_rate": 1.6011554978705807e-06,
+      "loss": 0.3906,
+      "num_input_tokens_seen": 369224072,
+      "step": 6415,
+      "train_runtime": 35969.2844,
+      "train_tokens_per_second": 10264.982
+    },
+    {
+      "epoch": 1.4772204325816842,
+      "grad_norm": 0.8509885059823917,
+      "learning_rate": 1.594533903347489e-06,
+      "loss": 0.3879,
+      "num_input_tokens_seen": 369529248,
+      "step": 6420,
+      "train_runtime": 36000.8612,
+      "train_tokens_per_second": 10264.456
+    },
+    {
+      "epoch": 1.4783709157846294,
+      "grad_norm": 0.8384596724474684,
+      "learning_rate": 1.587923430667434e-06,
+      "loss": 0.3835,
+      "num_input_tokens_seen": 369787704,
+      "step": 6425,
+      "train_runtime": 36023.7756,
+      "train_tokens_per_second": 10265.101
+    },
+    {
+      "epoch": 1.4795213989875748,
+      "grad_norm": 0.8459382851566605,
+      "learning_rate": 1.5813241014194252e-06,
+      "loss": 0.3846,
+      "num_input_tokens_seen": 370070944,
+      "step": 6430,
+      "train_runtime": 36053.8689,
+      "train_tokens_per_second": 10264.389
+    },
+    {
+      "epoch": 1.48067188219052,
+      "grad_norm": 0.8181797271295612,
+      "learning_rate": 1.574735937156079e-06,
+      "loss": 0.3815,
+      "num_input_tokens_seen": 370340352,
+      "step": 6435,
+      "train_runtime": 36075.294,
+      "train_tokens_per_second": 10265.761
+    },
+    {
+      "epoch": 1.4818223653934652,
+      "grad_norm": 0.8012943781757003,
+      "learning_rate": 1.5681589593935476e-06,
+      "loss": 0.3812,
+      "num_input_tokens_seen": 370620824,
+      "step": 6440,
+      "train_runtime": 36100.9636,
+      "train_tokens_per_second": 10266.231
+    },
+    {
+      "epoch": 1.4829728485964104,
+      "grad_norm": 0.7087214231683348,
+      "learning_rate": 1.5615931896114527e-06,
+      "loss": 0.3984,
+      "num_input_tokens_seen": 370919272,
+      "step": 6445,
+      "train_runtime": 36123.5099,
+      "train_tokens_per_second": 10268.085
+    },
+    {
+      "epoch": 1.4841233317993558,
+      "grad_norm": 0.9045730581227293,
+      "learning_rate": 1.5550386492528053e-06,
+      "loss": 0.3965,
+      "num_input_tokens_seen": 371213464,
+      "step": 6450,
+      "train_runtime": 36152.0606,
+      "train_tokens_per_second": 10268.114
+    },
+    {
+      "epoch": 1.485273815002301,
+      "grad_norm": 0.854571287397404,
+      "learning_rate": 1.548495359723946e-06,
+      "loss": 0.4005,
+      "num_input_tokens_seen": 371500560,
+      "step": 6455,
+      "train_runtime": 36181.8456,
+      "train_tokens_per_second": 10267.596
+    },
+    {
+      "epoch": 1.4864242982052462,
+      "grad_norm": 0.7494316376604265,
+      "learning_rate": 1.5419633423944785e-06,
+      "loss": 0.3931,
+      "num_input_tokens_seen": 371800624,
+      "step": 6460,
+      "train_runtime": 36207.0631,
+      "train_tokens_per_second": 10268.732
+    },
+    {
+      "epoch": 1.4875747814081914,
+      "grad_norm": 0.7717701906388188,
+      "learning_rate": 1.5354426185971804e-06,
+      "loss": 0.3766,
+      "num_input_tokens_seen": 372100208,
+      "step": 6465,
+      "train_runtime": 36236.2382,
+      "train_tokens_per_second": 10268.732
+    },
+    {
+      "epoch": 1.4887252646111366,
+      "grad_norm": 0.8250700212180457,
+      "learning_rate": 1.5289332096279546e-06,
+      "loss": 0.3848,
+      "num_input_tokens_seen": 372394944,
+      "step": 6470,
+      "train_runtime": 36263.9624,
+      "train_tokens_per_second": 10269.009
+    },
+    {
+      "epoch": 1.4898757478140818,
+      "grad_norm": 0.8185527162709576,
+      "learning_rate": 1.5224351367457473e-06,
+      "loss": 0.3981,
+      "num_input_tokens_seen": 372704016,
+      "step": 6475,
+      "train_runtime": 36292.0653,
+      "train_tokens_per_second": 10269.573
+    },
+    {
+      "epoch": 1.491026231017027,
+      "grad_norm": 0.8147122386129088,
+      "learning_rate": 1.5159484211724845e-06,
+      "loss": 0.4105,
+      "num_input_tokens_seen": 372999896,
+      "step": 6480,
+      "train_runtime": 36322.5617,
+      "train_tokens_per_second": 10269.097
+    },
+    {
+      "epoch": 1.4921767142199724,
+      "grad_norm": 0.9238925975898454,
+      "learning_rate": 1.5094730840929999e-06,
+      "loss": 0.3913,
+      "num_input_tokens_seen": 373256120,
+      "step": 6485,
+      "train_runtime": 36347.9702,
+      "train_tokens_per_second": 10268.967
+    },
+    {
+      "epoch": 1.4933271974229176,
+      "grad_norm": 0.7919508962246863,
+      "learning_rate": 1.5030091466549663e-06,
+      "loss": 0.3998,
+      "num_input_tokens_seen": 373552248,
+      "step": 6490,
+      "train_runtime": 36370.4479,
+      "train_tokens_per_second": 10270.763
+    },
+    {
+      "epoch": 1.4944776806258628,
+      "grad_norm": 0.8003590260557675,
+      "learning_rate": 1.4965566299688273e-06,
+      "loss": 0.3785,
+      "num_input_tokens_seen": 373826240,
+      "step": 6495,
+      "train_runtime": 36399.0625,
+      "train_tokens_per_second": 10270.216
+    },
+    {
+      "epoch": 1.495628163828808,
+      "grad_norm": 0.7917848454378508,
+      "learning_rate": 1.490115555107724e-06,
+      "loss": 0.4083,
+      "num_input_tokens_seen": 374121256,
+      "step": 6500,
+      "train_runtime": 36427.621,
+      "train_tokens_per_second": 10270.263
+    },
+    {
+      "epoch": 1.4967786470317535,
+      "grad_norm": 0.7884228292245306,
+      "learning_rate": 1.4836859431074358e-06,
+      "loss": 0.4018,
+      "num_input_tokens_seen": 374422792,
+      "step": 6505,
+      "train_runtime": 36453.7944,
+      "train_tokens_per_second": 10271.161
+    },
+    {
+      "epoch": 1.4979291302346986,
+      "grad_norm": 0.7570502616028848,
+      "learning_rate": 1.477267814966301e-06,
+      "loss": 0.3744,
+      "num_input_tokens_seen": 374721360,
+      "step": 6510,
+      "train_runtime": 36472.5912,
+      "train_tokens_per_second": 10274.054
+    },
+    {
+      "epoch": 1.4990796134376438,
+      "grad_norm": 0.7980531361944629,
+      "learning_rate": 1.4708611916451565e-06,
+      "loss": 0.3791,
+      "num_input_tokens_seen": 374985832,
+      "step": 6515,
+      "train_runtime": 36502.9568,
+      "train_tokens_per_second": 10272.752
+    },
+    {
+      "epoch": 1.500230096640589,
+      "grad_norm": 0.7895102543893451,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.3806,
+      "num_input_tokens_seen": 375295016,
+      "step": 6520,
+      "train_runtime": 36534.1286,
+      "train_tokens_per_second": 10272.45
+    },
+    {
+      "epoch": 1.5013805798435342,
+      "grad_norm": 0.8422126045914385,
+      "learning_rate": 1.458082543118241e-06,
+      "loss": 0.3812,
+      "num_input_tokens_seen": 375579496,
+      "step": 6525,
+      "train_runtime": 36554.9671,
+      "train_tokens_per_second": 10274.377
+    },
+    {
+      "epoch": 1.5025310630464794,
+      "grad_norm": 0.7252873327066667,
+      "learning_rate": 1.4517105596460008e-06,
+      "loss": 0.3817,
+      "num_input_tokens_seen": 375861312,
+      "step": 6530,
+      "train_runtime": 36585.5999,
+      "train_tokens_per_second": 10273.477
+    },
+    {
+      "epoch": 1.5036815462494246,
+      "grad_norm": 0.7065460061952743,
+      "learning_rate": 1.4453501644606754e-06,
+      "loss": 0.379,
+      "num_input_tokens_seen": 376143248,
+      "step": 6535,
+      "train_runtime": 36609.9343,
+      "train_tokens_per_second": 10274.349
+    },
+    {
+      "epoch": 1.50483202945237,
+      "grad_norm": 0.8746984575711356,
+      "learning_rate": 1.4390013783345503e-06,
+      "loss": 0.4134,
+      "num_input_tokens_seen": 376434968,
+      "step": 6540,
+      "train_runtime": 36640.9383,
+      "train_tokens_per_second": 10273.617
+    },
+    {
+      "epoch": 1.5059825126553152,
+      "grad_norm": 0.8238982389883732,
+      "learning_rate": 1.4326642220019976e-06,
+      "loss": 0.3995,
+      "num_input_tokens_seen": 376742720,
+      "step": 6545,
+      "train_runtime": 36674.5568,
+      "train_tokens_per_second": 10272.591
+    },
+    {
+      "epoch": 1.5071329958582604,
+      "grad_norm": 0.7539588459294444,
+      "learning_rate": 1.426338716159411e-06,
+      "loss": 0.4116,
+      "num_input_tokens_seen": 377048488,
+      "step": 6550,
+      "train_runtime": 36704.2697,
+      "train_tokens_per_second": 10272.606
+    },
+    {
+      "epoch": 1.5082834790612059,
+      "grad_norm": 0.8198799778817825,
+      "learning_rate": 1.4200248814651296e-06,
+      "loss": 0.375,
+      "num_input_tokens_seen": 377346256,
+      "step": 6555,
+      "train_runtime": 36733.5188,
+      "train_tokens_per_second": 10272.532
+    },
+    {
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.851123437604095,
+      "learning_rate": 1.4137227385393787e-06,
+      "loss": 0.394,
+      "num_input_tokens_seen": 377640864,
+      "step": 6560,
+      "train_runtime": 36761.2627,
+      "train_tokens_per_second": 10272.794
+    },
+    {
+      "epoch": 1.5105844454670962,
+      "grad_norm": 0.9042515407154946,
+      "learning_rate": 1.407432307964201e-06,
+      "loss": 0.378,
+      "num_input_tokens_seen": 377941880,
+      "step": 6565,
+      "train_runtime": 36787.629,
+      "train_tokens_per_second": 10273.613
+    },
+    {
+      "epoch": 1.5117349286700414,
+      "grad_norm": 0.8052726813802006,
+      "learning_rate": 1.4011536102833872e-06,
+      "loss": 0.3959,
+      "num_input_tokens_seen": 378239040,
+      "step": 6570,
+      "train_runtime": 36818.1982,
+      "train_tokens_per_second": 10273.155
+    },
+    {
+      "epoch": 1.5128854118729866,
+      "grad_norm": 0.7914343850658984,
+      "learning_rate": 1.394886666002408e-06,
+      "loss": 0.3852,
+      "num_input_tokens_seen": 378533440,
+      "step": 6575,
+      "train_runtime": 36841.3239,
+      "train_tokens_per_second": 10274.697
+    },
+    {
+      "epoch": 1.5140358950759318,
+      "grad_norm": 0.8246016855967352,
+      "learning_rate": 1.3886314955883511e-06,
+      "loss": 0.406,
+      "num_input_tokens_seen": 378838136,
+      "step": 6580,
+      "train_runtime": 36874.5004,
+      "train_tokens_per_second": 10273.716
+    },
+    {
+      "epoch": 1.515186378278877,
+      "grad_norm": 0.7458553268883757,
+      "learning_rate": 1.3823881194698535e-06,
+      "loss": 0.3949,
+      "num_input_tokens_seen": 379158336,
+      "step": 6585,
+      "train_runtime": 36904.9008,
+      "train_tokens_per_second": 10273.929
+    },
+    {
+      "epoch": 1.5163368614818222,
+      "grad_norm": 0.7980813823255363,
+      "learning_rate": 1.3761565580370262e-06,
+      "loss": 0.3828,
+      "num_input_tokens_seen": 379474744,
+      "step": 6590,
+      "train_runtime": 36930.0338,
+      "train_tokens_per_second": 10275.505
+    },
+    {
+      "epoch": 1.5174873446847676,
+      "grad_norm": 0.7624609176850852,
+      "learning_rate": 1.3699368316414041e-06,
+      "loss": 0.3856,
+      "num_input_tokens_seen": 379774480,
+      "step": 6595,
+      "train_runtime": 36961.4043,
+      "train_tokens_per_second": 10274.893
+    },
+    {
+      "epoch": 1.5186378278877128,
+      "grad_norm": 0.7200824486176366,
+      "learning_rate": 1.3637289605958665e-06,
+      "loss": 0.418,
+      "num_input_tokens_seen": 380069976,
+      "step": 6600,
+      "train_runtime": 36988.4027,
+      "train_tokens_per_second": 10275.382
+    },
+    {
+      "epoch": 1.519788311090658,
+      "grad_norm": 0.7471146241766161,
+      "learning_rate": 1.3575329651745718e-06,
+      "loss": 0.3891,
+      "num_input_tokens_seen": 380385120,
+      "step": 6605,
+      "train_runtime": 37016.7177,
+      "train_tokens_per_second": 10276.036
+    },
+    {
+      "epoch": 1.5209387942936035,
+      "grad_norm": 0.9147422541069357,
+      "learning_rate": 1.351348865612897e-06,
+      "loss": 0.3576,
+      "num_input_tokens_seen": 380663272,
+      "step": 6610,
+      "train_runtime": 37049.9848,
+      "train_tokens_per_second": 10274.317
+    },
+    {
+      "epoch": 1.5220892774965487,
+      "grad_norm": 0.817176013861196,
+      "learning_rate": 1.3451766821073692e-06,
+      "loss": 0.413,
+      "num_input_tokens_seen": 380949416,
+      "step": 6615,
+      "train_runtime": 37081.0739,
+      "train_tokens_per_second": 10273.419
+    },
+    {
+      "epoch": 1.5232397606994938,
+      "grad_norm": 0.7739257996532357,
+      "learning_rate": 1.3390164348155971e-06,
+      "loss": 0.4082,
+      "num_input_tokens_seen": 381249048,
+      "step": 6620,
+      "train_runtime": 37107.4384,
+      "train_tokens_per_second": 10274.195
+    },
+    {
+      "epoch": 1.524390243902439,
+      "grad_norm": 0.852807437856639,
+      "learning_rate": 1.332868143856208e-06,
+      "loss": 0.3823,
+      "num_input_tokens_seen": 381543072,
+      "step": 6625,
+      "train_runtime": 37139.3428,
+      "train_tokens_per_second": 10273.286
+    },
+    {
+      "epoch": 1.5255407271053842,
+      "grad_norm": 0.7419153417872477,
+      "learning_rate": 1.3267318293087821e-06,
+      "loss": 0.4079,
+      "num_input_tokens_seen": 381853976,
+      "step": 6630,
+      "train_runtime": 37166.0639,
+      "train_tokens_per_second": 10274.265
+    },
+    {
+      "epoch": 1.5266912103083294,
+      "grad_norm": 0.8090922872286956,
+      "learning_rate": 1.3206075112137868e-06,
+      "loss": 0.3752,
+      "num_input_tokens_seen": 382141328,
+      "step": 6635,
+      "train_runtime": 37198.1799,
+      "train_tokens_per_second": 10273.119
+    },
+    {
+      "epoch": 1.5278416935112746,
+      "grad_norm": 0.9942345841120283,
+      "learning_rate": 1.3144952095725055e-06,
+      "loss": 0.4035,
+      "num_input_tokens_seen": 382433992,
+      "step": 6640,
+      "train_runtime": 37228.6872,
+      "train_tokens_per_second": 10272.562
+    },
+    {
+      "epoch": 1.5289921767142198,
+      "grad_norm": 0.7274624489448284,
+      "learning_rate": 1.3083949443469812e-06,
+      "loss": 0.4002,
+      "num_input_tokens_seen": 382739088,
+      "step": 6645,
+      "train_runtime": 37256.1676,
+      "train_tokens_per_second": 10273.174
+    },
+    {
+      "epoch": 1.5301426599171652,
+      "grad_norm": 0.8850535117281844,
+      "learning_rate": 1.3023067354599512e-06,
+      "loss": 0.4119,
+      "num_input_tokens_seen": 383029264,
+      "step": 6650,
+      "train_runtime": 37281.8086,
+      "train_tokens_per_second": 10273.892
+    },
+    {
+      "epoch": 1.5312931431201104,
+      "grad_norm": 0.7135656843720358,
+      "learning_rate": 1.2962306027947702e-06,
+      "loss": 0.3844,
+      "num_input_tokens_seen": 383315944,
+      "step": 6655,
+      "train_runtime": 37302.4317,
+      "train_tokens_per_second": 10275.897
+    },
+    {
+      "epoch": 1.5324436263230556,
+      "grad_norm": 0.8346759353083981,
+      "learning_rate": 1.2901665661953578e-06,
+      "loss": 0.3978,
+      "num_input_tokens_seen": 383613984,
+      "step": 6660,
+      "train_runtime": 37332.4026,
+      "train_tokens_per_second": 10275.631
+    },
+    {
+      "epoch": 1.533594109526001,
+      "grad_norm": 0.9508112663195791,
+      "learning_rate": 1.2841146454661303e-06,
+      "loss": 0.4046,
+      "num_input_tokens_seen": 383899880,
+      "step": 6665,
+      "train_runtime": 37364.0731,
+      "train_tokens_per_second": 10274.573
+    },
+    {
+      "epoch": 1.5347445927289463,
+      "grad_norm": 0.6892259197737968,
+      "learning_rate": 1.2780748603719317e-06,
+      "loss": 0.3652,
+      "num_input_tokens_seen": 384200976,
+      "step": 6670,
+      "train_runtime": 37386.4029,
+      "train_tokens_per_second": 10276.489
+    },
+    {
+      "epoch": 1.5358950759318915,
+      "grad_norm": 0.8299359101923944,
+      "learning_rate": 1.2720472306379762e-06,
+      "loss": 0.3736,
+      "num_input_tokens_seen": 384502816,
+      "step": 6675,
+      "train_runtime": 37413.489,
+      "train_tokens_per_second": 10277.117
+    },
+    {
+      "epoch": 1.5370455591348366,
+      "grad_norm": 0.9270384634194787,
+      "learning_rate": 1.2660317759497763e-06,
+      "loss": 0.3887,
+      "num_input_tokens_seen": 384808536,
+      "step": 6680,
+      "train_runtime": 37440.8758,
+      "train_tokens_per_second": 10277.765
+    },
+    {
+      "epoch": 1.5381960423377818,
+      "grad_norm": 0.7717318394030848,
+      "learning_rate": 1.2600285159530861e-06,
+      "loss": 0.3932,
+      "num_input_tokens_seen": 385089960,
+      "step": 6685,
+      "train_runtime": 37471.0135,
+      "train_tokens_per_second": 10277.009
+    },
+    {
+      "epoch": 1.539346525540727,
+      "grad_norm": 0.7906992710286411,
+      "learning_rate": 1.254037470253831e-06,
+      "loss": 0.3727,
+      "num_input_tokens_seen": 385366256,
+      "step": 6690,
+      "train_runtime": 37503.3891,
+      "train_tokens_per_second": 10275.505
+    },
+    {
+      "epoch": 1.5404970087436722,
+      "grad_norm": 0.8522286728801836,
+      "learning_rate": 1.2480586584180448e-06,
+      "loss": 0.3835,
+      "num_input_tokens_seen": 385642224,
+      "step": 6695,
+      "train_runtime": 37529.3878,
+      "train_tokens_per_second": 10275.74
+    },
+    {
+      "epoch": 1.5416474919466174,
+      "grad_norm": 0.9440637617458236,
+      "learning_rate": 1.2420920999718079e-06,
+      "loss": 0.389,
+      "num_input_tokens_seen": 385934832,
+      "step": 6700,
+      "train_runtime": 37555.1179,
+      "train_tokens_per_second": 10276.491
+    },
+    {
+      "epoch": 1.5427979751495629,
+      "grad_norm": 0.8600257976640353,
+      "learning_rate": 1.236137814401187e-06,
+      "loss": 0.3909,
+      "num_input_tokens_seen": 386224952,
+      "step": 6705,
+      "train_runtime": 37581.501,
+      "train_tokens_per_second": 10276.996
+    },
+    {
+      "epoch": 1.543948458352508,
+      "grad_norm": 0.7698053520728887,
+      "learning_rate": 1.2301958211521602e-06,
+      "loss": 0.3619,
+      "num_input_tokens_seen": 386519744,
+      "step": 6710,
+      "train_runtime": 37610.8383,
+      "train_tokens_per_second": 10276.818
+    },
+    {
+      "epoch": 1.5450989415554532,
+      "grad_norm": 0.7743009765953179,
+      "learning_rate": 1.2242661396305633e-06,
+      "loss": 0.3761,
+      "num_input_tokens_seen": 386812160,
+      "step": 6715,
+      "train_runtime": 37647.196,
+      "train_tokens_per_second": 10274.661
+    },
+    {
+      "epoch": 1.5462494247583987,
+      "grad_norm": 0.8844700431338475,
+      "learning_rate": 1.218348789202025e-06,
+      "loss": 0.3791,
+      "num_input_tokens_seen": 387069136,
+      "step": 6720,
+      "train_runtime": 37674.0702,
+      "train_tokens_per_second": 10274.152
+    },
+    {
+      "epoch": 1.5473999079613439,
+      "grad_norm": 0.760305797779611,
+      "learning_rate": 1.2124437891918995e-06,
+      "loss": 0.378,
+      "num_input_tokens_seen": 387352008,
+      "step": 6725,
+      "train_runtime": 37706.0646,
+      "train_tokens_per_second": 10272.937
+    },
+    {
+      "epoch": 1.548550391164289,
+      "grad_norm": 0.8228074976035217,
+      "learning_rate": 1.206551158885208e-06,
+      "loss": 0.4053,
+      "num_input_tokens_seen": 387651184,
+      "step": 6730,
+      "train_runtime": 37738.6109,
+      "train_tokens_per_second": 10272.005
+    },
+    {
+      "epoch": 1.5497008743672342,
+      "grad_norm": 0.7957089249560045,
+      "learning_rate": 1.200670917526573e-06,
+      "loss": 0.3875,
+      "num_input_tokens_seen": 387959592,
+      "step": 6735,
+      "train_runtime": 37765.5313,
+      "train_tokens_per_second": 10272.849
+    },
+    {
+      "epoch": 1.5508513575701794,
+      "grad_norm": 0.8078911153131725,
+      "learning_rate": 1.1948030843201574e-06,
+      "loss": 0.3801,
+      "num_input_tokens_seen": 388223920,
+      "step": 6740,
+      "train_runtime": 37801.3513,
+      "train_tokens_per_second": 10270.107
+    },
+    {
+      "epoch": 1.5520018407731246,
+      "grad_norm": 0.7733156615397773,
+      "learning_rate": 1.1889476784295962e-06,
+      "loss": 0.3721,
+      "num_input_tokens_seen": 388515560,
+      "step": 6745,
+      "train_runtime": 37823.5774,
+      "train_tokens_per_second": 10271.782
+    },
+    {
+      "epoch": 1.5531523239760698,
+      "grad_norm": 0.717112161372567,
+      "learning_rate": 1.1831047189779426e-06,
+      "loss": 0.3901,
+      "num_input_tokens_seen": 388796432,
+      "step": 6750,
+      "train_runtime": 37851.1845,
+      "train_tokens_per_second": 10271.711
+    },
+    {
+      "epoch": 1.554302807179015,
+      "grad_norm": 0.8168536463931357,
+      "learning_rate": 1.177274225047601e-06,
+      "loss": 0.4147,
+      "num_input_tokens_seen": 389083800,
+      "step": 6755,
+      "train_runtime": 37872.4841,
+      "train_tokens_per_second": 10273.522
+    },
+    {
+      "epoch": 1.5554532903819605,
+      "grad_norm": 0.7719545039846679,
+      "learning_rate": 1.1714562156802634e-06,
+      "loss": 0.3806,
+      "num_input_tokens_seen": 389379056,
+      "step": 6760,
+      "train_runtime": 37895.3448,
+      "train_tokens_per_second": 10275.116
+    },
+    {
+      "epoch": 1.5566037735849056,
+      "grad_norm": 0.73469201563824,
+      "learning_rate": 1.1656507098768489e-06,
+      "loss": 0.383,
+      "num_input_tokens_seen": 389671672,
+      "step": 6765,
+      "train_runtime": 37927.0403,
+      "train_tokens_per_second": 10274.244
+    },
+    {
+      "epoch": 1.5577542567878508,
+      "grad_norm": 0.7987722293133451,
+      "learning_rate": 1.159857726597442e-06,
+      "loss": 0.3793,
+      "num_input_tokens_seen": 389962424,
+      "step": 6770,
+      "train_runtime": 37956.2798,
+      "train_tokens_per_second": 10273.99
+    },
+    {
+      "epoch": 1.5589047399907963,
+      "grad_norm": 0.7743917993241336,
+      "learning_rate": 1.154077284761232e-06,
+      "loss": 0.4078,
+      "num_input_tokens_seen": 390242920,
+      "step": 6775,
+      "train_runtime": 37979.1098,
+      "train_tokens_per_second": 10275.199
+    },
+    {
+      "epoch": 1.5600552231937415,
+      "grad_norm": 0.700896740472449,
+      "learning_rate": 1.1483094032464421e-06,
+      "loss": 0.3907,
+      "num_input_tokens_seen": 390506888,
+      "step": 6780,
+      "train_runtime": 38003.6282,
+      "train_tokens_per_second": 10275.516
+    },
+    {
+      "epoch": 1.5612057063966867,
+      "grad_norm": 0.928209260963755,
+      "learning_rate": 1.1425541008902852e-06,
+      "loss": 0.3858,
+      "num_input_tokens_seen": 390778312,
+      "step": 6785,
+      "train_runtime": 38031.9513,
+      "train_tokens_per_second": 10275.0
+    },
+    {
+      "epoch": 1.5623561895996319,
+      "grad_norm": 0.8805491747941411,
+      "learning_rate": 1.1368113964888866e-06,
+      "loss": 0.3875,
+      "num_input_tokens_seen": 391066640,
+      "step": 6790,
+      "train_runtime": 38058.6956,
+      "train_tokens_per_second": 10275.356
+    },
+    {
+      "epoch": 1.563506672802577,
+      "grad_norm": 0.8933026807595094,
+      "learning_rate": 1.1310813087972262e-06,
+      "loss": 0.3832,
+      "num_input_tokens_seen": 391333520,
+      "step": 6795,
+      "train_runtime": 38090.2856,
+      "train_tokens_per_second": 10273.841
+    },
+    {
+      "epoch": 1.5646571560055222,
+      "grad_norm": 0.6953353321347652,
+      "learning_rate": 1.1253638565290837e-06,
+      "loss": 0.4112,
+      "num_input_tokens_seen": 391621576,
+      "step": 6800,
+      "train_runtime": 38121.0716,
+      "train_tokens_per_second": 10273.1
+    },
+    {
+      "epoch": 1.5658076392084674,
+      "grad_norm": 0.7557421485078251,
+      "learning_rate": 1.11965905835697e-06,
+      "loss": 0.3916,
+      "num_input_tokens_seen": 391923848,
+      "step": 6805,
+      "train_runtime": 38153.2308,
+      "train_tokens_per_second": 10272.363
+    },
+    {
+      "epoch": 1.5669581224114126,
+      "grad_norm": 0.7845926813356168,
+      "learning_rate": 1.113966932912071e-06,
+      "loss": 0.3742,
+      "num_input_tokens_seen": 392206376,
+      "step": 6810,
+      "train_runtime": 38183.1071,
+      "train_tokens_per_second": 10271.725
+    },
+    {
+      "epoch": 1.568108605614358,
+      "grad_norm": 0.8029674392078301,
+      "learning_rate": 1.1082874987841852e-06,
+      "loss": 0.3944,
+      "num_input_tokens_seen": 392507504,
+      "step": 6815,
+      "train_runtime": 38208.9192,
+      "train_tokens_per_second": 10272.667
+    },
+    {
+      "epoch": 1.5692590888173032,
+      "grad_norm": 0.8682707616269303,
+      "learning_rate": 1.1026207745216611e-06,
+      "loss": 0.3778,
+      "num_input_tokens_seen": 392778136,
+      "step": 6820,
+      "train_runtime": 38239.4719,
+      "train_tokens_per_second": 10271.537
+    },
+    {
+      "epoch": 1.5704095720202487,
+      "grad_norm": 0.7653986858374723,
+      "learning_rate": 1.0969667786313388e-06,
+      "loss": 0.4016,
+      "num_input_tokens_seen": 393072632,
+      "step": 6825,
+      "train_runtime": 38259.1629,
+      "train_tokens_per_second": 10273.948
+    },
+    {
+      "epoch": 1.5715600552231939,
+      "grad_norm": 0.8089806664303577,
+      "learning_rate": 1.0913255295784913e-06,
+      "loss": 0.3842,
+      "num_input_tokens_seen": 393366888,
+      "step": 6830,
+      "train_runtime": 38290.4822,
+      "train_tokens_per_second": 10273.229
+    },
+    {
+      "epoch": 1.572710538426139,
+      "grad_norm": 0.7139102742386213,
+      "learning_rate": 1.085697045786756e-06,
+      "loss": 0.3807,
+      "num_input_tokens_seen": 393668648,
+      "step": 6835,
+      "train_runtime": 38316.565,
+      "train_tokens_per_second": 10274.111
+    },
+    {
+      "epoch": 1.5738610216290843,
+      "grad_norm": 0.782588407032302,
+      "learning_rate": 1.0800813456380892e-06,
+      "loss": 0.3589,
+      "num_input_tokens_seen": 393904032,
+      "step": 6840,
+      "train_runtime": 38343.9939,
+      "train_tokens_per_second": 10272.9
+    },
+    {
+      "epoch": 1.5750115048320295,
+      "grad_norm": 0.8633547686811831,
+      "learning_rate": 1.0744784474726915e-06,
+      "loss": 0.3962,
+      "num_input_tokens_seen": 394196984,
+      "step": 6845,
+      "train_runtime": 38373.4757,
+      "train_tokens_per_second": 10272.642
+    },
+    {
+      "epoch": 1.5761619880349746,
+      "grad_norm": 0.8814508720853452,
+      "learning_rate": 1.0688883695889535e-06,
+      "loss": 0.3795,
+      "num_input_tokens_seen": 394464376,
+      "step": 6850,
+      "train_runtime": 38394.8983,
+      "train_tokens_per_second": 10273.875
+    },
+    {
+      "epoch": 1.5773124712379198,
+      "grad_norm": 0.7535912438086815,
+      "learning_rate": 1.0633111302433985e-06,
+      "loss": 0.3816,
+      "num_input_tokens_seen": 394757584,
+      "step": 6855,
+      "train_runtime": 38417.7591,
+      "train_tokens_per_second": 10275.393
+    },
+    {
+      "epoch": 1.578462954440865,
+      "grad_norm": 0.8585854446125815,
+      "learning_rate": 1.0577467476506203e-06,
+      "loss": 0.3911,
+      "num_input_tokens_seen": 395052008,
+      "step": 6860,
+      "train_runtime": 38443.6591,
+      "train_tokens_per_second": 10276.129
+    },
+    {
+      "epoch": 1.5796134376438102,
+      "grad_norm": 0.7605750870373118,
+      "learning_rate": 1.0521952399832236e-06,
+      "loss": 0.3907,
+      "num_input_tokens_seen": 395332272,
+      "step": 6865,
+      "train_runtime": 38475.5049,
+      "train_tokens_per_second": 10274.908
+    },
+    {
+      "epoch": 1.5807639208467557,
+      "grad_norm": 0.9276072318195607,
+      "learning_rate": 1.0466566253717653e-06,
+      "loss": 0.3877,
+      "num_input_tokens_seen": 395595648,
+      "step": 6870,
+      "train_runtime": 38506.6066,
+      "train_tokens_per_second": 10273.449
+    },
+    {
+      "epoch": 1.5819144040497009,
+      "grad_norm": 0.7730779034981271,
+      "learning_rate": 1.041130921904695e-06,
+      "loss": 0.3941,
+      "num_input_tokens_seen": 395887744,
+      "step": 6875,
+      "train_runtime": 38534.3491,
+      "train_tokens_per_second": 10273.633
+    },
+    {
+      "epoch": 1.5830648872526463,
+      "grad_norm": 0.8470954488642445,
+      "learning_rate": 1.0356181476282979e-06,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 396176256,
+      "step": 6880,
+      "train_runtime": 38560.608,
+      "train_tokens_per_second": 10274.118
+    },
+    {
+      "epoch": 1.5842153704555915,
+      "grad_norm": 0.7368723520679286,
+      "learning_rate": 1.0301183205466292e-06,
+      "loss": 0.4023,
+      "num_input_tokens_seen": 396460304,
+      "step": 6885,
+      "train_runtime": 38581.9594,
+      "train_tokens_per_second": 10275.795
+    },
+    {
+      "epoch": 1.5853658536585367,
+      "grad_norm": 0.8073246217755503,
+      "learning_rate": 1.0246314586214635e-06,
+      "loss": 0.3721,
+      "num_input_tokens_seen": 396730864,
+      "step": 6890,
+      "train_runtime": 38606.9049,
+      "train_tokens_per_second": 10276.163
+    },
+    {
+      "epoch": 1.5865163368614819,
+      "grad_norm": 1.0050864307237637,
+      "learning_rate": 1.0191575797722368e-06,
+      "loss": 0.3809,
+      "num_input_tokens_seen": 396995064,
+      "step": 6895,
+      "train_runtime": 38632.8701,
+      "train_tokens_per_second": 10276.096
+    },
+    {
+      "epoch": 1.587666820064427,
+      "grad_norm": 0.8606918912228743,
+      "learning_rate": 1.0136967018759751e-06,
+      "loss": 0.3949,
+      "num_input_tokens_seen": 397280792,
+      "step": 6900,
+      "train_runtime": 38650.7236,
+      "train_tokens_per_second": 10278.741
+    },
+    {
+      "epoch": 1.5888173032673722,
+      "grad_norm": 1.2349771287598408,
+      "learning_rate": 1.0082488427672516e-06,
+      "loss": 0.3759,
+      "num_input_tokens_seen": 397580360,
+      "step": 6905,
+      "train_runtime": 38676.3379,
+      "train_tokens_per_second": 10279.68
+    },
+    {
+      "epoch": 1.5899677864703174,
+      "grad_norm": 0.7828262447919112,
+      "learning_rate": 1.002814020238118e-06,
+      "loss": 0.382,
+      "num_input_tokens_seen": 397888872,
+      "step": 6910,
+      "train_runtime": 38710.0329,
+      "train_tokens_per_second": 10278.701
+    },
+    {
+      "epoch": 1.5911182696732626,
+      "grad_norm": 0.7967089919432391,
+      "learning_rate": 9.973922520380526e-07,
+      "loss": 0.3596,
+      "num_input_tokens_seen": 398156192,
+      "step": 6915,
+      "train_runtime": 38736.1787,
+      "train_tokens_per_second": 10278.665
+    },
+    {
+      "epoch": 1.592268752876208,
+      "grad_norm": 1.0704648531496757,
+      "learning_rate": 9.919835558738989e-07,
+      "loss": 0.4024,
+      "num_input_tokens_seen": 398461520,
+      "step": 6920,
+      "train_runtime": 38758.2257,
+      "train_tokens_per_second": 10280.696
+    },
+    {
+      "epoch": 1.5934192360791533,
+      "grad_norm": 0.7414517247265502,
+      "learning_rate": 9.86587949409808e-07,
+      "loss": 0.3787,
+      "num_input_tokens_seen": 398758704,
+      "step": 6925,
+      "train_runtime": 38784.3181,
+      "train_tokens_per_second": 10281.442
+    },
+    {
+      "epoch": 1.5945697192820985,
+      "grad_norm": 0.7002632307701714,
+      "learning_rate": 9.812054502671835e-07,
+      "loss": 0.35,
+      "num_input_tokens_seen": 399036168,
+      "step": 6930,
+      "train_runtime": 38815.9591,
+      "train_tokens_per_second": 10280.209
+    },
+    {
+      "epoch": 1.5957202024850439,
+      "grad_norm": 0.7478325653323588,
+      "learning_rate": 9.758360760246195e-07,
+      "loss": 0.3716,
+      "num_input_tokens_seen": 399349136,
+      "step": 6935,
+      "train_runtime": 38847.9601,
+      "train_tokens_per_second": 10279.797
+    },
+    {
+      "epoch": 1.596870685687989,
+      "grad_norm": 0.765718537371214,
+      "learning_rate": 9.704798442178465e-07,
+      "loss": 0.3857,
+      "num_input_tokens_seen": 399628008,
+      "step": 6940,
+      "train_runtime": 38875.8824,
+      "train_tokens_per_second": 10279.587
+    },
+    {
+      "epoch": 1.5980211688909343,
+      "grad_norm": 0.8350640153031355,
+      "learning_rate": 9.651367723396749e-07,
+      "loss": 0.3994,
+      "num_input_tokens_seen": 399912096,
+      "step": 6945,
+      "train_runtime": 38902.8965,
+      "train_tokens_per_second": 10279.751
+    },
+    {
+      "epoch": 1.5991716520938795,
+      "grad_norm": 0.8527956330781709,
+      "learning_rate": 9.598068778399356e-07,
+      "loss": 0.3746,
+      "num_input_tokens_seen": 400212776,
+      "step": 6950,
+      "train_runtime": 38936.1252,
+      "train_tokens_per_second": 10278.701
+    },
+    {
+      "epoch": 1.6003221352968247,
+      "grad_norm": 0.7885705026058011,
+      "learning_rate": 9.544901781254228e-07,
+      "loss": 0.3589,
+      "num_input_tokens_seen": 400490760,
+      "step": 6955,
+      "train_runtime": 38959.1603,
+      "train_tokens_per_second": 10279.759
+    },
+    {
+      "epoch": 1.6014726184997699,
+      "grad_norm": 0.9256641921889759,
+      "learning_rate": 9.491866905598396e-07,
+      "loss": 0.387,
+      "num_input_tokens_seen": 400778152,
+      "step": 6960,
+      "train_runtime": 38990.4065,
+      "train_tokens_per_second": 10278.891
+    },
+    {
+      "epoch": 1.602623101702715,
+      "grad_norm": 0.7636849497852651,
+      "learning_rate": 9.438964324637395e-07,
+      "loss": 0.3761,
+      "num_input_tokens_seen": 401085592,
+      "step": 6965,
+      "train_runtime": 39017.4662,
+      "train_tokens_per_second": 10279.642
+    },
+    {
+      "epoch": 1.6037735849056602,
+      "grad_norm": 0.8242652378983875,
+      "learning_rate": 9.386194211144667e-07,
+      "loss": 0.3992,
+      "num_input_tokens_seen": 401383480,
+      "step": 6970,
+      "train_runtime": 39042.8257,
+      "train_tokens_per_second": 10280.595
+    },
+    {
+      "epoch": 1.6049240681086057,
+      "grad_norm": 0.7686224129514398,
+      "learning_rate": 9.333556737461097e-07,
+      "loss": 0.3967,
+      "num_input_tokens_seen": 401687104,
+      "step": 6975,
+      "train_runtime": 39077.0339,
+      "train_tokens_per_second": 10279.365
+    },
+    {
+      "epoch": 1.6060745513115509,
+      "grad_norm": 1.4525775801139236,
+      "learning_rate": 9.281052075494334e-07,
+      "loss": 0.3893,
+      "num_input_tokens_seen": 401972280,
+      "step": 6980,
+      "train_runtime": 39105.3857,
+      "train_tokens_per_second": 10279.205
+    },
+    {
+      "epoch": 1.607225034514496,
+      "grad_norm": 0.7256026856711933,
+      "learning_rate": 9.228680396718309e-07,
+      "loss": 0.3938,
+      "num_input_tokens_seen": 402269928,
+      "step": 6985,
+      "train_runtime": 39131.4434,
+      "train_tokens_per_second": 10279.967
+    },
+    {
+      "epoch": 1.6083755177174415,
+      "grad_norm": 0.7614063159505507,
+      "learning_rate": 9.176441872172603e-07,
+      "loss": 0.3693,
+      "num_input_tokens_seen": 402571400,
+      "step": 6990,
+      "train_runtime": 39166.6756,
+      "train_tokens_per_second": 10278.416
+    },
+    {
+      "epoch": 1.6095260009203867,
+      "grad_norm": 0.9490828185690343,
+      "learning_rate": 9.124336672461981e-07,
+      "loss": 0.387,
+      "num_input_tokens_seen": 402838064,
+      "step": 6995,
+      "train_runtime": 39192.0973,
+      "train_tokens_per_second": 10278.553
+    },
+    {
+      "epoch": 1.6106764841233319,
+      "grad_norm": 0.7401031521694099,
+      "learning_rate": 9.072364967755754e-07,
+      "loss": 0.3799,
+      "num_input_tokens_seen": 403117080,
+      "step": 7000,
+      "train_runtime": 39219.8113,
+      "train_tokens_per_second": 10278.404
+    },
+    {
+      "epoch": 1.611826967326277,
+      "grad_norm": 0.8472377291853119,
+      "learning_rate": 9.020526927787265e-07,
+      "loss": 0.382,
+      "num_input_tokens_seen": 403399912,
+      "step": 7005,
+      "train_runtime": 39254.4485,
+      "train_tokens_per_second": 10276.54
+    },
+    {
+      "epoch": 1.6129774505292223,
+      "grad_norm": 0.8329801268749932,
+      "learning_rate": 8.968822721853326e-07,
+      "loss": 0.4091,
+      "num_input_tokens_seen": 403675040,
+      "step": 7010,
+      "train_runtime": 39277.6495,
+      "train_tokens_per_second": 10277.474
+    },
+    {
+      "epoch": 1.6141279337321675,
+      "grad_norm": 0.789218906978949,
+      "learning_rate": 8.917252518813662e-07,
+      "loss": 0.3722,
+      "num_input_tokens_seen": 403966608,
+      "step": 7015,
+      "train_runtime": 39308.1499,
+      "train_tokens_per_second": 10276.917
+    },
+    {
+      "epoch": 1.6152784169351126,
+      "grad_norm": 0.8085659877206497,
+      "learning_rate": 8.865816487090373e-07,
+      "loss": 0.3921,
+      "num_input_tokens_seen": 404265744,
+      "step": 7020,
+      "train_runtime": 39337.5512,
+      "train_tokens_per_second": 10276.841
+    },
+    {
+      "epoch": 1.6164289001380578,
+      "grad_norm": 0.7914402015961632,
+      "learning_rate": 8.814514794667312e-07,
+      "loss": 0.388,
+      "num_input_tokens_seen": 404557280,
+      "step": 7025,
+      "train_runtime": 39361.6411,
+      "train_tokens_per_second": 10277.958
+    },
+    {
+      "epoch": 1.6175793833410033,
+      "grad_norm": 0.8054705133972674,
+      "learning_rate": 8.763347609089684e-07,
+      "loss": 0.3889,
+      "num_input_tokens_seen": 404833312,
+      "step": 7030,
+      "train_runtime": 39388.5677,
+      "train_tokens_per_second": 10277.939
+    },
+    {
+      "epoch": 1.6187298665439485,
+      "grad_norm": 0.8011944695301956,
+      "learning_rate": 8.71231509746337e-07,
+      "loss": 0.3925,
+      "num_input_tokens_seen": 405148048,
+      "step": 7035,
+      "train_runtime": 39416.3251,
+      "train_tokens_per_second": 10278.686
+    },
+    {
+      "epoch": 1.6198803497468937,
+      "grad_norm": 0.7789394719870619,
+      "learning_rate": 8.661417426454394e-07,
+      "loss": 0.3747,
+      "num_input_tokens_seen": 405456416,
+      "step": 7040,
+      "train_runtime": 39442.3725,
+      "train_tokens_per_second": 10279.717
+    },
+    {
+      "epoch": 1.621030832949839,
+      "grad_norm": 0.7911249867965333,
+      "learning_rate": 8.61065476228844e-07,
+      "loss": 0.3793,
+      "num_input_tokens_seen": 405765648,
+      "step": 7045,
+      "train_runtime": 39474.0626,
+      "train_tokens_per_second": 10279.298
+    },
+    {
+      "epoch": 1.6221813161527843,
+      "grad_norm": 0.8303578783213703,
+      "learning_rate": 8.560027270750276e-07,
+      "loss": 0.3581,
+      "num_input_tokens_seen": 406070976,
+      "step": 7050,
+      "train_runtime": 39504.4528,
+      "train_tokens_per_second": 10279.119
+    },
+    {
+      "epoch": 1.6233317993557295,
+      "grad_norm": 0.889608996752646,
+      "learning_rate": 8.509535117183204e-07,
+      "loss": 0.4021,
+      "num_input_tokens_seen": 406336304,
+      "step": 7055,
+      "train_runtime": 39528.6641,
+      "train_tokens_per_second": 10279.535
+    },
+    {
+      "epoch": 1.6244822825586747,
+      "grad_norm": 0.7539518682765002,
+      "learning_rate": 8.459178466488532e-07,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 406635656,
+      "step": 7060,
+      "train_runtime": 39563.2068,
+      "train_tokens_per_second": 10278.127
+    },
+    {
+      "epoch": 1.6256327657616199,
+      "grad_norm": 0.8364268787394277,
+      "learning_rate": 8.408957483125024e-07,
+      "loss": 0.3626,
+      "num_input_tokens_seen": 406933720,
+      "step": 7065,
+      "train_runtime": 39586.3117,
+      "train_tokens_per_second": 10279.657
+    },
+    {
+      "epoch": 1.626783248964565,
+      "grad_norm": 0.800081055560894,
+      "learning_rate": 8.358872331108392e-07,
+      "loss": 0.3866,
+      "num_input_tokens_seen": 407244776,
+      "step": 7070,
+      "train_runtime": 39618.024,
+      "train_tokens_per_second": 10279.28
+    },
+    {
+      "epoch": 1.6279337321675103,
+      "grad_norm": 0.822704833596379,
+      "learning_rate": 8.308923174010702e-07,
+      "loss": 0.3756,
+      "num_input_tokens_seen": 407493096,
+      "step": 7075,
+      "train_runtime": 39642.776,
+      "train_tokens_per_second": 10279.126
+    },
+    {
+      "epoch": 1.6290842153704554,
+      "grad_norm": 0.780499862994479,
+      "learning_rate": 8.259110174959894e-07,
+      "loss": 0.3926,
+      "num_input_tokens_seen": 407796888,
+      "step": 7080,
+      "train_runtime": 39668.2211,
+      "train_tokens_per_second": 10280.191
+    },
+    {
+      "epoch": 1.6302346985734009,
+      "grad_norm": 0.7473565380844901,
+      "learning_rate": 8.209433496639274e-07,
+      "loss": 0.3947,
+      "num_input_tokens_seen": 408057608,
+      "step": 7085,
+      "train_runtime": 39693.7383,
+      "train_tokens_per_second": 10280.151
+    },
+    {
+      "epoch": 1.631385181776346,
+      "grad_norm": 0.7912978135379808,
+      "learning_rate": 8.15989330128687e-07,
+      "loss": 0.3918,
+      "num_input_tokens_seen": 408314952,
+      "step": 7090,
+      "train_runtime": 39715.8477,
+      "train_tokens_per_second": 10280.907
+    },
+    {
+      "epoch": 1.6325356649792913,
+      "grad_norm": 0.9187389832239469,
+      "learning_rate": 8.110489750695022e-07,
+      "loss": 0.3778,
+      "num_input_tokens_seen": 408595328,
+      "step": 7095,
+      "train_runtime": 39741.1633,
+      "train_tokens_per_second": 10281.413
+    },
+    {
+      "epoch": 1.6336861481822367,
+      "grad_norm": 0.8837223488304119,
+      "learning_rate": 8.061223006209779e-07,
+      "loss": 0.3975,
+      "num_input_tokens_seen": 408875768,
+      "step": 7100,
+      "train_runtime": 39763.9917,
+      "train_tokens_per_second": 10282.563
+    },
+    {
+      "epoch": 1.6348366313851819,
+      "grad_norm": 0.6822166975653342,
+      "learning_rate": 8.012093228730416e-07,
+      "loss": 0.3862,
+      "num_input_tokens_seen": 409167752,
+      "step": 7105,
+      "train_runtime": 39793.0546,
+      "train_tokens_per_second": 10282.391
+    },
+    {
+      "epoch": 1.635987114588127,
+      "grad_norm": 0.7772635405883309,
+      "learning_rate": 7.963100578708882e-07,
+      "loss": 0.3952,
+      "num_input_tokens_seen": 409470192,
+      "step": 7110,
+      "train_runtime": 39818.2916,
+      "train_tokens_per_second": 10283.47
+    },
+    {
+      "epoch": 1.6371375977910723,
+      "grad_norm": 0.8012613022133598,
+      "learning_rate": 7.91424521614928e-07,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 409740640,
+      "step": 7115,
+      "train_runtime": 39843.0433,
+      "train_tokens_per_second": 10283.869
+    },
+    {
+      "epoch": 1.6382880809940175,
+      "grad_norm": 0.7954364708858213,
+      "learning_rate": 7.865527300607356e-07,
+      "loss": 0.3843,
+      "num_input_tokens_seen": 410037904,
+      "step": 7120,
+      "train_runtime": 39868.6193,
+      "train_tokens_per_second": 10284.728
+    },
+    {
+      "epoch": 1.6394385641969627,
+      "grad_norm": 0.793380558057108,
+      "learning_rate": 7.816946991189983e-07,
+      "loss": 0.3859,
+      "num_input_tokens_seen": 410320024,
+      "step": 7125,
+      "train_runtime": 39896.5873,
+      "train_tokens_per_second": 10284.59
+    },
+    {
+      "epoch": 1.6405890473999079,
+      "grad_norm": 0.8439535943742152,
+      "learning_rate": 7.768504446554581e-07,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 410592120,
+      "step": 7130,
+      "train_runtime": 39922.0454,
+      "train_tokens_per_second": 10284.847
+    },
+    {
+      "epoch": 1.641739530602853,
+      "grad_norm": 0.8025282828448552,
+      "learning_rate": 7.720199824908692e-07,
+      "loss": 0.417,
+      "num_input_tokens_seen": 410889704,
+      "step": 7135,
+      "train_runtime": 39943.0281,
+      "train_tokens_per_second": 10286.894
+    },
+    {
+      "epoch": 1.6428900138057985,
+      "grad_norm": 0.7933165867424761,
+      "learning_rate": 7.672033284009428e-07,
+      "loss": 0.3884,
+      "num_input_tokens_seen": 411176968,
+      "step": 7140,
+      "train_runtime": 39967.8633,
+      "train_tokens_per_second": 10287.69
+    },
+    {
+      "epoch": 1.6440404970087437,
+      "grad_norm": 0.8300838941511794,
+      "learning_rate": 7.624004981162902e-07,
+      "loss": 0.3735,
+      "num_input_tokens_seen": 411461400,
+      "step": 7145,
+      "train_runtime": 39999.537,
+      "train_tokens_per_second": 10286.654
+    },
+    {
+      "epoch": 1.6451909802116889,
+      "grad_norm": 0.8056498903939717,
+      "learning_rate": 7.576115073223778e-07,
+      "loss": 0.3812,
+      "num_input_tokens_seen": 411756856,
+      "step": 7150,
+      "train_runtime": 40030.7513,
+      "train_tokens_per_second": 10286.014
+    },
+    {
+      "epoch": 1.6463414634146343,
+      "grad_norm": 0.7516259258902676,
+      "learning_rate": 7.528363716594745e-07,
+      "loss": 0.3808,
+      "num_input_tokens_seen": 412054232,
+      "step": 7155,
+      "train_runtime": 40060.4094,
+      "train_tokens_per_second": 10285.822
+    },
+    {
+      "epoch": 1.6474919466175795,
+      "grad_norm": 0.7702926997413523,
+      "learning_rate": 7.480751067226e-07,
+      "loss": 0.4035,
+      "num_input_tokens_seen": 412346120,
+      "step": 7160,
+      "train_runtime": 40089.356,
+      "train_tokens_per_second": 10285.676
+    },
+    {
+      "epoch": 1.6486424298205247,
+      "grad_norm": 0.7579876516064216,
+      "learning_rate": 7.433277280614726e-07,
+      "loss": 0.3712,
+      "num_input_tokens_seen": 412646544,
+      "step": 7165,
+      "train_runtime": 40119.0232,
+      "train_tokens_per_second": 10285.558
+    },
+    {
+      "epoch": 1.6497929130234699,
+      "grad_norm": 0.7575718959776322,
+      "learning_rate": 7.385942511804611e-07,
+      "loss": 0.3914,
+      "num_input_tokens_seen": 412921616,
+      "step": 7170,
+      "train_runtime": 40148.0609,
+      "train_tokens_per_second": 10284.97
+    },
+    {
+      "epoch": 1.650943396226415,
+      "grad_norm": 0.7258096590879551,
+      "learning_rate": 7.33874691538533e-07,
+      "loss": 0.3839,
+      "num_input_tokens_seen": 413214304,
+      "step": 7175,
+      "train_runtime": 40170.6708,
+      "train_tokens_per_second": 10286.468
+    },
+    {
+      "epoch": 1.6520938794293603,
+      "grad_norm": 0.9599739829954606,
+      "learning_rate": 7.291690645492011e-07,
+      "loss": 0.3866,
+      "num_input_tokens_seen": 413509904,
+      "step": 7180,
+      "train_runtime": 40199.2274,
+      "train_tokens_per_second": 10286.514
+    },
+    {
+      "epoch": 1.6532443626323055,
+      "grad_norm": 0.8631273871934643,
+      "learning_rate": 7.244773855804788e-07,
+      "loss": 0.4125,
+      "num_input_tokens_seen": 413784368,
+      "step": 7185,
+      "train_runtime": 40225.4768,
+      "train_tokens_per_second": 10286.624
+    },
+    {
+      "epoch": 1.6543948458352506,
+      "grad_norm": 0.7617090035485522,
+      "learning_rate": 7.197996699548255e-07,
+      "loss": 0.3823,
+      "num_input_tokens_seen": 414075192,
+      "step": 7190,
+      "train_runtime": 40262.4099,
+      "train_tokens_per_second": 10284.412
+    },
+    {
+      "epoch": 1.655545329038196,
+      "grad_norm": 0.7622774663817032,
+      "learning_rate": 7.151359329490981e-07,
+      "loss": 0.3736,
+      "num_input_tokens_seen": 414374200,
+      "step": 7195,
+      "train_runtime": 40293.4767,
+      "train_tokens_per_second": 10283.903
+    },
+    {
+      "epoch": 1.6566958122411413,
+      "grad_norm": 0.8452180562470519,
+      "learning_rate": 7.104861897945014e-07,
+      "loss": 0.3715,
+      "num_input_tokens_seen": 414647872,
+      "step": 7200,
+      "train_runtime": 40319.8826,
+      "train_tokens_per_second": 10283.955
+    },
+    {
+      "epoch": 1.6578462954440867,
+      "grad_norm": 0.7851088962689995,
+      "learning_rate": 7.058504556765383e-07,
+      "loss": 0.3875,
+      "num_input_tokens_seen": 414944960,
+      "step": 7205,
+      "train_runtime": 40348.2754,
+      "train_tokens_per_second": 10284.082
+    },
+    {
+      "epoch": 1.6589967786470319,
+      "grad_norm": 0.7451312998945385,
+      "learning_rate": 7.012287457349598e-07,
+      "loss": 0.3868,
+      "num_input_tokens_seen": 415228608,
+      "step": 7210,
+      "train_runtime": 40376.4871,
+      "train_tokens_per_second": 10283.921
+    },
+    {
+      "epoch": 1.660147261849977,
+      "grad_norm": 0.7351416368255037,
+      "learning_rate": 6.966210750637121e-07,
+      "loss": 0.3717,
+      "num_input_tokens_seen": 415549136,
+      "step": 7215,
+      "train_runtime": 40402.1642,
+      "train_tokens_per_second": 10285.319
+    },
+    {
+      "epoch": 1.6612977450529223,
+      "grad_norm": 1.2386689046075936,
+      "learning_rate": 6.920274587108972e-07,
+      "loss": 0.3555,
+      "num_input_tokens_seen": 415852240,
+      "step": 7220,
+      "train_runtime": 40423.3785,
+      "train_tokens_per_second": 10287.419
+    },
+    {
+      "epoch": 1.6624482282558675,
+      "grad_norm": 0.8366526900855967,
+      "learning_rate": 6.874479116787142e-07,
+      "loss": 0.4116,
+      "num_input_tokens_seen": 416132320,
+      "step": 7225,
+      "train_runtime": 40454.2863,
+      "train_tokens_per_second": 10286.483
+    },
+    {
+      "epoch": 1.6635987114588127,
+      "grad_norm": 0.715639133577145,
+      "learning_rate": 6.828824489234109e-07,
+      "loss": 0.3808,
+      "num_input_tokens_seen": 416432968,
+      "step": 7230,
+      "train_runtime": 40484.6262,
+      "train_tokens_per_second": 10286.2
+    },
+    {
+      "epoch": 1.6647491946617579,
+      "grad_norm": 0.7719378252787551,
+      "learning_rate": 6.783310853552411e-07,
+      "loss": 0.3898,
+      "num_input_tokens_seen": 416712560,
+      "step": 7235,
+      "train_runtime": 40511.858,
+      "train_tokens_per_second": 10286.187
+    },
+    {
+      "epoch": 1.665899677864703,
+      "grad_norm": 0.8245206911072641,
+      "learning_rate": 6.737938358384122e-07,
+      "loss": 0.3878,
+      "num_input_tokens_seen": 416987968,
+      "step": 7240,
+      "train_runtime": 40541.6948,
+      "train_tokens_per_second": 10285.41
+    },
+    {
+      "epoch": 1.6670501610676483,
+      "grad_norm": 0.7397842922420265,
+      "learning_rate": 6.692707151910344e-07,
+      "loss": 0.3916,
+      "num_input_tokens_seen": 417279232,
+      "step": 7245,
+      "train_runtime": 40574.6403,
+      "train_tokens_per_second": 10284.237
+    },
+    {
+      "epoch": 1.6682006442705937,
+      "grad_norm": 0.8354674467109593,
+      "learning_rate": 6.647617381850774e-07,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 417559192,
+      "step": 7250,
+      "train_runtime": 40600.4644,
+      "train_tokens_per_second": 10284.592
+    },
+    {
+      "epoch": 1.6693511274735389,
+      "grad_norm": 0.8407466165490526,
+      "learning_rate": 6.60266919546318e-07,
+      "loss": 0.3952,
+      "num_input_tokens_seen": 417843728,
+      "step": 7255,
+      "train_runtime": 40627.9191,
+      "train_tokens_per_second": 10284.645
+    },
+    {
+      "epoch": 1.6705016106764843,
+      "grad_norm": 0.9765314333483613,
+      "learning_rate": 6.557862739542942e-07,
+      "loss": 0.4233,
+      "num_input_tokens_seen": 418119800,
+      "step": 7260,
+      "train_runtime": 40663.3536,
+      "train_tokens_per_second": 10282.472
+    },
+    {
+      "epoch": 1.6716520938794295,
+      "grad_norm": 0.7884020705475865,
+      "learning_rate": 6.513198160422546e-07,
+      "loss": 0.3818,
+      "num_input_tokens_seen": 418412144,
+      "step": 7265,
+      "train_runtime": 40689.961,
+      "train_tokens_per_second": 10282.933
+    },
+    {
+      "epoch": 1.6728025770823747,
+      "grad_norm": 0.8268555122109376,
+      "learning_rate": 6.468675603971137e-07,
+      "loss": 0.3906,
+      "num_input_tokens_seen": 418706000,
+      "step": 7270,
+      "train_runtime": 40718.1741,
+      "train_tokens_per_second": 10283.025
+    },
+    {
+      "epoch": 1.6739530602853199,
+      "grad_norm": 0.7978884686354187,
+      "learning_rate": 6.424295215594057e-07,
+      "loss": 0.3746,
+      "num_input_tokens_seen": 418992896,
+      "step": 7275,
+      "train_runtime": 40749.4768,
+      "train_tokens_per_second": 10282.166
+    },
+    {
+      "epoch": 1.675103543488265,
+      "grad_norm": 0.7366578825758895,
+      "learning_rate": 6.38005714023232e-07,
+      "loss": 0.3683,
+      "num_input_tokens_seen": 419287432,
+      "step": 7280,
+      "train_runtime": 40780.9517,
+      "train_tokens_per_second": 10281.453
+    },
+    {
+      "epoch": 1.6762540266912103,
+      "grad_norm": 0.74029970560341,
+      "learning_rate": 6.335961522362144e-07,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 419579384,
+      "step": 7285,
+      "train_runtime": 40802.658,
+      "train_tokens_per_second": 10283.139
+    },
+    {
+      "epoch": 1.6774045098941555,
+      "grad_norm": 0.7697441120225352,
+      "learning_rate": 6.292008505994524e-07,
+      "loss": 0.4003,
+      "num_input_tokens_seen": 419859960,
+      "step": 7290,
+      "train_runtime": 40831.1922,
+      "train_tokens_per_second": 10282.824
+    },
+    {
+      "epoch": 1.6785549930971007,
+      "grad_norm": 0.8289671228864346,
+      "learning_rate": 6.248198234674735e-07,
+      "loss": 0.3839,
+      "num_input_tokens_seen": 420141200,
+      "step": 7295,
+      "train_runtime": 40856.363,
+      "train_tokens_per_second": 10283.373
+    },
+    {
+      "epoch": 1.679705476300046,
+      "grad_norm": 0.8019159955668383,
+      "learning_rate": 6.204530851481861e-07,
+      "loss": 0.3839,
+      "num_input_tokens_seen": 420452544,
+      "step": 7300,
+      "train_runtime": 40880.9733,
+      "train_tokens_per_second": 10284.798
+    },
+    {
+      "epoch": 1.6808559595029913,
+      "grad_norm": 0.710562060517948,
+      "learning_rate": 6.161006499028316e-07,
+      "loss": 0.3822,
+      "num_input_tokens_seen": 420754304,
+      "step": 7305,
+      "train_runtime": 40908.6824,
+      "train_tokens_per_second": 10285.208
+    },
+    {
+      "epoch": 1.6820064427059365,
+      "grad_norm": 0.8647374530255871,
+      "learning_rate": 6.11762531945942e-07,
+      "loss": 0.3798,
+      "num_input_tokens_seen": 421018320,
+      "step": 7310,
+      "train_runtime": 40934.0991,
+      "train_tokens_per_second": 10285.271
+    },
+    {
+      "epoch": 1.6831569259088819,
+      "grad_norm": 0.8063428840575269,
+      "learning_rate": 6.074387454452891e-07,
+      "loss": 0.3695,
+      "num_input_tokens_seen": 421299296,
+      "step": 7315,
+      "train_runtime": 40962.3765,
+      "train_tokens_per_second": 10285.031
+    },
+    {
+      "epoch": 1.684307409111827,
+      "grad_norm": 0.8250795588923827,
+      "learning_rate": 6.031293045218384e-07,
+      "loss": 0.3996,
+      "num_input_tokens_seen": 421591872,
+      "step": 7320,
+      "train_runtime": 40992.6998,
+      "train_tokens_per_second": 10284.56
+    },
+    {
+      "epoch": 1.6854578923147723,
+      "grad_norm": 0.7495627185612652,
+      "learning_rate": 5.988342232497058e-07,
+      "loss": 0.3773,
+      "num_input_tokens_seen": 421877832,
+      "step": 7325,
+      "train_runtime": 41030.6838,
+      "train_tokens_per_second": 10282.008
+    },
+    {
+      "epoch": 1.6866083755177175,
+      "grad_norm": 0.8291085866676273,
+      "learning_rate": 5.945535156561144e-07,
+      "loss": 0.3777,
+      "num_input_tokens_seen": 422140264,
+      "step": 7330,
+      "train_runtime": 41059.6233,
+      "train_tokens_per_second": 10281.153
+    },
+    {
+      "epoch": 1.6877588587206627,
+      "grad_norm": 0.6919740247055045,
+      "learning_rate": 5.902871957213374e-07,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 422426456,
+      "step": 7335,
+      "train_runtime": 41092.7064,
+      "train_tokens_per_second": 10279.84
+    },
+    {
+      "epoch": 1.6889093419236079,
+      "grad_norm": 0.8419721197054296,
+      "learning_rate": 5.860352773786632e-07,
+      "loss": 0.3622,
+      "num_input_tokens_seen": 422711944,
+      "step": 7340,
+      "train_runtime": 41124.2641,
+      "train_tokens_per_second": 10278.894
+    },
+    {
+      "epoch": 1.690059825126553,
+      "grad_norm": 0.9344009365479368,
+      "learning_rate": 5.817977745143477e-07,
+      "loss": 0.3693,
+      "num_input_tokens_seen": 422982792,
+      "step": 7345,
+      "train_runtime": 41146.7607,
+      "train_tokens_per_second": 10279.856
+    },
+    {
+      "epoch": 1.6912103083294983,
+      "grad_norm": 0.8237498828805256,
+      "learning_rate": 5.775747009675648e-07,
+      "loss": 0.3794,
+      "num_input_tokens_seen": 423245696,
+      "step": 7350,
+      "train_runtime": 41172.4426,
+      "train_tokens_per_second": 10279.83
+    },
+    {
+      "epoch": 1.6923607915324437,
+      "grad_norm": 0.7553169751756835,
+      "learning_rate": 5.733660705303651e-07,
+      "loss": 0.3838,
+      "num_input_tokens_seen": 423548120,
+      "step": 7355,
+      "train_runtime": 41197.521,
+      "train_tokens_per_second": 10280.913
+    },
+    {
+      "epoch": 1.6935112747353889,
+      "grad_norm": 0.8782001430613734,
+      "learning_rate": 5.691718969476312e-07,
+      "loss": 0.3802,
+      "num_input_tokens_seen": 423833184,
+      "step": 7360,
+      "train_runtime": 41227.0196,
+      "train_tokens_per_second": 10280.471
+    },
+    {
+      "epoch": 1.694661757938334,
+      "grad_norm": 0.8661191294423588,
+      "learning_rate": 5.649921939170295e-07,
+      "loss": 0.3953,
+      "num_input_tokens_seen": 424095496,
+      "step": 7365,
+      "train_runtime": 41256.9837,
+      "train_tokens_per_second": 10279.363
+    },
+    {
+      "epoch": 1.6958122411412795,
+      "grad_norm": 0.7361437933945663,
+      "learning_rate": 5.608269750889673e-07,
+      "loss": 0.3815,
+      "num_input_tokens_seen": 424396912,
+      "step": 7370,
+      "train_runtime": 41285.6461,
+      "train_tokens_per_second": 10279.527
+    },
+    {
+      "epoch": 1.6969627243442247,
+      "grad_norm": 0.7880124575681219,
+      "learning_rate": 5.566762540665483e-07,
+      "loss": 0.3876,
+      "num_input_tokens_seen": 424685864,
+      "step": 7375,
+      "train_runtime": 41314.1805,
+      "train_tokens_per_second": 10279.421
+    },
+    {
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.7979837069505088,
+      "learning_rate": 5.52540044405529e-07,
+      "loss": 0.384,
+      "num_input_tokens_seen": 424958136,
+      "step": 7380,
+      "train_runtime": 41343.1689,
+      "train_tokens_per_second": 10278.799
+    },
+    {
+      "epoch": 1.699263690750115,
+      "grad_norm": 0.7636141242484338,
+      "learning_rate": 5.484183596142739e-07,
+      "loss": 0.3999,
+      "num_input_tokens_seen": 425240360,
+      "step": 7385,
+      "train_runtime": 41376.2019,
+      "train_tokens_per_second": 10277.414
+    },
+    {
+      "epoch": 1.7004141739530603,
+      "grad_norm": 0.7097447383958857,
+      "learning_rate": 5.443112131537093e-07,
+      "loss": 0.3563,
+      "num_input_tokens_seen": 425534056,
+      "step": 7390,
+      "train_runtime": 41404.3036,
+      "train_tokens_per_second": 10277.532
+    },
+    {
+      "epoch": 1.7015646571560055,
+      "grad_norm": 0.6827205802161698,
+      "learning_rate": 5.402186184372826e-07,
+      "loss": 0.3745,
+      "num_input_tokens_seen": 425798816,
+      "step": 7395,
+      "train_runtime": 41434.2157,
+      "train_tokens_per_second": 10276.502
+    },
+    {
+      "epoch": 1.7027151403589507,
+      "grad_norm": 0.7576126759047791,
+      "learning_rate": 5.361405888309174e-07,
+      "loss": 0.3981,
+      "num_input_tokens_seen": 426073200,
+      "step": 7400,
+      "train_runtime": 41462.1096,
+      "train_tokens_per_second": 10276.207
+    },
+    {
+      "epoch": 1.7038656235618959,
+      "grad_norm": 0.7644625363176181,
+      "learning_rate": 5.320771376529643e-07,
+      "loss": 0.4016,
+      "num_input_tokens_seen": 426362392,
+      "step": 7405,
+      "train_runtime": 41490.0371,
+      "train_tokens_per_second": 10276.26
+    },
+    {
+      "epoch": 1.7050161067648413,
+      "grad_norm": 0.9594463410322395,
+      "learning_rate": 5.280282781741702e-07,
+      "loss": 0.3943,
+      "num_input_tokens_seen": 426611136,
+      "step": 7410,
+      "train_runtime": 41515.4587,
+      "train_tokens_per_second": 10275.959
+    },
+    {
+      "epoch": 1.7061665899677865,
+      "grad_norm": 0.764527023796408,
+      "learning_rate": 5.239940236176233e-07,
+      "loss": 0.3853,
+      "num_input_tokens_seen": 426912728,
+      "step": 7415,
+      "train_runtime": 41547.1556,
+      "train_tokens_per_second": 10275.378
+    },
+    {
+      "epoch": 1.7073170731707317,
+      "grad_norm": 0.7702190035883508,
+      "learning_rate": 5.199743871587143e-07,
+      "loss": 0.3901,
+      "num_input_tokens_seen": 427212928,
+      "step": 7420,
+      "train_runtime": 41573.3179,
+      "train_tokens_per_second": 10276.133
+    },
+    {
+      "epoch": 1.708467556373677,
+      "grad_norm": 0.7234233447435587,
+      "learning_rate": 5.159693819250911e-07,
+      "loss": 0.3859,
+      "num_input_tokens_seen": 427509216,
+      "step": 7425,
+      "train_runtime": 41606.1175,
+      "train_tokens_per_second": 10275.153
+    },
+    {
+      "epoch": 1.7096180395766223,
+      "grad_norm": 0.7725130654665208,
+      "learning_rate": 5.119790209966191e-07,
+      "loss": 0.3804,
+      "num_input_tokens_seen": 427781152,
+      "step": 7430,
+      "train_runtime": 41634.0954,
+      "train_tokens_per_second": 10274.78
+    },
+    {
+      "epoch": 1.7107685227795675,
+      "grad_norm": 0.7917738899200925,
+      "learning_rate": 5.080033174053412e-07,
+      "loss": 0.3826,
+      "num_input_tokens_seen": 428053144,
+      "step": 7435,
+      "train_runtime": 41670.6082,
+      "train_tokens_per_second": 10272.304
+    },
+    {
+      "epoch": 1.7119190059825127,
+      "grad_norm": 0.8309316268242334,
+      "learning_rate": 5.040422841354248e-07,
+      "loss": 0.3805,
+      "num_input_tokens_seen": 428347976,
+      "step": 7440,
+      "train_runtime": 41700.6305,
+      "train_tokens_per_second": 10271.978
+    },
+    {
+      "epoch": 1.7130694891854579,
+      "grad_norm": 0.8622787910294952,
+      "learning_rate": 5.000959341231287e-07,
+      "loss": 0.4156,
+      "num_input_tokens_seen": 428659200,
+      "step": 7445,
+      "train_runtime": 41729.0718,
+      "train_tokens_per_second": 10272.436
+    },
+    {
+      "epoch": 1.714219972388403,
+      "grad_norm": 0.7365437282011078,
+      "learning_rate": 4.961642802567578e-07,
+      "loss": 0.387,
+      "num_input_tokens_seen": 428972936,
+      "step": 7450,
+      "train_runtime": 41756.7827,
+      "train_tokens_per_second": 10273.132
+    },
+    {
+      "epoch": 1.7153704555913483,
+      "grad_norm": 0.744131914948768,
+      "learning_rate": 4.922473353766232e-07,
+      "loss": 0.3722,
+      "num_input_tokens_seen": 429280584,
+      "step": 7455,
+      "train_runtime": 41789.9222,
+      "train_tokens_per_second": 10272.347
+    },
+    {
+      "epoch": 1.7165209387942935,
+      "grad_norm": 0.7678695465936678,
+      "learning_rate": 4.883451122749921e-07,
+      "loss": 0.3787,
+      "num_input_tokens_seen": 429577104,
+      "step": 7460,
+      "train_runtime": 41818.6471,
+      "train_tokens_per_second": 10272.382
+    },
+    {
+      "epoch": 1.7176714219972389,
+      "grad_norm": 0.828111070390171,
+      "learning_rate": 4.844576236960597e-07,
+      "loss": 0.3911,
+      "num_input_tokens_seen": 429861992,
+      "step": 7465,
+      "train_runtime": 41854.398,
+      "train_tokens_per_second": 10270.414
+    },
+    {
+      "epoch": 1.718821905200184,
+      "grad_norm": 0.8149063347928397,
+      "learning_rate": 4.805848823358966e-07,
+      "loss": 0.3857,
+      "num_input_tokens_seen": 430142552,
+      "step": 7470,
+      "train_runtime": 41887.5677,
+      "train_tokens_per_second": 10268.979
+    },
+    {
+      "epoch": 1.7199723884031293,
+      "grad_norm": 0.7670895050981109,
+      "learning_rate": 4.7672690084240856e-07,
+      "loss": 0.404,
+      "num_input_tokens_seen": 430448344,
+      "step": 7475,
+      "train_runtime": 41913.5634,
+      "train_tokens_per_second": 10269.906
+    },
+    {
+      "epoch": 1.7211228716060747,
+      "grad_norm": 0.7508117093664783,
+      "learning_rate": 4.728836918153007e-07,
+      "loss": 0.3976,
+      "num_input_tokens_seen": 430738544,
+      "step": 7480,
+      "train_runtime": 41942.0391,
+      "train_tokens_per_second": 10269.852
+    },
+    {
+      "epoch": 1.7222733548090199,
+      "grad_norm": 0.8002075698817599,
+      "learning_rate": 4.690552678060317e-07,
+      "loss": 0.4172,
+      "num_input_tokens_seen": 431049600,
+      "step": 7485,
+      "train_runtime": 41976.5377,
+      "train_tokens_per_second": 10268.822
+    },
+    {
+      "epoch": 1.723423838011965,
+      "grad_norm": 0.7545596718803462,
+      "learning_rate": 4.652416413177746e-07,
+      "loss": 0.4001,
+      "num_input_tokens_seen": 431325024,
+      "step": 7490,
+      "train_runtime": 42010.4704,
+      "train_tokens_per_second": 10267.084
+    },
+    {
+      "epoch": 1.7245743212149103,
+      "grad_norm": 0.9430866576256823,
+      "learning_rate": 4.614428248053743e-07,
+      "loss": 0.4071,
+      "num_input_tokens_seen": 431591048,
+      "step": 7495,
+      "train_runtime": 42038.8911,
+      "train_tokens_per_second": 10266.471
+    },
+    {
+      "epoch": 1.7257248044178555,
+      "grad_norm": 0.9362270968381525,
+      "learning_rate": 4.576588306753099e-07,
+      "loss": 0.4159,
+      "num_input_tokens_seen": 431896184,
+      "step": 7500,
+      "train_runtime": 42069.6306,
+      "train_tokens_per_second": 10266.222
+    },
+    {
+      "epoch": 1.7268752876208007,
+      "grad_norm": 0.8784269388147379,
+      "learning_rate": 4.538896712856528e-07,
+      "loss": 0.3809,
+      "num_input_tokens_seen": 432173736,
+      "step": 7505,
+      "train_runtime": 42100.9243,
+      "train_tokens_per_second": 10265.184
+    },
+    {
+      "epoch": 1.7280257708237459,
+      "grad_norm": 0.8316236794315593,
+      "learning_rate": 4.501353589460217e-07,
+      "loss": 0.3564,
+      "num_input_tokens_seen": 432461392,
+      "step": 7510,
+      "train_runtime": 42129.9373,
+      "train_tokens_per_second": 10264.943
+    },
+    {
+      "epoch": 1.729176254026691,
+      "grad_norm": 0.7367201561519939,
+      "learning_rate": 4.4639590591754986e-07,
+      "loss": 0.3763,
+      "num_input_tokens_seen": 432752240,
+      "step": 7515,
+      "train_runtime": 42163.4697,
+      "train_tokens_per_second": 10263.677
+    },
+    {
+      "epoch": 1.7303267372296365,
+      "grad_norm": 0.8610353422551513,
+      "learning_rate": 4.4267132441284446e-07,
+      "loss": 0.3827,
+      "num_input_tokens_seen": 433057416,
+      "step": 7520,
+      "train_runtime": 42189.568,
+      "train_tokens_per_second": 10264.562
+    },
+    {
+      "epoch": 1.7314772204325817,
+      "grad_norm": 0.8424189666629835,
+      "learning_rate": 4.38961626595939e-07,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 433340568,
+      "step": 7525,
+      "train_runtime": 42229.347,
+      "train_tokens_per_second": 10261.598
+    },
+    {
+      "epoch": 1.7326277036355269,
+      "grad_norm": 0.8019937518652265,
+      "learning_rate": 4.3526682458226053e-07,
+      "loss": 0.3804,
+      "num_input_tokens_seen": 433622592,
+      "step": 7530,
+      "train_runtime": 42257.0208,
+      "train_tokens_per_second": 10261.551
+    },
+    {
+      "epoch": 1.7337781868384723,
+      "grad_norm": 0.7735248604757327,
+      "learning_rate": 4.315869304385889e-07,
+      "loss": 0.4027,
+      "num_input_tokens_seen": 433904736,
+      "step": 7535,
+      "train_runtime": 42288.7361,
+      "train_tokens_per_second": 10260.527
+    },
+    {
+      "epoch": 1.7349286700414175,
+      "grad_norm": 0.8582663594423506,
+      "learning_rate": 4.279219561830161e-07,
+      "loss": 0.3763,
+      "num_input_tokens_seen": 434214904,
+      "step": 7540,
+      "train_runtime": 42313.3795,
+      "train_tokens_per_second": 10261.882
+    },
+    {
+      "epoch": 1.7360791532443627,
+      "grad_norm": 0.8293299183904065,
+      "learning_rate": 4.242719137849077e-07,
+      "loss": 0.4051,
+      "num_input_tokens_seen": 434498304,
+      "step": 7545,
+      "train_runtime": 42337.7437,
+      "train_tokens_per_second": 10262.67
+    },
+    {
+      "epoch": 1.7372296364473079,
+      "grad_norm": 0.8519998563935477,
+      "learning_rate": 4.2063681516486377e-07,
+      "loss": 0.3789,
+      "num_input_tokens_seen": 434763344,
+      "step": 7550,
+      "train_runtime": 42366.2512,
+      "train_tokens_per_second": 10262.021
+    },
+    {
+      "epoch": 1.738380119650253,
+      "grad_norm": 0.8258407374994876,
+      "learning_rate": 4.1701667219468e-07,
+      "loss": 0.3799,
+      "num_input_tokens_seen": 435045928,
+      "step": 7555,
+      "train_runtime": 42395.9569,
+      "train_tokens_per_second": 10261.496
+    },
+    {
+      "epoch": 1.7395306028531983,
+      "grad_norm": 0.7540785800602019,
+      "learning_rate": 4.1341149669730687e-07,
+      "loss": 0.4099,
+      "num_input_tokens_seen": 435358312,
+      "step": 7560,
+      "train_runtime": 42421.2138,
+      "train_tokens_per_second": 10262.75
+    },
+    {
+      "epoch": 1.7406810860561435,
+      "grad_norm": 0.6999873785142079,
+      "learning_rate": 4.098213004468149e-07,
+      "loss": 0.3725,
+      "num_input_tokens_seen": 435631016,
+      "step": 7565,
+      "train_runtime": 42448.1261,
+      "train_tokens_per_second": 10262.668
+    },
+    {
+      "epoch": 1.7418315692590887,
+      "grad_norm": 0.8306355205604321,
+      "learning_rate": 4.0624609516835335e-07,
+      "loss": 0.3674,
+      "num_input_tokens_seen": 435877664,
+      "step": 7570,
+      "train_runtime": 42473.2599,
+      "train_tokens_per_second": 10262.402
+    },
+    {
+      "epoch": 1.742982052462034,
+      "grad_norm": 0.7024538940887601,
+      "learning_rate": 4.0268589253811494e-07,
+      "loss": 0.3716,
+      "num_input_tokens_seen": 436171656,
+      "step": 7575,
+      "train_runtime": 42499.9357,
+      "train_tokens_per_second": 10262.878
+    },
+    {
+      "epoch": 1.7441325356649793,
+      "grad_norm": 0.8275355209687048,
+      "learning_rate": 3.9914070418329123e-07,
+      "loss": 0.3668,
+      "num_input_tokens_seen": 436476432,
+      "step": 7580,
+      "train_runtime": 42526.1827,
+      "train_tokens_per_second": 10263.711
+    },
+    {
+      "epoch": 1.7452830188679245,
+      "grad_norm": 0.8472293163556103,
+      "learning_rate": 3.956105416820427e-07,
+      "loss": 0.3943,
+      "num_input_tokens_seen": 436755304,
+      "step": 7585,
+      "train_runtime": 42559.9007,
+      "train_tokens_per_second": 10262.132
+    },
+    {
+      "epoch": 1.74643350207087,
+      "grad_norm": 0.7223250562936765,
+      "learning_rate": 3.920954165634555e-07,
+      "loss": 0.3875,
+      "num_input_tokens_seen": 437057640,
+      "step": 7590,
+      "train_runtime": 42589.096,
+      "train_tokens_per_second": 10262.196
+    },
+    {
+      "epoch": 1.747583985273815,
+      "grad_norm": 0.7368299463391748,
+      "learning_rate": 3.885953403075055e-07,
+      "loss": 0.3889,
+      "num_input_tokens_seen": 437347256,
+      "step": 7595,
+      "train_runtime": 42615.65,
+      "train_tokens_per_second": 10262.597
+    },
+    {
+      "epoch": 1.7487344684767603,
+      "grad_norm": 0.8225450697573361,
+      "learning_rate": 3.851103243450216e-07,
+      "loss": 0.3801,
+      "num_input_tokens_seen": 437662040,
+      "step": 7600,
+      "train_runtime": 42638.5151,
+      "train_tokens_per_second": 10264.477
+    },
+    {
+      "epoch": 1.7498849516797055,
+      "grad_norm": 0.7797186419107279,
+      "learning_rate": 3.816403800576468e-07,
+      "loss": 0.3742,
+      "num_input_tokens_seen": 437981344,
+      "step": 7605,
+      "train_runtime": 42672.2528,
+      "train_tokens_per_second": 10263.844
+    },
+    {
+      "epoch": 1.7510354348826507,
+      "grad_norm": 0.759931117039561,
+      "learning_rate": 3.7818551877780327e-07,
+      "loss": 0.403,
+      "num_input_tokens_seen": 438268120,
+      "step": 7610,
+      "train_runtime": 42697.3575,
+      "train_tokens_per_second": 10264.526
+    },
+    {
+      "epoch": 1.7521859180855959,
+      "grad_norm": 0.8087257561966026,
+      "learning_rate": 3.74745751788651e-07,
+      "loss": 0.3857,
+      "num_input_tokens_seen": 438561136,
+      "step": 7615,
+      "train_runtime": 42728.4304,
+      "train_tokens_per_second": 10263.919
+    },
+    {
+      "epoch": 1.753336401288541,
+      "grad_norm": 0.8006819009796143,
+      "learning_rate": 3.713210903240566e-07,
+      "loss": 0.3751,
+      "num_input_tokens_seen": 438835272,
+      "step": 7620,
+      "train_runtime": 42753.8059,
+      "train_tokens_per_second": 10264.239
+    },
+    {
+      "epoch": 1.7544868844914863,
+      "grad_norm": 0.8528444857946174,
+      "learning_rate": 3.679115455685528e-07,
+      "loss": 0.3792,
+      "num_input_tokens_seen": 439123608,
+      "step": 7625,
+      "train_runtime": 42780.7908,
+      "train_tokens_per_second": 10264.504
+    },
+    {
+      "epoch": 1.7556373676944317,
+      "grad_norm": 0.8120623359032298,
+      "learning_rate": 3.645171286573035e-07,
+      "loss": 0.4088,
+      "num_input_tokens_seen": 439413480,
+      "step": 7630,
+      "train_runtime": 42811.2204,
+      "train_tokens_per_second": 10263.979
+    },
+    {
+      "epoch": 1.7567878508973769,
+      "grad_norm": 0.8065600619370311,
+      "learning_rate": 3.6113785067606543e-07,
+      "loss": 0.4028,
+      "num_input_tokens_seen": 439703104,
+      "step": 7635,
+      "train_runtime": 42844.7378,
+      "train_tokens_per_second": 10262.71
+    },
+    {
+      "epoch": 1.7579383341003223,
+      "grad_norm": 0.8032905316047377,
+      "learning_rate": 3.5777372266115496e-07,
+      "loss": 0.4146,
+      "num_input_tokens_seen": 440003712,
+      "step": 7640,
+      "train_runtime": 42876.3428,
+      "train_tokens_per_second": 10262.156
+    },
+    {
+      "epoch": 1.7590888173032675,
+      "grad_norm": 0.9840038166800669,
+      "learning_rate": 3.5442475559941013e-07,
+      "loss": 0.3697,
+      "num_input_tokens_seen": 440283928,
+      "step": 7645,
+      "train_runtime": 42908.0026,
+      "train_tokens_per_second": 10261.115
+    },
+    {
+      "epoch": 1.7602393005062127,
+      "grad_norm": 0.805099524826727,
+      "learning_rate": 3.510909604281537e-07,
+      "loss": 0.3787,
+      "num_input_tokens_seen": 440595544,
+      "step": 7650,
+      "train_runtime": 42943.66,
+      "train_tokens_per_second": 10259.851
+    },
+    {
+      "epoch": 1.7613897837091579,
+      "grad_norm": 0.7251379658997159,
+      "learning_rate": 3.477723480351619e-07,
+      "loss": 0.3698,
+      "num_input_tokens_seen": 440876048,
+      "step": 7655,
+      "train_runtime": 42975.8788,
+      "train_tokens_per_second": 10258.686
+    },
+    {
+      "epoch": 1.762540266912103,
+      "grad_norm": 0.7929938045924546,
+      "learning_rate": 3.444689292586234e-07,
+      "loss": 0.3829,
+      "num_input_tokens_seen": 441182944,
+      "step": 7660,
+      "train_runtime": 43000.7937,
+      "train_tokens_per_second": 10259.879
+    },
+    {
+      "epoch": 1.7636907501150483,
+      "grad_norm": 0.7815965233989962,
+      "learning_rate": 3.411807148871066e-07,
+      "loss": 0.3706,
+      "num_input_tokens_seen": 441491368,
+      "step": 7665,
+      "train_runtime": 43030.1288,
+      "train_tokens_per_second": 10260.052
+    },
+    {
+      "epoch": 1.7648412333179935,
+      "grad_norm": 0.756368051026391,
+      "learning_rate": 3.3790771565952395e-07,
+      "loss": 0.3887,
+      "num_input_tokens_seen": 441788752,
+      "step": 7670,
+      "train_runtime": 43056.3976,
+      "train_tokens_per_second": 10260.699
+    },
+    {
+      "epoch": 1.7659917165209387,
+      "grad_norm": 0.9429354740409847,
+      "learning_rate": 3.346499422650984e-07,
+      "loss": 0.3974,
+      "num_input_tokens_seen": 442047936,
+      "step": 7675,
+      "train_runtime": 43087.9889,
+      "train_tokens_per_second": 10259.192
+    },
+    {
+      "epoch": 1.7671421997238839,
+      "grad_norm": 0.7733650315370391,
+      "learning_rate": 3.3140740534332637e-07,
+      "loss": 0.388,
+      "num_input_tokens_seen": 442344768,
+      "step": 7680,
+      "train_runtime": 43121.1223,
+      "train_tokens_per_second": 10258.192
+    },
+    {
+      "epoch": 1.7682926829268293,
+      "grad_norm": 0.8793520665673636,
+      "learning_rate": 3.281801154839437e-07,
+      "loss": 0.3917,
+      "num_input_tokens_seen": 442628864,
+      "step": 7685,
+      "train_runtime": 43146.6099,
+      "train_tokens_per_second": 10258.717
+    },
+    {
+      "epoch": 1.7694431661297745,
+      "grad_norm": 0.7558895405477133,
+      "learning_rate": 3.2496808322689086e-07,
+      "loss": 0.3828,
+      "num_input_tokens_seen": 442932656,
+      "step": 7690,
+      "train_runtime": 43168.6679,
+      "train_tokens_per_second": 10260.512
+    },
+    {
+      "epoch": 1.77059364933272,
+      "grad_norm": 0.7153114919709856,
+      "learning_rate": 3.2177131906228066e-07,
+      "loss": 0.3813,
+      "num_input_tokens_seen": 443234928,
+      "step": 7695,
+      "train_runtime": 43198.2378,
+      "train_tokens_per_second": 10260.486
+    },
+    {
+      "epoch": 1.771744132535665,
+      "grad_norm": 0.8358437152610306,
+      "learning_rate": 3.185898334303583e-07,
+      "loss": 0.3857,
+      "num_input_tokens_seen": 443529600,
+      "step": 7700,
+      "train_runtime": 43224.1296,
+      "train_tokens_per_second": 10261.157
+    },
+    {
+      "epoch": 1.7728946157386103,
+      "grad_norm": 0.8651149613116922,
+      "learning_rate": 3.1542363672147414e-07,
+      "loss": 0.3916,
+      "num_input_tokens_seen": 443839144,
+      "step": 7705,
+      "train_runtime": 43247.5277,
+      "train_tokens_per_second": 10262.763
+    },
+    {
+      "epoch": 1.7740450989415555,
+      "grad_norm": 0.7647075733919965,
+      "learning_rate": 3.1227273927604826e-07,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 444134608,
+      "step": 7710,
+      "train_runtime": 43280.0351,
+      "train_tokens_per_second": 10261.882
+    },
+    {
+      "epoch": 1.7751955821445007,
+      "grad_norm": 0.735202395502504,
+      "learning_rate": 3.091371513845315e-07,
+      "loss": 0.3849,
+      "num_input_tokens_seen": 444406512,
+      "step": 7715,
+      "train_runtime": 43307.7772,
+      "train_tokens_per_second": 10261.587
+    },
+    {
+      "epoch": 1.7763460653474459,
+      "grad_norm": 0.8564344475316903,
+      "learning_rate": 3.060168832873772e-07,
+      "loss": 0.3937,
+      "num_input_tokens_seen": 444674296,
+      "step": 7720,
+      "train_runtime": 43336.0811,
+      "train_tokens_per_second": 10261.064
+    },
+    {
+      "epoch": 1.777496548550391,
+      "grad_norm": 0.7679571287788206,
+      "learning_rate": 3.0291194517500565e-07,
+      "loss": 0.3765,
+      "num_input_tokens_seen": 444970208,
+      "step": 7725,
+      "train_runtime": 43367.3523,
+      "train_tokens_per_second": 10260.488
+    },
+    {
+      "epoch": 1.7786470317533363,
+      "grad_norm": 0.7631195924453881,
+      "learning_rate": 2.9982234718777205e-07,
+      "loss": 0.396,
+      "num_input_tokens_seen": 445254384,
+      "step": 7730,
+      "train_runtime": 43399.9101,
+      "train_tokens_per_second": 10259.339
+    },
+    {
+      "epoch": 1.7797975149562817,
+      "grad_norm": 0.7751919197056084,
+      "learning_rate": 2.967480994159322e-07,
+      "loss": 0.3811,
+      "num_input_tokens_seen": 445555248,
+      "step": 7735,
+      "train_runtime": 43429.7404,
+      "train_tokens_per_second": 10259.22
+    },
+    {
+      "epoch": 1.7809479981592269,
+      "grad_norm": 0.820805340091573,
+      "learning_rate": 2.9368921189960985e-07,
+      "loss": 0.372,
+      "num_input_tokens_seen": 445812224,
+      "step": 7740,
+      "train_runtime": 43451.5014,
+      "train_tokens_per_second": 10259.996
+    },
+    {
+      "epoch": 1.782098481362172,
+      "grad_norm": 0.8594960424145742,
+      "learning_rate": 2.906456946287639e-07,
+      "loss": 0.3868,
+      "num_input_tokens_seen": 446090696,
+      "step": 7745,
+      "train_runtime": 43481.7841,
+      "train_tokens_per_second": 10259.255
+    },
+    {
+      "epoch": 1.7832489645651175,
+      "grad_norm": 0.8673619025018974,
+      "learning_rate": 2.8761755754315666e-07,
+      "loss": 0.3881,
+      "num_input_tokens_seen": 446384576,
+      "step": 7750,
+      "train_runtime": 43511.1382,
+      "train_tokens_per_second": 10259.088
+    },
+    {
+      "epoch": 1.7843994477680627,
+      "grad_norm": 0.7988176424430542,
+      "learning_rate": 2.8460481053231905e-07,
+      "loss": 0.3805,
+      "num_input_tokens_seen": 446672640,
+      "step": 7755,
+      "train_runtime": 43539.1353,
+      "train_tokens_per_second": 10259.107
+    },
+    {
+      "epoch": 1.785549930971008,
+      "grad_norm": 0.7353695905263102,
+      "learning_rate": 2.8160746343551983e-07,
+      "loss": 0.3876,
+      "num_input_tokens_seen": 446972520,
+      "step": 7760,
+      "train_runtime": 43559.5923,
+      "train_tokens_per_second": 10261.173
+    },
+    {
+      "epoch": 1.786700414173953,
+      "grad_norm": 0.8467251440277036,
+      "learning_rate": 2.786255260417375e-07,
+      "loss": 0.3687,
+      "num_input_tokens_seen": 447254016,
+      "step": 7765,
+      "train_runtime": 43591.1212,
+      "train_tokens_per_second": 10260.209
+    },
+    {
+      "epoch": 1.7878508973768983,
+      "grad_norm": 0.727389711475817,
+      "learning_rate": 2.7565900808961876e-07,
+      "loss": 0.3973,
+      "num_input_tokens_seen": 447553576,
+      "step": 7770,
+      "train_runtime": 43613.9133,
+      "train_tokens_per_second": 10261.716
+    },
+    {
+      "epoch": 1.7890013805798435,
+      "grad_norm": 0.7852081860698105,
+      "learning_rate": 2.7270791926745534e-07,
+      "loss": 0.3936,
+      "num_input_tokens_seen": 447844192,
+      "step": 7775,
+      "train_runtime": 43637.7278,
+      "train_tokens_per_second": 10262.775
+    },
+    {
+      "epoch": 1.7901518637827887,
+      "grad_norm": 0.7077950409093857,
+      "learning_rate": 2.697722692131488e-07,
+      "loss": 0.3827,
+      "num_input_tokens_seen": 448117920,
+      "step": 7780,
+      "train_runtime": 43671.5023,
+      "train_tokens_per_second": 10261.106
+    },
+    {
+      "epoch": 1.7913023469857339,
+      "grad_norm": 0.8201351894652101,
+      "learning_rate": 2.6685206751417946e-07,
+      "loss": 0.3728,
+      "num_input_tokens_seen": 448439376,
+      "step": 7785,
+      "train_runtime": 43697.9977,
+      "train_tokens_per_second": 10262.241
+    },
+    {
+      "epoch": 1.7924528301886793,
+      "grad_norm": 0.6988067899310945,
+      "learning_rate": 2.639473237075751e-07,
+      "loss": 0.3995,
+      "num_input_tokens_seen": 448749264,
+      "step": 7790,
+      "train_runtime": 43732.5633,
+      "train_tokens_per_second": 10261.216
+    },
+    {
+      "epoch": 1.7936033133916245,
+      "grad_norm": 0.7180512665562236,
+      "learning_rate": 2.6105804727988017e-07,
+      "loss": 0.3657,
+      "num_input_tokens_seen": 449052920,
+      "step": 7795,
+      "train_runtime": 43757.9786,
+      "train_tokens_per_second": 10262.195
+    },
+    {
+      "epoch": 1.7947537965945697,
+      "grad_norm": 0.8384841607177868,
+      "learning_rate": 2.581842476671248e-07,
+      "loss": 0.3967,
+      "num_input_tokens_seen": 449332912,
+      "step": 7800,
+      "train_runtime": 43782.7307,
+      "train_tokens_per_second": 10262.789
+    },
+    {
+      "epoch": 1.795904279797515,
+      "grad_norm": 0.7753666992792244,
+      "learning_rate": 2.5532593425479145e-07,
+      "loss": 0.3748,
+      "num_input_tokens_seen": 449618432,
+      "step": 7805,
+      "train_runtime": 43801.1104,
+      "train_tokens_per_second": 10265.001
+    },
+    {
+      "epoch": 1.7970547630004603,
+      "grad_norm": 0.8114046817517676,
+      "learning_rate": 2.524831163777897e-07,
+      "loss": 0.3724,
+      "num_input_tokens_seen": 449884008,
+      "step": 7810,
+      "train_runtime": 43831.2488,
+      "train_tokens_per_second": 10264.002
+    },
+    {
+      "epoch": 1.7982052462034055,
+      "grad_norm": 0.8747366258768403,
+      "learning_rate": 2.496558033204205e-07,
+      "loss": 0.3923,
+      "num_input_tokens_seen": 450185408,
+      "step": 7815,
+      "train_runtime": 43854.652,
+      "train_tokens_per_second": 10265.397
+    },
+    {
+      "epoch": 1.7993557294063507,
+      "grad_norm": 0.8664339602914535,
+      "learning_rate": 2.4684400431634835e-07,
+      "loss": 0.4039,
+      "num_input_tokens_seen": 450467432,
+      "step": 7820,
+      "train_runtime": 43879.564,
+      "train_tokens_per_second": 10265.996
+    },
+    {
+      "epoch": 1.8005062126092959,
+      "grad_norm": 0.7525187035740254,
+      "learning_rate": 2.440477285485715e-07,
+      "loss": 0.3825,
+      "num_input_tokens_seen": 450744960,
+      "step": 7825,
+      "train_runtime": 43909.1167,
+      "train_tokens_per_second": 10265.407
+    },
+    {
+      "epoch": 1.801656695812241,
+      "grad_norm": 1.0598054465773903,
+      "learning_rate": 2.412669851493904e-07,
+      "loss": 0.3588,
+      "num_input_tokens_seen": 451013096,
+      "step": 7830,
+      "train_runtime": 43935.3984,
+      "train_tokens_per_second": 10265.369
+    },
+    {
+      "epoch": 1.8028071790151863,
+      "grad_norm": 0.7860278501473493,
+      "learning_rate": 2.3850178320037944e-07,
+      "loss": 0.3855,
+      "num_input_tokens_seen": 451308136,
+      "step": 7835,
+      "train_runtime": 43964.7274,
+      "train_tokens_per_second": 10265.232
+    },
+    {
+      "epoch": 1.8039576622181315,
+      "grad_norm": 0.8579401623204803,
+      "learning_rate": 2.3575213173235411e-07,
+      "loss": 0.3784,
+      "num_input_tokens_seen": 451610352,
+      "step": 7840,
+      "train_runtime": 43995.394,
+      "train_tokens_per_second": 10264.946
+    },
+    {
+      "epoch": 1.805108145421077,
+      "grad_norm": 0.8647390007260074,
+      "learning_rate": 2.330180397253473e-07,
+      "loss": 0.3931,
+      "num_input_tokens_seen": 451907248,
+      "step": 7845,
+      "train_runtime": 44014.1545,
+      "train_tokens_per_second": 10267.316
+    },
+    {
+      "epoch": 1.806258628624022,
+      "grad_norm": 0.781895913675853,
+      "learning_rate": 2.302995161085747e-07,
+      "loss": 0.3941,
+      "num_input_tokens_seen": 452215256,
+      "step": 7850,
+      "train_runtime": 44041.2441,
+      "train_tokens_per_second": 10267.995
+    },
+    {
+      "epoch": 1.8074091118269673,
+      "grad_norm": 0.8801163908796872,
+      "learning_rate": 2.2759656976040712e-07,
+      "loss": 0.3758,
+      "num_input_tokens_seen": 452501032,
+      "step": 7855,
+      "train_runtime": 44073.8922,
+      "train_tokens_per_second": 10266.872
+    },
+    {
+      "epoch": 1.8085595950299127,
+      "grad_norm": 0.7482439676426204,
+      "learning_rate": 2.2490920950834329e-07,
+      "loss": 0.388,
+      "num_input_tokens_seen": 452781360,
+      "step": 7860,
+      "train_runtime": 44102.8337,
+      "train_tokens_per_second": 10266.491
+    },
+    {
+      "epoch": 1.809710078232858,
+      "grad_norm": 0.8168237374501867,
+      "learning_rate": 2.2223744412897707e-07,
+      "loss": 0.3761,
+      "num_input_tokens_seen": 453058280,
+      "step": 7865,
+      "train_runtime": 44122.2386,
+      "train_tokens_per_second": 10268.252
+    },
+    {
+      "epoch": 1.810860561435803,
+      "grad_norm": 0.8047464223180236,
+      "learning_rate": 2.195812823479765e-07,
+      "loss": 0.3931,
+      "num_input_tokens_seen": 453339304,
+      "step": 7870,
+      "train_runtime": 44155.0801,
+      "train_tokens_per_second": 10266.98
+    },
+    {
+      "epoch": 1.8120110446387483,
+      "grad_norm": 0.9224847979667109,
+      "learning_rate": 2.1694073284004357e-07,
+      "loss": 0.3605,
+      "num_input_tokens_seen": 453639608,
+      "step": 7875,
+      "train_runtime": 44187.1042,
+      "train_tokens_per_second": 10266.335
+    },
+    {
+      "epoch": 1.8131615278416935,
+      "grad_norm": 0.8078201468651051,
+      "learning_rate": 2.1431580422889732e-07,
+      "loss": 0.3691,
+      "num_input_tokens_seen": 453928432,
+      "step": 7880,
+      "train_runtime": 44216.2869,
+      "train_tokens_per_second": 10266.091
+    },
+    {
+      "epoch": 1.8143120110446387,
+      "grad_norm": 0.8594658351833375,
+      "learning_rate": 2.11706505087238e-07,
+      "loss": 0.3975,
+      "num_input_tokens_seen": 454214264,
+      "step": 7885,
+      "train_runtime": 44243.0995,
+      "train_tokens_per_second": 10266.33
+    },
+    {
+      "epoch": 1.8154624942475839,
+      "grad_norm": 0.6768348862857614,
+      "learning_rate": 2.0911284393672515e-07,
+      "loss": 0.3751,
+      "num_input_tokens_seen": 454512776,
+      "step": 7890,
+      "train_runtime": 44274.3927,
+      "train_tokens_per_second": 10265.816
+    },
+    {
+      "epoch": 1.816612977450529,
+      "grad_norm": 0.7111060866818695,
+      "learning_rate": 2.0653482924794132e-07,
+      "loss": 0.389,
+      "num_input_tokens_seen": 454816440,
+      "step": 7895,
+      "train_runtime": 44299.6466,
+      "train_tokens_per_second": 10266.819
+    },
+    {
+      "epoch": 1.8177634606534745,
+      "grad_norm": 0.7309519296301149,
+      "learning_rate": 2.0397246944037497e-07,
+      "loss": 0.3857,
+      "num_input_tokens_seen": 455093304,
+      "step": 7900,
+      "train_runtime": 44323.5003,
+      "train_tokens_per_second": 10267.54
+    },
+    {
+      "epoch": 1.8189139438564197,
+      "grad_norm": 0.9933341648300974,
+      "learning_rate": 2.0142577288238542e-07,
+      "loss": 0.3965,
+      "num_input_tokens_seen": 455374272,
+      "step": 7905,
+      "train_runtime": 44353.9347,
+      "train_tokens_per_second": 10266.829
+    },
+    {
+      "epoch": 1.8200644270593649,
+      "grad_norm": 0.7686901293150897,
+      "learning_rate": 1.9889474789117625e-07,
+      "loss": 0.3658,
+      "num_input_tokens_seen": 455650936,
+      "step": 7910,
+      "train_runtime": 44376.4052,
+      "train_tokens_per_second": 10267.865
+    },
+    {
+      "epoch": 1.8212149102623103,
+      "grad_norm": 0.8491390010894678,
+      "learning_rate": 1.96379402732772e-07,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 455954608,
+      "step": 7915,
+      "train_runtime": 44405.025,
+      "train_tokens_per_second": 10268.086
+    },
+    {
+      "epoch": 1.8223653934652555,
+      "grad_norm": 0.6822149749555213,
+      "learning_rate": 1.9387974562198652e-07,
+      "loss": 0.3997,
+      "num_input_tokens_seen": 456247784,
+      "step": 7920,
+      "train_runtime": 44430.4467,
+      "train_tokens_per_second": 10268.809
+    },
+    {
+      "epoch": 1.8235158766682007,
+      "grad_norm": 0.8118867087474307,
+      "learning_rate": 1.9139578472240072e-07,
+      "loss": 0.3962,
+      "num_input_tokens_seen": 456540616,
+      "step": 7925,
+      "train_runtime": 44453.2181,
+      "train_tokens_per_second": 10270.136
+    },
+    {
+      "epoch": 1.824666359871146,
+      "grad_norm": 0.669250488253454,
+      "learning_rate": 1.8892752814633208e-07,
+      "loss": 0.3579,
+      "num_input_tokens_seen": 456825720,
+      "step": 7930,
+      "train_runtime": 44479.4718,
+      "train_tokens_per_second": 10270.484
+    },
+    {
+      "epoch": 1.825816843074091,
+      "grad_norm": 0.8145307011932224,
+      "learning_rate": 1.8647498395480968e-07,
+      "loss": 0.3585,
+      "num_input_tokens_seen": 457146504,
+      "step": 7935,
+      "train_runtime": 44510.5734,
+      "train_tokens_per_second": 10270.515
+    },
+    {
+      "epoch": 1.8269673262770363,
+      "grad_norm": 0.690635286742687,
+      "learning_rate": 1.840381601575486e-07,
+      "loss": 0.3887,
+      "num_input_tokens_seen": 457417048,
+      "step": 7940,
+      "train_runtime": 44543.372,
+      "train_tokens_per_second": 10269.026
+    },
+    {
+      "epoch": 1.8281178094799815,
+      "grad_norm": 0.9173569900684793,
+      "learning_rate": 1.8161706471292174e-07,
+      "loss": 0.4017,
+      "num_input_tokens_seen": 457703456,
+      "step": 7945,
+      "train_runtime": 44570.0952,
+      "train_tokens_per_second": 10269.295
+    },
+    {
+      "epoch": 1.8292682926829267,
+      "grad_norm": 0.9129713084950523,
+      "learning_rate": 1.7921170552793633e-07,
+      "loss": 0.3906,
+      "num_input_tokens_seen": 458000392,
+      "step": 7950,
+      "train_runtime": 44594.1901,
+      "train_tokens_per_second": 10270.405
+    },
+    {
+      "epoch": 1.830418775885872,
+      "grad_norm": 0.7683983019820931,
+      "learning_rate": 1.7682209045820687e-07,
+      "loss": 0.3814,
+      "num_input_tokens_seen": 458277016,
+      "step": 7955,
+      "train_runtime": 44621.6979,
+      "train_tokens_per_second": 10270.273
+    },
+    {
+      "epoch": 1.8315692590888173,
+      "grad_norm": 0.8453410193571166,
+      "learning_rate": 1.7444822730792843e-07,
+      "loss": 0.3688,
+      "num_input_tokens_seen": 458578936,
+      "step": 7960,
+      "train_runtime": 44648.8938,
+      "train_tokens_per_second": 10270.779
+    },
+    {
+      "epoch": 1.8327197422917625,
+      "grad_norm": 0.8389872152730282,
+      "learning_rate": 1.720901238298528e-07,
+      "loss": 0.3946,
+      "num_input_tokens_seen": 458856384,
+      "step": 7965,
+      "train_runtime": 44682.4289,
+      "train_tokens_per_second": 10269.28
+    },
+    {
+      "epoch": 1.833870225494708,
+      "grad_norm": 0.752303822031851,
+      "learning_rate": 1.697477877252629e-07,
+      "loss": 0.3825,
+      "num_input_tokens_seen": 459157176,
+      "step": 7970,
+      "train_runtime": 44713.0467,
+      "train_tokens_per_second": 10268.975
+    },
+    {
+      "epoch": 1.835020708697653,
+      "grad_norm": 0.8139030086706649,
+      "learning_rate": 1.6742122664394677e-07,
+      "loss": 0.4134,
+      "num_input_tokens_seen": 459440440,
+      "step": 7975,
+      "train_runtime": 44742.0539,
+      "train_tokens_per_second": 10268.649
+    },
+    {
+      "epoch": 1.8361711919005983,
+      "grad_norm": 0.7839276882386372,
+      "learning_rate": 1.651104481841731e-07,
+      "loss": 0.3737,
+      "num_input_tokens_seen": 459729736,
+      "step": 7980,
+      "train_runtime": 44769.4754,
+      "train_tokens_per_second": 10268.821
+    },
+    {
+      "epoch": 1.8373216751035435,
+      "grad_norm": 0.8541231192598258,
+      "learning_rate": 1.6281545989266678e-07,
+      "loss": 0.3646,
+      "num_input_tokens_seen": 460010056,
+      "step": 7985,
+      "train_runtime": 44801.982,
+      "train_tokens_per_second": 10267.627
+    },
+    {
+      "epoch": 1.8384721583064887,
+      "grad_norm": 0.814686798538712,
+      "learning_rate": 1.6053626926458509e-07,
+      "loss": 0.3631,
+      "num_input_tokens_seen": 460288856,
+      "step": 7990,
+      "train_runtime": 44832.8241,
+      "train_tokens_per_second": 10266.783
+    },
+    {
+      "epoch": 1.8396226415094339,
+      "grad_norm": 0.7943980356028599,
+      "learning_rate": 1.5827288374348825e-07,
+      "loss": 0.3555,
+      "num_input_tokens_seen": 460574576,
+      "step": 7995,
+      "train_runtime": 44862.1694,
+      "train_tokens_per_second": 10266.436
+    },
+    {
+      "epoch": 1.840773124712379,
+      "grad_norm": 0.8946189234825646,
+      "learning_rate": 1.5602531072132331e-07,
+      "loss": 0.3933,
+      "num_input_tokens_seen": 460877824,
+      "step": 8000,
+      "train_runtime": 44888.5445,
+      "train_tokens_per_second": 10267.159
+    },
+    {
+      "epoch": 1.8419236079153243,
+      "grad_norm": 0.733555882476511,
+      "learning_rate": 1.5379355753839197e-07,
+      "loss": 0.3755,
+      "num_input_tokens_seen": 461178112,
+      "step": 8005,
+      "train_runtime": 44917.9984,
+      "train_tokens_per_second": 10267.112
+    },
+    {
+      "epoch": 1.8430740911182697,
+      "grad_norm": 0.7912510035387609,
+      "learning_rate": 1.515776314833345e-07,
+      "loss": 0.3683,
+      "num_input_tokens_seen": 461483232,
+      "step": 8010,
+      "train_runtime": 44941.7646,
+      "train_tokens_per_second": 10268.472
+    },
+    {
+      "epoch": 1.844224574321215,
+      "grad_norm": 0.839653667988869,
+      "learning_rate": 1.4937753979309688e-07,
+      "loss": 0.3919,
+      "num_input_tokens_seen": 461767352,
+      "step": 8015,
+      "train_runtime": 44963.7473,
+      "train_tokens_per_second": 10269.77
+    },
+    {
+      "epoch": 1.8453750575241603,
+      "grad_norm": 0.8513219667232966,
+      "learning_rate": 1.4719328965291436e-07,
+      "loss": 0.381,
+      "num_input_tokens_seen": 462028976,
+      "step": 8020,
+      "train_runtime": 44992.0045,
+      "train_tokens_per_second": 10269.135
+    },
+    {
+      "epoch": 1.8465255407271055,
+      "grad_norm": 0.7466040720604709,
+      "learning_rate": 1.450248881962868e-07,
+      "loss": 0.4088,
+      "num_input_tokens_seen": 462339312,
+      "step": 8025,
+      "train_runtime": 45023.6165,
+      "train_tokens_per_second": 10268.818
+    },
+    {
+      "epoch": 1.8476760239300507,
+      "grad_norm": 0.7177724190508968,
+      "learning_rate": 1.4287234250495163e-07,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 462635616,
+      "step": 8030,
+      "train_runtime": 45050.848,
+      "train_tokens_per_second": 10269.188
+    },
+    {
+      "epoch": 1.848826507132996,
+      "grad_norm": 0.7789775384372521,
+      "learning_rate": 1.4073565960886436e-07,
+      "loss": 0.3797,
+      "num_input_tokens_seen": 462925896,
+      "step": 8035,
+      "train_runtime": 45079.9748,
+      "train_tokens_per_second": 10268.992
+    },
+    {
+      "epoch": 1.849976990335941,
+      "grad_norm": 0.7823564413111995,
+      "learning_rate": 1.3861484648617528e-07,
+      "loss": 0.384,
+      "num_input_tokens_seen": 463184552,
+      "step": 8040,
+      "train_runtime": 45112.5889,
+      "train_tokens_per_second": 10267.302
+    },
+    {
+      "epoch": 1.8511274735388863,
+      "grad_norm": 0.8623810304061909,
+      "learning_rate": 1.3650991006320559e-07,
+      "loss": 0.403,
+      "num_input_tokens_seen": 463467680,
+      "step": 8045,
+      "train_runtime": 45141.9898,
+      "train_tokens_per_second": 10266.886
+    },
+    {
+      "epoch": 1.8522779567418315,
+      "grad_norm": 0.8860655156918921,
+      "learning_rate": 1.344208572144229e-07,
+      "loss": 0.3791,
+      "num_input_tokens_seen": 463734952,
+      "step": 8050,
+      "train_runtime": 45171.9766,
+      "train_tokens_per_second": 10265.988
+    },
+    {
+      "epoch": 1.8534284399447767,
+      "grad_norm": 0.7578751656012174,
+      "learning_rate": 1.3234769476242358e-07,
+      "loss": 0.3565,
+      "num_input_tokens_seen": 464037984,
+      "step": 8055,
+      "train_runtime": 45201.5827,
+      "train_tokens_per_second": 10265.968
+    },
+    {
+      "epoch": 1.8545789231477219,
+      "grad_norm": 0.7251452501701983,
+      "learning_rate": 1.3029042947790838e-07,
+      "loss": 0.3918,
+      "num_input_tokens_seen": 464333512,
+      "step": 8060,
+      "train_runtime": 45227.5128,
+      "train_tokens_per_second": 10266.616
+    },
+    {
+      "epoch": 1.8557294063506673,
+      "grad_norm": 1.0244303839204194,
+      "learning_rate": 1.2824906807965665e-07,
+      "loss": 0.3867,
+      "num_input_tokens_seen": 464607376,
+      "step": 8065,
+      "train_runtime": 45253.4209,
+      "train_tokens_per_second": 10266.79
+    },
+    {
+      "epoch": 1.8568798895536125,
+      "grad_norm": 0.990822767974288,
+      "learning_rate": 1.262236172345105e-07,
+      "loss": 0.4095,
+      "num_input_tokens_seen": 464878840,
+      "step": 8070,
+      "train_runtime": 45280.4506,
+      "train_tokens_per_second": 10266.657
+    },
+    {
+      "epoch": 1.858030372756558,
+      "grad_norm": 0.9278559018231897,
+      "learning_rate": 1.2421408355734854e-07,
+      "loss": 0.3768,
+      "num_input_tokens_seen": 465139816,
+      "step": 8075,
+      "train_runtime": 45302.1875,
+      "train_tokens_per_second": 10267.491
+    },
+    {
+      "epoch": 1.859180855959503,
+      "grad_norm": 0.7411689441246126,
+      "learning_rate": 1.2222047361106716e-07,
+      "loss": 0.3973,
+      "num_input_tokens_seen": 465445064,
+      "step": 8080,
+      "train_runtime": 45332.8313,
+      "train_tokens_per_second": 10267.284
+    },
+    {
+      "epoch": 1.8603313391624483,
+      "grad_norm": 0.8600792629124747,
+      "learning_rate": 1.202427939065548e-07,
+      "loss": 0.3893,
+      "num_input_tokens_seen": 465736000,
+      "step": 8085,
+      "train_runtime": 45358.9509,
+      "train_tokens_per_second": 10267.786
+    },
+    {
+      "epoch": 1.8614818223653935,
+      "grad_norm": 0.7235267992935049,
+      "learning_rate": 1.1828105090267717e-07,
+      "loss": 0.3912,
+      "num_input_tokens_seen": 466041776,
+      "step": 8090,
+      "train_runtime": 45388.6889,
+      "train_tokens_per_second": 10267.796
+    },
+    {
+      "epoch": 1.8626323055683387,
+      "grad_norm": 0.7730327303978245,
+      "learning_rate": 1.163352510062521e-07,
+      "loss": 0.3849,
+      "num_input_tokens_seen": 466341456,
+      "step": 8095,
+      "train_runtime": 45422.5514,
+      "train_tokens_per_second": 10266.738
+    },
+    {
+      "epoch": 1.863782788771284,
+      "grad_norm": 0.88949977254397,
+      "learning_rate": 1.1440540057202632e-07,
+      "loss": 0.3847,
+      "num_input_tokens_seen": 466645800,
+      "step": 8100,
+      "train_runtime": 45449.957,
+      "train_tokens_per_second": 10267.244
+    },
+    {
+      "epoch": 1.864933271974229,
+      "grad_norm": 0.7470425204081507,
+      "learning_rate": 1.1249150590266045e-07,
+      "loss": 0.3805,
+      "num_input_tokens_seen": 466912696,
+      "step": 8105,
+      "train_runtime": 45481.4464,
+      "train_tokens_per_second": 10266.004
+    },
+    {
+      "epoch": 1.8660837551771743,
+      "grad_norm": 0.8609583415692328,
+      "learning_rate": 1.1059357324870456e-07,
+      "loss": 0.3891,
+      "num_input_tokens_seen": 467173352,
+      "step": 8110,
+      "train_runtime": 45508.0432,
+      "train_tokens_per_second": 10265.731
+    },
+    {
+      "epoch": 1.8672342383801197,
+      "grad_norm": 0.7705097750533386,
+      "learning_rate": 1.0871160880857767e-07,
+      "loss": 0.3854,
+      "num_input_tokens_seen": 467449808,
+      "step": 8115,
+      "train_runtime": 45531.1019,
+      "train_tokens_per_second": 10266.604
+    },
+    {
+      "epoch": 1.868384721583065,
+      "grad_norm": 0.8904904664019506,
+      "learning_rate": 1.0684561872855103e-07,
+      "loss": 0.3986,
+      "num_input_tokens_seen": 467740136,
+      "step": 8120,
+      "train_runtime": 45557.6077,
+      "train_tokens_per_second": 10267.004
+    },
+    {
+      "epoch": 1.86953520478601,
+      "grad_norm": 0.7680644229892677,
+      "learning_rate": 1.0499560910272266e-07,
+      "loss": 0.3928,
+      "num_input_tokens_seen": 468027424,
+      "step": 8125,
+      "train_runtime": 45586.1642,
+      "train_tokens_per_second": 10266.874
+    },
+    {
+      "epoch": 1.8706856879889555,
+      "grad_norm": 0.8189452679012945,
+      "learning_rate": 1.0316158597300285e-07,
+      "loss": 0.3638,
+      "num_input_tokens_seen": 468272456,
+      "step": 8130,
+      "train_runtime": 45611.4441,
+      "train_tokens_per_second": 10266.556
+    },
+    {
+      "epoch": 1.8718361711919007,
+      "grad_norm": 0.9232967372729867,
+      "learning_rate": 1.0134355532908924e-07,
+      "loss": 0.3961,
+      "num_input_tokens_seen": 468550520,
+      "step": 8135,
+      "train_runtime": 45638.4322,
+      "train_tokens_per_second": 10266.578
+    },
+    {
+      "epoch": 1.872986654394846,
+      "grad_norm": 0.7864595351951325,
+      "learning_rate": 9.954152310845233e-08,
+      "loss": 0.3982,
+      "num_input_tokens_seen": 468836424,
+      "step": 8140,
+      "train_runtime": 45664.4956,
+      "train_tokens_per_second": 10266.979
+    },
+    {
+      "epoch": 1.874137137597791,
+      "grad_norm": 0.7083218909824701,
+      "learning_rate": 9.775549519631333e-08,
+      "loss": 0.3877,
+      "num_input_tokens_seen": 469130424,
+      "step": 8145,
+      "train_runtime": 45694.0284,
+      "train_tokens_per_second": 10266.778
+    },
+    {
+      "epoch": 1.8752876208007363,
+      "grad_norm": 0.8398593800655513,
+      "learning_rate": 9.598547742562414e-08,
+      "loss": 0.3681,
+      "num_input_tokens_seen": 469406248,
+      "step": 8150,
+      "train_runtime": 45721.3858,
+      "train_tokens_per_second": 10266.667
+    },
+    {
+      "epoch": 1.8764381040036815,
+      "grad_norm": 0.7837073765835032,
+      "learning_rate": 9.42314755770496e-08,
+      "loss": 0.3697,
+      "num_input_tokens_seen": 469697048,
+      "step": 8155,
+      "train_runtime": 45743.5116,
+      "train_tokens_per_second": 10268.058
+    },
+    {
+      "epoch": 1.8775885872066267,
+      "grad_norm": 0.7720850459739209,
+      "learning_rate": 9.249349537894969e-08,
+      "loss": 0.3609,
+      "num_input_tokens_seen": 469994200,
+      "step": 8160,
+      "train_runtime": 45767.9735,
+      "train_tokens_per_second": 10269.063
+    },
+    {
+      "epoch": 1.8787390704095719,
+      "grad_norm": 0.745502033741882,
+      "learning_rate": 9.077154250735908e-08,
+      "loss": 0.3825,
+      "num_input_tokens_seen": 470298480,
+      "step": 8165,
+      "train_runtime": 45794.9297,
+      "train_tokens_per_second": 10269.663
+    },
+    {
+      "epoch": 1.8798895536125173,
+      "grad_norm": 0.7543475101070007,
+      "learning_rate": 8.906562258596818e-08,
+      "loss": 0.3744,
+      "num_input_tokens_seen": 470581544,
+      "step": 8170,
+      "train_runtime": 45824.5043,
+      "train_tokens_per_second": 10269.212
+    },
+    {
+      "epoch": 1.8810400368154625,
+      "grad_norm": 0.7345279758690928,
+      "learning_rate": 8.737574118610704e-08,
+      "loss": 0.4052,
+      "num_input_tokens_seen": 470846408,
+      "step": 8175,
+      "train_runtime": 45854.1537,
+      "train_tokens_per_second": 10268.348
+    },
+    {
+      "epoch": 1.8821905200184077,
+      "grad_norm": 0.8318656995830941,
+      "learning_rate": 8.570190382672539e-08,
+      "loss": 0.383,
+      "num_input_tokens_seen": 471124816,
+      "step": 8180,
+      "train_runtime": 45882.6016,
+      "train_tokens_per_second": 10268.049
+    },
+    {
+      "epoch": 1.8833410032213531,
+      "grad_norm": 0.7507859368105789,
+      "learning_rate": 8.404411597437435e-08,
+      "loss": 0.3841,
+      "num_input_tokens_seen": 471414992,
+      "step": 8185,
+      "train_runtime": 45913.9919,
+      "train_tokens_per_second": 10267.349
+    },
+    {
+      "epoch": 1.8844914864242983,
+      "grad_norm": 0.8357400057431097,
+      "learning_rate": 8.24023830431897e-08,
+      "loss": 0.3582,
+      "num_input_tokens_seen": 471705008,
+      "step": 8190,
+      "train_runtime": 45951.4186,
+      "train_tokens_per_second": 10265.298
+    },
+    {
+      "epoch": 1.8856419696272435,
+      "grad_norm": 0.8817185627250049,
+      "learning_rate": 8.077671039487423e-08,
+      "loss": 0.3707,
+      "num_input_tokens_seen": 471995440,
+      "step": 8195,
+      "train_runtime": 45983.0615,
+      "train_tokens_per_second": 10264.55
+    },
+    {
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.7255925828865154,
+      "learning_rate": 7.916710333867982e-08,
+      "loss": 0.3813,
+      "num_input_tokens_seen": 472304880,
+      "step": 8200,
+      "train_runtime": 46014.2534,
+      "train_tokens_per_second": 10264.317
+    },
+    {
+      "epoch": 1.887942936033134,
+      "grad_norm": 0.7177309371361038,
+      "learning_rate": 7.757356713138931e-08,
+      "loss": 0.3843,
+      "num_input_tokens_seen": 472593352,
+      "step": 8205,
+      "train_runtime": 46046.4758,
+      "train_tokens_per_second": 10263.399
+    },
+    {
+      "epoch": 1.889093419236079,
+      "grad_norm": 0.7218825331853228,
+      "learning_rate": 7.59961069773002e-08,
+      "loss": 0.3717,
+      "num_input_tokens_seen": 472893584,
+      "step": 8210,
+      "train_runtime": 46082.0673,
+      "train_tokens_per_second": 10261.987
+    },
+    {
+      "epoch": 1.8902439024390243,
+      "grad_norm": 0.702928943550359,
+      "learning_rate": 7.443472802820817e-08,
+      "loss": 0.3682,
+      "num_input_tokens_seen": 473163984,
+      "step": 8215,
+      "train_runtime": 46113.1764,
+      "train_tokens_per_second": 10260.928
+    },
+    {
+      "epoch": 1.8913943856419695,
+      "grad_norm": 0.7031206654336563,
+      "learning_rate": 7.288943538338922e-08,
+      "loss": 0.3912,
+      "num_input_tokens_seen": 473471464,
+      "step": 8220,
+      "train_runtime": 46140.3597,
+      "train_tokens_per_second": 10261.547
+    },
+    {
+      "epoch": 1.892544868844915,
+      "grad_norm": 0.8523582345660332,
+      "learning_rate": 7.136023408958304e-08,
+      "loss": 0.401,
+      "num_input_tokens_seen": 473756120,
+      "step": 8225,
+      "train_runtime": 46168.1778,
+      "train_tokens_per_second": 10261.53
+    },
+    {
+      "epoch": 1.89369535204786,
+      "grad_norm": 0.7937503562642403,
+      "learning_rate": 6.984712914097801e-08,
+      "loss": 0.3732,
+      "num_input_tokens_seen": 474065160,
+      "step": 8230,
+      "train_runtime": 46193.7656,
+      "train_tokens_per_second": 10262.536
+    },
+    {
+      "epoch": 1.8948458352508053,
+      "grad_norm": 0.7054437650091988,
+      "learning_rate": 6.835012547919296e-08,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 474374912,
+      "step": 8235,
+      "train_runtime": 46223.6324,
+      "train_tokens_per_second": 10262.606
+    },
+    {
+      "epoch": 1.8959963184537507,
+      "grad_norm": 0.8188776987073859,
+      "learning_rate": 6.68692279932609e-08,
+      "loss": 0.408,
+      "num_input_tokens_seen": 474663240,
+      "step": 8240,
+      "train_runtime": 46248.9841,
+      "train_tokens_per_second": 10263.214
+    },
+    {
+      "epoch": 1.897146801656696,
+      "grad_norm": 0.8872662924792425,
+      "learning_rate": 6.540444151961533e-08,
+      "loss": 0.3846,
+      "num_input_tokens_seen": 474952592,
+      "step": 8245,
+      "train_runtime": 46275.2367,
+      "train_tokens_per_second": 10263.645
+    },
+    {
+      "epoch": 1.898297284859641,
+      "grad_norm": 0.7117765096368451,
+      "learning_rate": 6.395577084207461e-08,
+      "loss": 0.3783,
+      "num_input_tokens_seen": 475266024,
+      "step": 8250,
+      "train_runtime": 46299.5886,
+      "train_tokens_per_second": 10265.016
+    },
+    {
+      "epoch": 1.8994477680625863,
+      "grad_norm": 0.8026599858351988,
+      "learning_rate": 6.252322069182082e-08,
+      "loss": 0.3573,
+      "num_input_tokens_seen": 475569568,
+      "step": 8255,
+      "train_runtime": 46328.2757,
+      "train_tokens_per_second": 10265.212
+    },
+    {
+      "epoch": 1.9005982512655315,
+      "grad_norm": 0.8983760328913002,
+      "learning_rate": 6.110679574739208e-08,
+      "loss": 0.3871,
+      "num_input_tokens_seen": 475810000,
+      "step": 8260,
+      "train_runtime": 46355.4941,
+      "train_tokens_per_second": 10264.371
+    },
+    {
+      "epoch": 1.9017487344684767,
+      "grad_norm": 0.7312651883046695,
+      "learning_rate": 5.970650063466144e-08,
+      "loss": 0.3578,
+      "num_input_tokens_seen": 476107760,
+      "step": 8265,
+      "train_runtime": 46385.3414,
+      "train_tokens_per_second": 10264.186
+    },
+    {
+      "epoch": 1.902899217671422,
+      "grad_norm": 0.8243553836546688,
+      "learning_rate": 5.832233992682512e-08,
+      "loss": 0.3764,
+      "num_input_tokens_seen": 476389664,
+      "step": 8270,
+      "train_runtime": 46411.0115,
+      "train_tokens_per_second": 10264.583
+    },
+    {
+      "epoch": 1.904049700874367,
+      "grad_norm": 0.8050763442154591,
+      "learning_rate": 5.695431814438379e-08,
+      "loss": 0.3903,
+      "num_input_tokens_seen": 476678088,
+      "step": 8275,
+      "train_runtime": 46445.378,
+      "train_tokens_per_second": 10263.198
+    },
+    {
+      "epoch": 1.9052001840773125,
+      "grad_norm": 0.8491473905344739,
+      "learning_rate": 5.560243975513413e-08,
+      "loss": 0.3677,
+      "num_input_tokens_seen": 476970024,
+      "step": 8280,
+      "train_runtime": 46473.4749,
+      "train_tokens_per_second": 10263.274
+    },
+    {
+      "epoch": 1.9063506672802577,
+      "grad_norm": 0.8598633854323477,
+      "learning_rate": 5.4266709174148335e-08,
+      "loss": 0.4189,
+      "num_input_tokens_seen": 477252488,
+      "step": 8285,
+      "train_runtime": 46504.1157,
+      "train_tokens_per_second": 10262.586
+    },
+    {
+      "epoch": 1.907501150483203,
+      "grad_norm": 0.9238810503558949,
+      "learning_rate": 5.294713076376134e-08,
+      "loss": 0.3655,
+      "num_input_tokens_seen": 477532960,
+      "step": 8290,
+      "train_runtime": 46536.0482,
+      "train_tokens_per_second": 10261.571
+    },
+    {
+      "epoch": 1.9086516336861483,
+      "grad_norm": 0.7411343773915282,
+      "learning_rate": 5.1643708833557515e-08,
+      "loss": 0.378,
+      "num_input_tokens_seen": 477833248,
+      "step": 8295,
+      "train_runtime": 46566.3243,
+      "train_tokens_per_second": 10261.348
+    },
+    {
+      "epoch": 1.9098021168890935,
+      "grad_norm": 0.7596927715799423,
+      "learning_rate": 5.0356447640356185e-08,
+      "loss": 0.3791,
+      "num_input_tokens_seen": 478128560,
+      "step": 8300,
+      "train_runtime": 46592.9472,
+      "train_tokens_per_second": 10261.823
+    },
+    {
+      "epoch": 1.9109526000920387,
+      "grad_norm": 0.8293109560824727,
+      "learning_rate": 4.908535138819892e-08,
+      "loss": 0.3837,
+      "num_input_tokens_seen": 478406624,
+      "step": 8305,
+      "train_runtime": 46613.853,
+      "train_tokens_per_second": 10263.186
+    },
+    {
+      "epoch": 1.912103083294984,
+      "grad_norm": 0.7481410296495528,
+      "learning_rate": 4.7830424228332284e-08,
+      "loss": 0.397,
+      "num_input_tokens_seen": 478693968,
+      "step": 8310,
+      "train_runtime": 46642.2292,
+      "train_tokens_per_second": 10263.102
+    },
+    {
+      "epoch": 1.913253566497929,
+      "grad_norm": 0.8985642239069692,
+      "learning_rate": 4.6591670259197864e-08,
+      "loss": 0.3754,
+      "num_input_tokens_seen": 478987352,
+      "step": 8315,
+      "train_runtime": 46669.5759,
+      "train_tokens_per_second": 10263.375
+    },
+    {
+      "epoch": 1.9144040497008743,
+      "grad_norm": 0.763312333501863,
+      "learning_rate": 4.536909352641727e-08,
+      "loss": 0.3989,
+      "num_input_tokens_seen": 479278648,
+      "step": 8320,
+      "train_runtime": 46701.1999,
+      "train_tokens_per_second": 10262.662
+    },
+    {
+      "epoch": 1.9155545329038195,
+      "grad_norm": 0.7654384631420995,
+      "learning_rate": 4.416269802277884e-08,
+      "loss": 0.3608,
+      "num_input_tokens_seen": 479557496,
+      "step": 8325,
+      "train_runtime": 46730.1265,
+      "train_tokens_per_second": 10262.277
+    },
+    {
+      "epoch": 1.9167050161067647,
+      "grad_norm": 0.8102341718081502,
+      "learning_rate": 4.2972487688225376e-08,
+      "loss": 0.3702,
+      "num_input_tokens_seen": 479853632,
+      "step": 8330,
+      "train_runtime": 46763.9134,
+      "train_tokens_per_second": 10261.195
+    },
+    {
+      "epoch": 1.91785549930971,
+      "grad_norm": 0.8993986260813253,
+      "learning_rate": 4.179846640984142e-08,
+      "loss": 0.3897,
+      "num_input_tokens_seen": 480137456,
+      "step": 8335,
+      "train_runtime": 46797.0469,
+      "train_tokens_per_second": 10259.995
+    },
+    {
+      "epoch": 1.9190059825126553,
+      "grad_norm": 0.71506398695041,
+      "learning_rate": 4.064063802183937e-08,
+      "loss": 0.3826,
+      "num_input_tokens_seen": 480405928,
+      "step": 8340,
+      "train_runtime": 46825.1067,
+      "train_tokens_per_second": 10259.58
+    },
+    {
+      "epoch": 1.9201564657156005,
+      "grad_norm": 0.8235679358126546,
+      "learning_rate": 3.9499006305546705e-08,
+      "loss": 0.3614,
+      "num_input_tokens_seen": 480709960,
+      "step": 8345,
+      "train_runtime": 46860.1934,
+      "train_tokens_per_second": 10258.386
+    },
+    {
+      "epoch": 1.921306948918546,
+      "grad_norm": 0.7999206131465968,
+      "learning_rate": 3.837357498939598e-08,
+      "loss": 0.4015,
+      "num_input_tokens_seen": 480988960,
+      "step": 8350,
+      "train_runtime": 46887.0306,
+      "train_tokens_per_second": 10258.465
+    },
+    {
+      "epoch": 1.9224574321214911,
+      "grad_norm": 0.7531420543849067,
+      "learning_rate": 3.7264347748910967e-08,
+      "loss": 0.4105,
+      "num_input_tokens_seen": 481251952,
+      "step": 8355,
+      "train_runtime": 46916.3171,
+      "train_tokens_per_second": 10257.667
+    },
+    {
+      "epoch": 1.9236079153244363,
+      "grad_norm": 0.7245676200594492,
+      "learning_rate": 3.617132820669389e-08,
+      "loss": 0.3532,
+      "num_input_tokens_seen": 481520784,
+      "step": 8360,
+      "train_runtime": 46945.1452,
+      "train_tokens_per_second": 10257.094
+    },
+    {
+      "epoch": 1.9247583985273815,
+      "grad_norm": 0.8274421617419991,
+      "learning_rate": 3.5094519932415417e-08,
+      "loss": 0.3831,
+      "num_input_tokens_seen": 481803544,
+      "step": 8365,
+      "train_runtime": 46969.4782,
+      "train_tokens_per_second": 10257.801
+    },
+    {
+      "epoch": 1.9259088817303267,
+      "grad_norm": 0.8179317380222653,
+      "learning_rate": 3.40339264428019e-08,
+      "loss": 0.3923,
+      "num_input_tokens_seen": 482085192,
+      "step": 8370,
+      "train_runtime": 46997.3549,
+      "train_tokens_per_second": 10257.709
+    },
+    {
+      "epoch": 1.927059364933272,
+      "grad_norm": 0.7494249724488083,
+      "learning_rate": 3.2989551201624836e-08,
+      "loss": 0.3908,
+      "num_input_tokens_seen": 482360016,
+      "step": 8375,
+      "train_runtime": 47026.9964,
+      "train_tokens_per_second": 10257.087
+    },
+    {
+      "epoch": 1.928209848136217,
+      "grad_norm": 0.7503315020057478,
+      "learning_rate": 3.1961397619687526e-08,
+      "loss": 0.3935,
+      "num_input_tokens_seen": 482658328,
+      "step": 8380,
+      "train_runtime": 47056.9382,
+      "train_tokens_per_second": 10256.9
+    },
+    {
+      "epoch": 1.9293603313391623,
+      "grad_norm": 0.9847011325418676,
+      "learning_rate": 3.0949469054816216e-08,
+      "loss": 0.384,
+      "num_input_tokens_seen": 482953616,
+      "step": 8385,
+      "train_runtime": 47087.476,
+      "train_tokens_per_second": 10256.519
+    },
+    {
+      "epoch": 1.9305108145421077,
+      "grad_norm": 0.8003529046579337,
+      "learning_rate": 2.995376881184841e-08,
+      "loss": 0.3653,
+      "num_input_tokens_seen": 483233032,
+      "step": 8390,
+      "train_runtime": 47119.0496,
+      "train_tokens_per_second": 10255.577
+    },
+    {
+      "epoch": 1.931661297745053,
+      "grad_norm": 0.8289552284207415,
+      "learning_rate": 2.8974300142620703e-08,
+      "loss": 0.3757,
+      "num_input_tokens_seen": 483515112,
+      "step": 8395,
+      "train_runtime": 47151.5056,
+      "train_tokens_per_second": 10254.5
+    },
+    {
+      "epoch": 1.932811780947998,
+      "grad_norm": 0.7680562827816213,
+      "learning_rate": 2.801106624596095e-08,
+      "loss": 0.393,
+      "num_input_tokens_seen": 483779848,
+      "step": 8400,
+      "train_runtime": 47177.5522,
+      "train_tokens_per_second": 10254.45
+    },
+    {
+      "epoch": 1.9339622641509435,
+      "grad_norm": 0.7954314453551361,
+      "learning_rate": 2.7064070267674434e-08,
+      "loss": 0.4037,
+      "num_input_tokens_seen": 484075256,
+      "step": 8405,
+      "train_runtime": 47195.2402,
+      "train_tokens_per_second": 10256.866
+    },
+    {
+      "epoch": 1.9351127473538887,
+      "grad_norm": 0.8178026395347452,
+      "learning_rate": 2.6133315300536644e-08,
+      "loss": 0.3578,
+      "num_input_tokens_seen": 484364760,
+      "step": 8410,
+      "train_runtime": 47221.1284,
+      "train_tokens_per_second": 10257.374
+    },
+    {
+      "epoch": 1.936263230556834,
+      "grad_norm": 1.2263739150022919,
+      "learning_rate": 2.5218804384281593e-08,
+      "loss": 0.392,
+      "num_input_tokens_seen": 484650464,
+      "step": 8415,
+      "train_runtime": 47248.7124,
+      "train_tokens_per_second": 10257.432
+    },
+    {
+      "epoch": 1.937413713759779,
+      "grad_norm": 0.8160520286055322,
+      "learning_rate": 2.432054050559185e-08,
+      "loss": 0.4059,
+      "num_input_tokens_seen": 484900328,
+      "step": 8420,
+      "train_runtime": 47274.5787,
+      "train_tokens_per_second": 10257.105
+    },
+    {
+      "epoch": 1.9385641969627243,
+      "grad_norm": 0.8170320560206203,
+      "learning_rate": 2.343852659808965e-08,
+      "loss": 0.3926,
+      "num_input_tokens_seen": 485177584,
+      "step": 8425,
+      "train_runtime": 47299.9485,
+      "train_tokens_per_second": 10257.465
+    },
+    {
+      "epoch": 1.9397146801656695,
+      "grad_norm": 0.8620705416318972,
+      "learning_rate": 2.2572765542325792e-08,
+      "loss": 0.3841,
+      "num_input_tokens_seen": 485484824,
+      "step": 8430,
+      "train_runtime": 47328.5824,
+      "train_tokens_per_second": 10257.751
+    },
+    {
+      "epoch": 1.9408651633686147,
+      "grad_norm": 0.9029809954062497,
+      "learning_rate": 2.1723260165771864e-08,
+      "loss": 0.4052,
+      "num_input_tokens_seen": 485781624,
+      "step": 8435,
+      "train_runtime": 47348.8107,
+      "train_tokens_per_second": 10259.637
+    },
+    {
+      "epoch": 1.94201564657156,
+      "grad_norm": 0.7992879085961136,
+      "learning_rate": 2.089001324281137e-08,
+      "loss": 0.3944,
+      "num_input_tokens_seen": 486058480,
+      "step": 8440,
+      "train_runtime": 47372.8622,
+      "train_tokens_per_second": 10260.273
+    },
+    {
+      "epoch": 1.9431661297745053,
+      "grad_norm": 0.8046974646493217,
+      "learning_rate": 2.007302749472695e-08,
+      "loss": 0.3948,
+      "num_input_tokens_seen": 486358784,
+      "step": 8445,
+      "train_runtime": 47404.1666,
+      "train_tokens_per_second": 10259.832
+    },
+    {
+      "epoch": 1.9443166129774505,
+      "grad_norm": 0.9544298535810893,
+      "learning_rate": 1.9272305589697615e-08,
+      "loss": 0.3794,
+      "num_input_tokens_seen": 486624304,
+      "step": 8450,
+      "train_runtime": 47433.6468,
+      "train_tokens_per_second": 10259.053
+    },
+    {
+      "epoch": 1.945467096180396,
+      "grad_norm": 0.9139772885141813,
+      "learning_rate": 1.848785014278376e-08,
+      "loss": 0.3814,
+      "num_input_tokens_seen": 486904120,
+      "step": 8455,
+      "train_runtime": 47460.6947,
+      "train_tokens_per_second": 10259.102
+    },
+    {
+      "epoch": 1.9466175793833411,
+      "grad_norm": 0.692814838661373,
+      "learning_rate": 1.7719663715922707e-08,
+      "loss": 0.3708,
+      "num_input_tokens_seen": 487216336,
+      "step": 8460,
+      "train_runtime": 47491.8315,
+      "train_tokens_per_second": 10258.95
+    },
+    {
+      "epoch": 1.9477680625862863,
+      "grad_norm": 0.75592300271525,
+      "learning_rate": 1.6967748817919292e-08,
+      "loss": 0.3835,
+      "num_input_tokens_seen": 487521152,
+      "step": 8465,
+      "train_runtime": 47520.2425,
+      "train_tokens_per_second": 10259.231
+    },
+    {
+      "epoch": 1.9489185457892315,
+      "grad_norm": 0.883595204527807,
+      "learning_rate": 1.6232107904437522e-08,
+      "loss": 0.3976,
+      "num_input_tokens_seen": 487808104,
+      "step": 8470,
+      "train_runtime": 47545.1244,
+      "train_tokens_per_second": 10259.898
+    },
+    {
+      "epoch": 1.9500690289921767,
+      "grad_norm": 0.6965740014816086,
+      "learning_rate": 1.551274337799169e-08,
+      "loss": 0.373,
+      "num_input_tokens_seen": 488114256,
+      "step": 8475,
+      "train_runtime": 47573.8568,
+      "train_tokens_per_second": 10260.136
+    },
+    {
+      "epoch": 1.951219512195122,
+      "grad_norm": 0.7608437388294115,
+      "learning_rate": 1.4809657587940286e-08,
+      "loss": 0.3555,
+      "num_input_tokens_seen": 488392768,
+      "step": 8480,
+      "train_runtime": 47606.9903,
+      "train_tokens_per_second": 10258.846
+    },
+    {
+      "epoch": 1.952369995398067,
+      "grad_norm": 0.8697102951311344,
+      "learning_rate": 1.4122852830475985e-08,
+      "loss": 0.3913,
+      "num_input_tokens_seen": 488675648,
+      "step": 8485,
+      "train_runtime": 47638.0907,
+      "train_tokens_per_second": 10258.086
+    },
+    {
+      "epoch": 1.9535204786010123,
+      "grad_norm": 0.773678877042324,
+      "learning_rate": 1.3452331348621784e-08,
+      "loss": 0.3882,
+      "num_input_tokens_seen": 488961192,
+      "step": 8490,
+      "train_runtime": 47663.8734,
+      "train_tokens_per_second": 10258.528
+    },
+    {
+      "epoch": 1.9546709618039575,
+      "grad_norm": 0.8935470744644787,
+      "learning_rate": 1.2798095332220429e-08,
+      "loss": 0.3523,
+      "num_input_tokens_seen": 489236840,
+      "step": 8495,
+      "train_runtime": 47692.4438,
+      "train_tokens_per_second": 10258.163
+    },
+    {
+      "epoch": 1.955821445006903,
+      "grad_norm": 0.8555134331534722,
+      "learning_rate": 1.2160146917927217e-08,
+      "loss": 0.3858,
+      "num_input_tokens_seen": 489512400,
+      "step": 8500,
+      "train_runtime": 47727.4763,
+      "train_tokens_per_second": 10256.407
+    },
+    {
+      "epoch": 1.956971928209848,
+      "grad_norm": 0.7962637383362804,
+      "learning_rate": 1.1538488189204999e-08,
+      "loss": 0.3862,
+      "num_input_tokens_seen": 489790624,
+      "step": 8505,
+      "train_runtime": 47756.4122,
+      "train_tokens_per_second": 10256.018
+    },
+    {
+      "epoch": 1.9581224114127935,
+      "grad_norm": 0.8565527014176711,
+      "learning_rate": 1.0933121176316952e-08,
+      "loss": 0.3547,
+      "num_input_tokens_seen": 490054336,
+      "step": 8510,
+      "train_runtime": 47786.8803,
+      "train_tokens_per_second": 10254.997
+    },
+    {
+      "epoch": 1.9592728946157387,
+      "grad_norm": 0.7629070153310843,
+      "learning_rate": 1.0344047856319373e-08,
+      "loss": 0.3908,
+      "num_input_tokens_seen": 490351664,
+      "step": 8515,
+      "train_runtime": 47811.796,
+      "train_tokens_per_second": 10255.872
+    },
+    {
+      "epoch": 1.960423377818684,
+      "grad_norm": 0.7825098826504876,
+      "learning_rate": 9.771270153053902e-09,
+      "loss": 0.3933,
+      "num_input_tokens_seen": 490635760,
+      "step": 8520,
+      "train_runtime": 47832.818,
+      "train_tokens_per_second": 10257.304
+    },
+    {
+      "epoch": 1.9615738610216291,
+      "grad_norm": 0.791824996517274,
+      "learning_rate": 9.214789937144753e-09,
+      "loss": 0.394,
+      "num_input_tokens_seen": 490917960,
+      "step": 8525,
+      "train_runtime": 47859.3575,
+      "train_tokens_per_second": 10257.513
+    },
+    {
+      "epoch": 1.9627243442245743,
+      "grad_norm": 0.7867492001274388,
+      "learning_rate": 8.674609025988712e-09,
+      "loss": 0.3938,
+      "num_input_tokens_seen": 491213880,
+      "step": 8530,
+      "train_runtime": 47884.3215,
+      "train_tokens_per_second": 10258.345
+    },
+    {
+      "epoch": 1.9638748274275195,
+      "grad_norm": 0.8096417194051213,
+      "learning_rate": 8.150729183751815e-09,
+      "loss": 0.3868,
+      "num_input_tokens_seen": 491502632,
+      "step": 8535,
+      "train_runtime": 47918.3688,
+      "train_tokens_per_second": 10257.082
+    },
+    {
+      "epoch": 1.9650253106304647,
+      "grad_norm": 0.7481506136655819,
+      "learning_rate": 7.643152121362685e-09,
+      "loss": 0.3961,
+      "num_input_tokens_seen": 491805896,
+      "step": 8540,
+      "train_runtime": 47944.8088,
+      "train_tokens_per_second": 10257.751
+    },
+    {
+      "epoch": 1.96617579383341,
+      "grad_norm": 0.8059784898998039,
+      "learning_rate": 7.15187949650753e-09,
+      "loss": 0.3771,
+      "num_input_tokens_seen": 492086080,
+      "step": 8545,
+      "train_runtime": 47976.9156,
+      "train_tokens_per_second": 10256.726
+    },
+    {
+      "epoch": 1.9673262770363553,
+      "grad_norm": 0.9106943795768503,
+      "learning_rate": 6.676912913622935e-09,
+      "loss": 0.4207,
+      "num_input_tokens_seen": 492376880,
+      "step": 8550,
+      "train_runtime": 48007.2174,
+      "train_tokens_per_second": 10256.309
+    },
+    {
+      "epoch": 1.9684767602393005,
+      "grad_norm": 0.8246606907074029,
+      "learning_rate": 6.21825392389308e-09,
+      "loss": 0.3968,
+      "num_input_tokens_seen": 492660496,
+      "step": 8555,
+      "train_runtime": 48036.7937,
+      "train_tokens_per_second": 10255.899
+    },
+    {
+      "epoch": 1.9696272434422457,
+      "grad_norm": 0.7621481366508165,
+      "learning_rate": 5.775904025243084e-09,
+      "loss": 0.4009,
+      "num_input_tokens_seen": 492963808,
+      "step": 8560,
+      "train_runtime": 48063.1122,
+      "train_tokens_per_second": 10256.594
+    },
+    {
+      "epoch": 1.9707777266451911,
+      "grad_norm": 0.7952642087678853,
+      "learning_rate": 5.349864662334558e-09,
+      "loss": 0.3963,
+      "num_input_tokens_seen": 493244248,
+      "step": 8565,
+      "train_runtime": 48090.6958,
+      "train_tokens_per_second": 10256.542
+    },
+    {
+      "epoch": 1.9719282098481363,
+      "grad_norm": 0.7887362349168614,
+      "learning_rate": 4.940137226560615e-09,
+      "loss": 0.3765,
+      "num_input_tokens_seen": 493519304,
+      "step": 8570,
+      "train_runtime": 48116.832,
+      "train_tokens_per_second": 10256.687
+    },
+    {
+      "epoch": 1.9730786930510815,
+      "grad_norm": 0.7913338889793305,
+      "learning_rate": 4.54672305604198e-09,
+      "loss": 0.3863,
+      "num_input_tokens_seen": 493822184,
+      "step": 8575,
+      "train_runtime": 48144.7678,
+      "train_tokens_per_second": 10257.027
+    },
+    {
+      "epoch": 1.9742291762540267,
+      "grad_norm": 0.7180879947394211,
+      "learning_rate": 4.16962343562255e-09,
+      "loss": 0.3877,
+      "num_input_tokens_seen": 494109296,
+      "step": 8580,
+      "train_runtime": 48175.7331,
+      "train_tokens_per_second": 10256.394
+    },
+    {
+      "epoch": 1.975379659456972,
+      "grad_norm": 0.8016187291694565,
+      "learning_rate": 3.8088395968638445e-09,
+      "loss": 0.3612,
+      "num_input_tokens_seen": 494383056,
+      "step": 8585,
+      "train_runtime": 48200.9292,
+      "train_tokens_per_second": 10256.712
+    },
+    {
+      "epoch": 1.976530142659917,
+      "grad_norm": 0.8551888606413809,
+      "learning_rate": 3.4643727180427854e-09,
+      "loss": 0.3835,
+      "num_input_tokens_seen": 494678672,
+      "step": 8590,
+      "train_runtime": 48228.8805,
+      "train_tokens_per_second": 10256.897
+    },
+    {
+      "epoch": 1.9776806258628623,
+      "grad_norm": 0.748557175187064,
+      "learning_rate": 3.136223924146142e-09,
+      "loss": 0.3729,
+      "num_input_tokens_seen": 494954192,
+      "step": 8595,
+      "train_runtime": 48255.1377,
+      "train_tokens_per_second": 10257.026
+    },
+    {
+      "epoch": 1.9788311090658075,
+      "grad_norm": 0.7685745940093629,
+      "learning_rate": 2.8243942868694253e-09,
+      "loss": 0.3928,
+      "num_input_tokens_seen": 495228832,
+      "step": 8600,
+      "train_runtime": 48283.0749,
+      "train_tokens_per_second": 10256.779
+    },
+    {
+      "epoch": 1.979981592268753,
+      "grad_norm": 0.789467510464664,
+      "learning_rate": 2.5288848246102225e-09,
+      "loss": 0.3884,
+      "num_input_tokens_seen": 495511320,
+      "step": 8605,
+      "train_runtime": 48312.1027,
+      "train_tokens_per_second": 10256.464
+    },
+    {
+      "epoch": 1.9811320754716981,
+      "grad_norm": 0.913157819991065,
+      "learning_rate": 2.2496965024670913e-09,
+      "loss": 0.3739,
+      "num_input_tokens_seen": 495783824,
+      "step": 8610,
+      "train_runtime": 48341.6714,
+      "train_tokens_per_second": 10255.827
+    },
+    {
+      "epoch": 1.9822825586746433,
+      "grad_norm": 0.89091253586579,
+      "learning_rate": 1.986830232235115e-09,
+      "loss": 0.3833,
+      "num_input_tokens_seen": 496066608,
+      "step": 8615,
+      "train_runtime": 48363.3324,
+      "train_tokens_per_second": 10257.081
+    },
+    {
+      "epoch": 1.9834330418775887,
+      "grad_norm": 1.0581304470080295,
+      "learning_rate": 1.7402868724042399e-09,
+      "loss": 0.3877,
+      "num_input_tokens_seen": 496355552,
+      "step": 8620,
+      "train_runtime": 48383.715,
+      "train_tokens_per_second": 10258.732
+    },
+    {
+      "epoch": 1.984583525080534,
+      "grad_norm": 0.7608002031024231,
+      "learning_rate": 1.5100672281553874e-09,
+      "loss": 0.3964,
+      "num_input_tokens_seen": 496634080,
+      "step": 8625,
+      "train_runtime": 48416.6493,
+      "train_tokens_per_second": 10257.506
+    },
+    {
+      "epoch": 1.9857340082834791,
+      "grad_norm": 0.8498719833699132,
+      "learning_rate": 1.2961720513587905e-09,
+      "loss": 0.3961,
+      "num_input_tokens_seen": 496912632,
+      "step": 8630,
+      "train_runtime": 48445.4523,
+      "train_tokens_per_second": 10257.157
+    },
+    {
+      "epoch": 1.9868844914864243,
+      "grad_norm": 0.7503179889145187,
+      "learning_rate": 1.0986020405695518e-09,
+      "loss": 0.3761,
+      "num_input_tokens_seen": 497176200,
+      "step": 8635,
+      "train_runtime": 48471.986,
+      "train_tokens_per_second": 10256.98
+    },
+    {
+      "epoch": 1.9880349746893695,
+      "grad_norm": 0.871238499988888,
+      "learning_rate": 9.173578410281992e-10,
+      "loss": 0.3906,
+      "num_input_tokens_seen": 497450808,
+      "step": 8640,
+      "train_runtime": 48500.2273,
+      "train_tokens_per_second": 10256.67
+    },
+    {
+      "epoch": 1.9891854578923147,
+      "grad_norm": 0.8156359789785304,
+      "learning_rate": 7.524400446562441e-10,
+      "loss": 0.4167,
+      "num_input_tokens_seen": 497752720,
+      "step": 8645,
+      "train_runtime": 48533.683,
+      "train_tokens_per_second": 10255.82
+    },
+    {
+      "epoch": 1.99033594109526,
+      "grad_norm": 0.9096881118755317,
+      "learning_rate": 6.038491900556276e-10,
+      "loss": 0.3794,
+      "num_input_tokens_seen": 498048920,
+      "step": 8650,
+      "train_runtime": 48566.7759,
+      "train_tokens_per_second": 10254.931
+    },
+    {
+      "epoch": 1.991486424298205,
+      "grad_norm": 0.9950407406477443,
+      "learning_rate": 4.715857625059439e-10,
+      "loss": 0.3869,
+      "num_input_tokens_seen": 498339712,
+      "step": 8655,
+      "train_runtime": 48593.9537,
+      "train_tokens_per_second": 10255.179
+    },
+    {
+      "epoch": 1.9926369075011505,
+      "grad_norm": 0.7583342822774866,
+      "learning_rate": 3.5565019396333057e-10,
+      "loss": 0.3764,
+      "num_input_tokens_seen": 498628256,
+      "step": 8660,
+      "train_runtime": 48624.7829,
+      "train_tokens_per_second": 10254.611
+    },
+    {
+      "epoch": 1.9937873907040957,
+      "grad_norm": 0.7767256916064607,
+      "learning_rate": 2.5604286305991324e-10,
+      "loss": 0.4064,
+      "num_input_tokens_seen": 498923096,
+      "step": 8665,
+      "train_runtime": 48650.4781,
+      "train_tokens_per_second": 10255.256
+    },
+    {
+      "epoch": 1.994937873907041,
+      "grad_norm": 0.8292515455738418,
+      "learning_rate": 1.727640951015852e-10,
+      "loss": 0.3625,
+      "num_input_tokens_seen": 499229936,
+      "step": 8670,
+      "train_runtime": 48675.1529,
+      "train_tokens_per_second": 10256.361
+    },
+    {
+      "epoch": 1.9960883571099863,
+      "grad_norm": 0.7413620813035472,
+      "learning_rate": 1.058141620663422e-10,
+      "loss": 0.3954,
+      "num_input_tokens_seen": 499535560,
+      "step": 8675,
+      "train_runtime": 48701.1564,
+      "train_tokens_per_second": 10257.16
+    },
+    {
+      "epoch": 1.9972388403129315,
+      "grad_norm": 0.8019064331319312,
+      "learning_rate": 5.519328260428225e-11,
+      "loss": 0.383,
+      "num_input_tokens_seen": 499839616,
+      "step": 8680,
+      "train_runtime": 48730.6632,
+      "train_tokens_per_second": 10257.189
+    },
+    {
+      "epoch": 1.9983893235158767,
+      "grad_norm": 0.721707415102053,
+      "learning_rate": 2.0901622037605708e-11,
+      "loss": 0.3829,
+      "num_input_tokens_seen": 500113816,
+      "step": 8685,
+      "train_runtime": 48755.5927,
+      "train_tokens_per_second": 10257.568
+    },
+    {
+      "epoch": 1.999539806718822,
+      "grad_norm": 0.770545613623502,
+      "learning_rate": 2.939292358949963e-12,
+      "loss": 0.386,
+      "num_input_tokens_seen": 500379792,
+      "step": 8690,
+      "train_runtime": 48780.9508,
+      "train_tokens_per_second": 10257.688
+    },
+    {
+      "epoch": 2.0,
+      "num_input_tokens_seen": 500486680,
+      "step": 8692,
+      "total_flos": 925572038246400.0,
+      "train_loss": 0.4423057133433899,
+      "train_runtime": 48898.4483,
+      "train_samples_per_second": 0.711,
+      "train_steps_per_second": 0.178
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 8692,
+  "num_input_tokens_seen": 500486680,
+  "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": 925572038246400.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}