diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4684 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 2322,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004306632213608958,
+      "grad_norm": 1.5113693870546838,
+      "learning_rate": 9.999926779237713e-06,
+      "loss": 0.7518440246582031,
+      "num_input_tokens_seen": 516464,
+      "step": 5,
+      "train_runtime": 80.3069,
+      "train_tokens_per_second": 6431.131
+    },
+    {
+      "epoch": 0.008613264427217916,
+      "grad_norm": 1.1099479933103011,
+      "learning_rate": 9.999629323566323e-06,
+      "loss": 0.5895246028900146,
+      "num_input_tokens_seen": 1050992,
+      "step": 10,
+      "train_runtime": 156.0878,
+      "train_tokens_per_second": 6733.34
+    },
+    {
+      "epoch": 0.012919896640826873,
+      "grad_norm": 0.9566198183888653,
+      "learning_rate": 9.999103070290155e-06,
+      "loss": 0.5217817306518555,
+      "num_input_tokens_seen": 1591440,
+      "step": 15,
+      "train_runtime": 221.4671,
+      "train_tokens_per_second": 7185.897
+    },
+    {
+      "epoch": 0.017226528854435832,
+      "grad_norm": 0.8711180504651662,
+      "learning_rate": 9.998348043492093e-06,
+      "loss": 0.5254570960998535,
+      "num_input_tokens_seen": 2104328,
+      "step": 20,
+      "train_runtime": 281.0116,
+      "train_tokens_per_second": 7488.403
+    },
+    {
+      "epoch": 0.02153316106804479,
+      "grad_norm": 0.7568221566958359,
+      "learning_rate": 9.997364277724362e-06,
+      "loss": 0.49294347763061525,
+      "num_input_tokens_seen": 2659352,
+      "step": 25,
+      "train_runtime": 342.2303,
+      "train_tokens_per_second": 7770.65
+    },
+    {
+      "epoch": 0.025839793281653745,
+      "grad_norm": 1.110633668053048,
+      "learning_rate": 9.996151818006951e-06,
+      "loss": 0.4858407020568848,
+      "num_input_tokens_seen": 3199536,
+      "step": 30,
+      "train_runtime": 411.1034,
+      "train_tokens_per_second": 7782.802
+    },
+    {
+      "epoch": 0.030146425495262703,
+      "grad_norm": 1.307486068113022,
+      "learning_rate": 9.99471071982555e-06,
+      "loss": 0.4974493980407715,
+      "num_input_tokens_seen": 3665456,
+      "step": 35,
+      "train_runtime": 477.4255,
+      "train_tokens_per_second": 7677.545
+    },
+    {
+      "epoch": 0.034453057708871665,
+      "grad_norm": 0.7354991482453888,
+      "learning_rate": 9.993041049129005e-06,
+      "loss": 0.4639857292175293,
+      "num_input_tokens_seen": 4181744,
+      "step": 40,
+      "train_runtime": 548.6431,
+      "train_tokens_per_second": 7621.974
+    },
+    {
+      "epoch": 0.03875968992248062,
+      "grad_norm": 0.6823446795404668,
+      "learning_rate": 9.99114288232632e-06,
+      "loss": 0.4471879482269287,
+      "num_input_tokens_seen": 4725304,
+      "step": 45,
+      "train_runtime": 625.6381,
+      "train_tokens_per_second": 7552.775
+    },
+    {
+      "epoch": 0.04306632213608958,
+      "grad_norm": 0.6187958236434659,
+      "learning_rate": 9.989016306283132e-06,
+      "loss": 0.4194998264312744,
+      "num_input_tokens_seen": 5217928,
+      "step": 50,
+      "train_runtime": 695.1978,
+      "train_tokens_per_second": 7505.674
+    },
+    {
+      "epoch": 0.047372954349698536,
+      "grad_norm": 0.7217736924185857,
+      "learning_rate": 9.986661418317759e-06,
+      "loss": 0.4558732509613037,
+      "num_input_tokens_seen": 5720264,
+      "step": 55,
+      "train_runtime": 762.7465,
+      "train_tokens_per_second": 7499.561
+    },
+    {
+      "epoch": 0.05167958656330749,
+      "grad_norm": 1.1046030739670325,
+      "learning_rate": 9.984078326196736e-06,
+      "loss": 0.4496774673461914,
+      "num_input_tokens_seen": 6200784,
+      "step": 60,
+      "train_runtime": 829.9567,
+      "train_tokens_per_second": 7471.213
+    },
+    {
+      "epoch": 0.05598621877691645,
+      "grad_norm": 0.7074096009153457,
+      "learning_rate": 9.981267148129884e-06,
+      "loss": 0.4551366329193115,
+      "num_input_tokens_seen": 6721120,
+      "step": 65,
+      "train_runtime": 901.8675,
+      "train_tokens_per_second": 7452.448
+    },
+    {
+      "epoch": 0.06029285099052541,
+      "grad_norm": 0.6980358820573814,
+      "learning_rate": 9.978228012764904e-06,
+      "loss": 0.43771700859069823,
+      "num_input_tokens_seen": 7202096,
+      "step": 70,
+      "train_runtime": 965.2084,
+      "train_tokens_per_second": 7461.7
+    },
+    {
+      "epoch": 0.06459948320413436,
+      "grad_norm": 0.70967612822,
+      "learning_rate": 9.974961059181482e-06,
+      "loss": 0.44599485397338867,
+      "num_input_tokens_seen": 7721720,
+      "step": 75,
+      "train_runtime": 1035.2088,
+      "train_tokens_per_second": 7459.094
+    },
+    {
+      "epoch": 0.06890611541774333,
+      "grad_norm": 0.8195531593273334,
+      "learning_rate": 9.971466436884933e-06,
+      "loss": 0.4492579460144043,
+      "num_input_tokens_seen": 8215480,
+      "step": 80,
+      "train_runtime": 1100.699,
+      "train_tokens_per_second": 7463.875
+    },
+    {
+      "epoch": 0.07321274763135228,
+      "grad_norm": 0.6611353263896688,
+      "learning_rate": 9.967744305799358e-06,
+      "loss": 0.428343677520752,
+      "num_input_tokens_seen": 8751608,
+      "step": 85,
+      "train_runtime": 1175.1956,
+      "train_tokens_per_second": 7446.937
+    },
+    {
+      "epoch": 0.07751937984496124,
+      "grad_norm": 0.7344023965501802,
+      "learning_rate": 9.963794836260316e-06,
+      "loss": 0.44635858535766604,
+      "num_input_tokens_seen": 9220400,
+      "step": 90,
+      "train_runtime": 1249.0723,
+      "train_tokens_per_second": 7381.798
+    },
+    {
+      "epoch": 0.0818260120585702,
+      "grad_norm": 0.7993384059251473,
+      "learning_rate": 9.959618209007045e-06,
+      "loss": 0.41611676216125487,
+      "num_input_tokens_seen": 9752480,
+      "step": 95,
+      "train_runtime": 1328.8505,
+      "train_tokens_per_second": 7339.035
+    },
+    {
+      "epoch": 0.08613264427217916,
+      "grad_norm": 0.692666755511038,
+      "learning_rate": 9.955214615174174e-06,
+      "loss": 0.4301795959472656,
+      "num_input_tokens_seen": 10286000,
+      "step": 100,
+      "train_runtime": 1397.8403,
+      "train_tokens_per_second": 7358.494
+    },
+    {
+      "epoch": 0.09043927648578812,
+      "grad_norm": 0.8306336622178832,
+      "learning_rate": 9.950584256282988e-06,
+      "loss": 0.42357568740844725,
+      "num_input_tokens_seen": 10786992,
+      "step": 105,
+      "train_runtime": 1465.2741,
+      "train_tokens_per_second": 7361.757
+    },
+    {
+      "epoch": 0.09474590869939707,
+      "grad_norm": 0.7246154301219507,
+      "learning_rate": 9.945727344232202e-06,
+      "loss": 0.43195395469665526,
+      "num_input_tokens_seen": 11324728,
+      "step": 110,
+      "train_runtime": 1543.9809,
+      "train_tokens_per_second": 7334.759
+    },
+    {
+      "epoch": 0.09905254091300603,
+      "grad_norm": 0.7133273160650513,
+      "learning_rate": 9.940644101288259e-06,
+      "loss": 0.4098663806915283,
+      "num_input_tokens_seen": 11842384,
+      "step": 115,
+      "train_runtime": 1610.1312,
+      "train_tokens_per_second": 7354.919
+    },
+    {
+      "epoch": 0.10335917312661498,
+      "grad_norm": 0.7454776319580722,
+      "learning_rate": 9.93533476007517e-06,
+      "loss": 0.4260974884033203,
+      "num_input_tokens_seen": 12395312,
+      "step": 120,
+      "train_runtime": 1679.6709,
+      "train_tokens_per_second": 7379.607
+    },
+    {
+      "epoch": 0.10766580534022395,
+      "grad_norm": 0.7881403926209154,
+      "learning_rate": 9.929799563563858e-06,
+      "loss": 0.41834230422973634,
+      "num_input_tokens_seen": 12876440,
+      "step": 125,
+      "train_runtime": 1748.4163,
+      "train_tokens_per_second": 7364.63
+    },
+    {
+      "epoch": 0.1119724375538329,
+      "grad_norm": 0.7096308914101791,
+      "learning_rate": 9.924038765061042e-06,
+      "loss": 0.4106014251708984,
+      "num_input_tokens_seen": 13376496,
+      "step": 130,
+      "train_runtime": 1818.2885,
+      "train_tokens_per_second": 7356.641
+    },
+    {
+      "epoch": 0.11627906976744186,
+      "grad_norm": 0.7276233917576757,
+      "learning_rate": 9.918052628197645e-06,
+      "loss": 0.4307220458984375,
+      "num_input_tokens_seen": 13872280,
+      "step": 135,
+      "train_runtime": 1879.0083,
+      "train_tokens_per_second": 7382.767
+    },
+    {
+      "epoch": 0.12058570198105081,
+      "grad_norm": 0.6304682702555895,
+      "learning_rate": 9.911841426916733e-06,
+      "loss": 0.4186269760131836,
+      "num_input_tokens_seen": 14373496,
+      "step": 140,
+      "train_runtime": 1948.9705,
+      "train_tokens_per_second": 7374.917
+    },
+    {
+      "epoch": 0.12489233419465978,
+      "grad_norm": 0.7457639995354279,
+      "learning_rate": 9.905405445460972e-06,
+      "loss": 0.41518564224243165,
+      "num_input_tokens_seen": 14886552,
+      "step": 145,
+      "train_runtime": 2013.9467,
+      "train_tokens_per_second": 7391.731
+    },
+    {
+      "epoch": 0.12919896640826872,
+      "grad_norm": 0.7089829747760644,
+      "learning_rate": 9.898744978359627e-06,
+      "loss": 0.40087127685546875,
+      "num_input_tokens_seen": 15364928,
+      "step": 150,
+      "train_runtime": 2080.6533,
+      "train_tokens_per_second": 7384.665
+    },
+    {
+      "epoch": 0.1335055986218777,
+      "grad_norm": 0.7908692602439755,
+      "learning_rate": 9.891860330415077e-06,
+      "loss": 0.4161245346069336,
+      "num_input_tokens_seen": 15877064,
+      "step": 155,
+      "train_runtime": 2148.1134,
+      "train_tokens_per_second": 7391.166
+    },
+    {
+      "epoch": 0.13781223083548666,
+      "grad_norm": 0.7336628341974487,
+      "learning_rate": 9.884751816688873e-06,
+      "loss": 0.38726229667663575,
+      "num_input_tokens_seen": 16362424,
+      "step": 160,
+      "train_runtime": 2219.2408,
+      "train_tokens_per_second": 7372.983
+    },
+    {
+      "epoch": 0.1421188630490956,
+      "grad_norm": 0.7377502319782788,
+      "learning_rate": 9.877419762487312e-06,
+      "loss": 0.41555061340332033,
+      "num_input_tokens_seen": 16884272,
+      "step": 165,
+      "train_runtime": 2285.0222,
+      "train_tokens_per_second": 7389.106
+    },
+    {
+      "epoch": 0.14642549526270457,
+      "grad_norm": 0.763519054939035,
+      "learning_rate": 9.869864503346557e-06,
+      "loss": 0.42660012245178225,
+      "num_input_tokens_seen": 17395904,
+      "step": 170,
+      "train_runtime": 2353.6081,
+      "train_tokens_per_second": 7391.164
+    },
+    {
+      "epoch": 0.1507321274763135,
+      "grad_norm": 0.9454978823437594,
+      "learning_rate": 9.862086385017283e-06,
+      "loss": 0.3985039472579956,
+      "num_input_tokens_seen": 17935128,
+      "step": 175,
+      "train_runtime": 2425.5746,
+      "train_tokens_per_second": 7394.177
+    },
+    {
+      "epoch": 0.15503875968992248,
+      "grad_norm": 0.897983992254847,
+      "learning_rate": 9.854085763448844e-06,
+      "loss": 0.403258228302002,
+      "num_input_tokens_seen": 18441240,
+      "step": 180,
+      "train_runtime": 2491.2198,
+      "train_tokens_per_second": 7402.494
+    },
+    {
+      "epoch": 0.15934539190353145,
+      "grad_norm": 0.7028674441116541,
+      "learning_rate": 9.845863004772994e-06,
+      "loss": 0.4074747085571289,
+      "num_input_tokens_seen": 18950248,
+      "step": 185,
+      "train_runtime": 2562.4101,
+      "train_tokens_per_second": 7395.478
+    },
+    {
+      "epoch": 0.1636520241171404,
+      "grad_norm": 0.6890074181556956,
+      "learning_rate": 9.837418485287126e-06,
+      "loss": 0.3952253580093384,
+      "num_input_tokens_seen": 19443272,
+      "step": 190,
+      "train_runtime": 2629.3175,
+      "train_tokens_per_second": 7394.798
+    },
+    {
+      "epoch": 0.16795865633074936,
+      "grad_norm": 0.6864055327668929,
+      "learning_rate": 9.82875259143706e-06,
+      "loss": 0.4341114521026611,
+      "num_input_tokens_seen": 19974056,
+      "step": 195,
+      "train_runtime": 2701.8046,
+      "train_tokens_per_second": 7392.857
+    },
+    {
+      "epoch": 0.17226528854435832,
+      "grad_norm": 0.7549452956748491,
+      "learning_rate": 9.819865719799344e-06,
+      "loss": 0.39603476524353026,
+      "num_input_tokens_seen": 20505304,
+      "step": 200,
+      "train_runtime": 2760.8615,
+      "train_tokens_per_second": 7427.14
+    },
+    {
+      "epoch": 0.17657192075796727,
+      "grad_norm": 0.7461369795730732,
+      "learning_rate": 9.81075827706312e-06,
+      "loss": 0.3957651615142822,
+      "num_input_tokens_seen": 21041488,
+      "step": 205,
+      "train_runtime": 2833.0985,
+      "train_tokens_per_second": 7427.023
+    },
+    {
+      "epoch": 0.18087855297157623,
+      "grad_norm": 0.6726317666151067,
+      "learning_rate": 9.801430680011499e-06,
+      "loss": 0.3961822509765625,
+      "num_input_tokens_seen": 21525384,
+      "step": 210,
+      "train_runtime": 2901.9507,
+      "train_tokens_per_second": 7417.557
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 0.7153134316038106,
+      "learning_rate": 9.791883355502503e-06,
+      "loss": 0.42037005424499513,
+      "num_input_tokens_seen": 22015656,
+      "step": 215,
+      "train_runtime": 2972.2855,
+      "train_tokens_per_second": 7406.979
+    },
+    {
+      "epoch": 0.18949181739879414,
+      "grad_norm": 0.7441983537016906,
+      "learning_rate": 9.782116740449515e-06,
+      "loss": 0.3965612888336182,
+      "num_input_tokens_seen": 22529696,
+      "step": 220,
+      "train_runtime": 3044.2115,
+      "train_tokens_per_second": 7400.831
+    },
+    {
+      "epoch": 0.1937984496124031,
+      "grad_norm": 0.7350628011367314,
+      "learning_rate": 9.7721312818013e-06,
+      "loss": 0.44352130889892577,
+      "num_input_tokens_seen": 23038360,
+      "step": 225,
+      "train_runtime": 3116.1306,
+      "train_tokens_per_second": 7393.259
+    },
+    {
+      "epoch": 0.19810508182601205,
+      "grad_norm": 0.7563228965004609,
+      "learning_rate": 9.761927436521534e-06,
+      "loss": 0.38213863372802737,
+      "num_input_tokens_seen": 23532864,
+      "step": 230,
+      "train_runtime": 3179.4752,
+      "train_tokens_per_second": 7401.493
+    },
+    {
+      "epoch": 0.20241171403962102,
+      "grad_norm": 0.775450865404374,
+      "learning_rate": 9.751505671567914e-06,
+      "loss": 0.3937716484069824,
+      "num_input_tokens_seen": 24005936,
+      "step": 235,
+      "train_runtime": 3253.9708,
+      "train_tokens_per_second": 7377.428
+    },
+    {
+      "epoch": 0.20671834625322996,
+      "grad_norm": 0.7250225416398999,
+      "learning_rate": 9.740866463870762e-06,
+      "loss": 0.40571026802062987,
+      "num_input_tokens_seen": 24490872,
+      "step": 240,
+      "train_runtime": 3322.9821,
+      "train_tokens_per_second": 7370.149
+    },
+    {
+      "epoch": 0.21102497846683893,
+      "grad_norm": 0.7114153264209351,
+      "learning_rate": 9.730010300311226e-06,
+      "loss": 0.411739444732666,
+      "num_input_tokens_seen": 25014232,
+      "step": 245,
+      "train_runtime": 3392.0309,
+      "train_tokens_per_second": 7374.412
+    },
+    {
+      "epoch": 0.2153316106804479,
+      "grad_norm": 0.7013109032284867,
+      "learning_rate": 9.718937677698976e-06,
+      "loss": 0.40146331787109374,
+      "num_input_tokens_seen": 25507448,
+      "step": 250,
+      "train_runtime": 3457.5164,
+      "train_tokens_per_second": 7377.39
+    },
+    {
+      "epoch": 0.21963824289405684,
+      "grad_norm": 0.6984434047541229,
+      "learning_rate": 9.707649102749488e-06,
+      "loss": 0.3829845428466797,
+      "num_input_tokens_seen": 26003584,
+      "step": 255,
+      "train_runtime": 3526.0072,
+      "train_tokens_per_second": 7374.796
+    },
+    {
+      "epoch": 0.2239448751076658,
+      "grad_norm": 0.6930588317334605,
+      "learning_rate": 9.696145092060843e-06,
+      "loss": 0.37958288192749023,
+      "num_input_tokens_seen": 26523584,
+      "step": 260,
+      "train_runtime": 3586.5787,
+      "train_tokens_per_second": 7395.233
+    },
+    {
+      "epoch": 0.22825150732127478,
+      "grad_norm": 0.7883661370197577,
+      "learning_rate": 9.684426172090084e-06,
+      "loss": 0.4073479652404785,
+      "num_input_tokens_seen": 27002040,
+      "step": 265,
+      "train_runtime": 3663.8397,
+      "train_tokens_per_second": 7369.875
+    },
+    {
+      "epoch": 0.23255813953488372,
+      "grad_norm": 0.6795095661344687,
+      "learning_rate": 9.672492879129136e-06,
+      "loss": 0.40698766708374023,
+      "num_input_tokens_seen": 27505176,
+      "step": 270,
+      "train_runtime": 3735.7865,
+      "train_tokens_per_second": 7362.62
+    },
+    {
+      "epoch": 0.2368647717484927,
+      "grad_norm": 0.9228740109209783,
+      "learning_rate": 9.660345759280254e-06,
+      "loss": 0.4029866695404053,
+      "num_input_tokens_seen": 28022920,
+      "step": 275,
+      "train_runtime": 3801.2628,
+      "train_tokens_per_second": 7372.003
+    },
+    {
+      "epoch": 0.24117140396210163,
+      "grad_norm": 1.0690229200759769,
+      "learning_rate": 9.647985368431031e-06,
+      "loss": 0.4014443397521973,
+      "num_input_tokens_seen": 28534944,
+      "step": 280,
+      "train_runtime": 3879.684,
+      "train_tokens_per_second": 7354.966
+    },
+    {
+      "epoch": 0.2454780361757106,
+      "grad_norm": 0.7476961212143748,
+      "learning_rate": 9.63541227222897e-06,
+      "loss": 0.4226128101348877,
+      "num_input_tokens_seen": 29035280,
+      "step": 285,
+      "train_runtime": 3948.0122,
+      "train_tokens_per_second": 7354.405
+    },
+    {
+      "epoch": 0.24978466838931956,
+      "grad_norm": 0.7955697655841107,
+      "learning_rate": 9.622627046055584e-06,
+      "loss": 0.4024317264556885,
+      "num_input_tokens_seen": 29597328,
+      "step": 290,
+      "train_runtime": 4029.1263,
+      "train_tokens_per_second": 7345.843
+    },
+    {
+      "epoch": 0.2540913006029285,
+      "grad_norm": 0.6894431613335288,
+      "learning_rate": 9.609630275000072e-06,
+      "loss": 0.3940183162689209,
+      "num_input_tokens_seen": 30094560,
+      "step": 295,
+      "train_runtime": 4100.0206,
+      "train_tokens_per_second": 7340.1
+    },
+    {
+      "epoch": 0.25839793281653745,
+      "grad_norm": 0.6990387473878472,
+      "learning_rate": 9.596422553832547e-06,
+      "loss": 0.3797435760498047,
+      "num_input_tokens_seen": 30628392,
+      "step": 300,
+      "train_runtime": 4171.3368,
+      "train_tokens_per_second": 7342.584
+    },
+    {
+      "epoch": 0.26270456503014644,
+      "grad_norm": 0.740391991800684,
+      "learning_rate": 9.583004486976813e-06,
+      "loss": 0.3960827350616455,
+      "num_input_tokens_seen": 31155408,
+      "step": 305,
+      "train_runtime": 4239.711,
+      "train_tokens_per_second": 7348.474
+    },
+    {
+      "epoch": 0.2670111972437554,
+      "grad_norm": 0.6842903856341336,
+      "learning_rate": 9.5693766884827e-06,
+      "loss": 0.38438570499420166,
+      "num_input_tokens_seen": 31648472,
+      "step": 310,
+      "train_runtime": 4304.3552,
+      "train_tokens_per_second": 7352.663
+    },
+    {
+      "epoch": 0.2713178294573643,
+      "grad_norm": 0.7452536578366745,
+      "learning_rate": 9.555539781997978e-06,
+      "loss": 0.4010166645050049,
+      "num_input_tokens_seen": 32148440,
+      "step": 315,
+      "train_runtime": 4375.0353,
+      "train_tokens_per_second": 7348.156
+    },
+    {
+      "epoch": 0.2756244616709733,
+      "grad_norm": 0.6599734525587321,
+      "learning_rate": 9.541494400739799e-06,
+      "loss": 0.3938159465789795,
+      "num_input_tokens_seen": 32659456,
+      "step": 320,
+      "train_runtime": 4449.4205,
+      "train_tokens_per_second": 7340.159
+    },
+    {
+      "epoch": 0.27993109388458226,
+      "grad_norm": 0.6790075507627891,
+      "learning_rate": 9.527241187465735e-06,
+      "loss": 0.39118094444274903,
+      "num_input_tokens_seen": 33157552,
+      "step": 325,
+      "train_runtime": 4514.8539,
+      "train_tokens_per_second": 7344.103
+    },
+    {
+      "epoch": 0.2842377260981912,
+      "grad_norm": 0.6400320619933816,
+      "learning_rate": 9.51278079444435e-06,
+      "loss": 0.38675827980041505,
+      "num_input_tokens_seen": 33687624,
+      "step": 330,
+      "train_runtime": 4579.3488,
+      "train_tokens_per_second": 7356.422
+    },
+    {
+      "epoch": 0.2885443583118002,
+      "grad_norm": 0.8011737268019301,
+      "learning_rate": 9.498113883425364e-06,
+      "loss": 0.4036139488220215,
+      "num_input_tokens_seen": 34171792,
+      "step": 335,
+      "train_runtime": 4653.543,
+      "train_tokens_per_second": 7343.177
+    },
+    {
+      "epoch": 0.29285099052540914,
+      "grad_norm": 0.6748924299050266,
+      "learning_rate": 9.483241125609358e-06,
+      "loss": 0.41157045364379885,
+      "num_input_tokens_seen": 34691536,
+      "step": 340,
+      "train_runtime": 4729.7805,
+      "train_tokens_per_second": 7334.703
+    },
+    {
+      "epoch": 0.2971576227390181,
+      "grad_norm": 0.6981892601742133,
+      "learning_rate": 9.468163201617063e-06,
+      "loss": 0.42139220237731934,
+      "num_input_tokens_seen": 35206192,
+      "step": 345,
+      "train_runtime": 4801.0436,
+      "train_tokens_per_second": 7333.029
+    },
+    {
+      "epoch": 0.301464254952627,
+      "grad_norm": 0.7265361698204105,
+      "learning_rate": 9.45288080145821e-06,
+      "loss": 0.3920948028564453,
+      "num_input_tokens_seen": 35705960,
+      "step": 350,
+      "train_runtime": 4873.848,
+      "train_tokens_per_second": 7326.031
+    },
+    {
+      "epoch": 0.305770887166236,
+      "grad_norm": 0.6865303232424075,
+      "learning_rate": 9.437394624499957e-06,
+      "loss": 0.37868556976318357,
+      "num_input_tokens_seen": 36206960,
+      "step": 355,
+      "train_runtime": 4934.8248,
+      "train_tokens_per_second": 7337.03
+    },
+    {
+      "epoch": 0.31007751937984496,
+      "grad_norm": 0.8147162296164123,
+      "learning_rate": 9.421705379434887e-06,
+      "loss": 0.3680633544921875,
+      "num_input_tokens_seen": 36749336,
+      "step": 360,
+      "train_runtime": 5009.1119,
+      "train_tokens_per_second": 7336.497
+    },
+    {
+      "epoch": 0.3143841515934539,
+      "grad_norm": 0.6626000206050349,
+      "learning_rate": 9.405813784248562e-06,
+      "loss": 0.386813759803772,
+      "num_input_tokens_seen": 37255480,
+      "step": 365,
+      "train_runtime": 5069.0004,
+      "train_tokens_per_second": 7349.67
+    },
+    {
+      "epoch": 0.3186907838070629,
+      "grad_norm": 0.7212410819832733,
+      "learning_rate": 9.38972056618668e-06,
+      "loss": 0.38484036922454834,
+      "num_input_tokens_seen": 37760024,
+      "step": 370,
+      "train_runtime": 5130.9861,
+      "train_tokens_per_second": 7359.214
+    },
+    {
+      "epoch": 0.32299741602067183,
+      "grad_norm": 0.6530771248339148,
+      "learning_rate": 9.37342646172179e-06,
+      "loss": 0.3900888919830322,
+      "num_input_tokens_seen": 38314528,
+      "step": 375,
+      "train_runtime": 5198.0649,
+      "train_tokens_per_second": 7370.921
+    },
+    {
+      "epoch": 0.3273040482342808,
+      "grad_norm": 0.6310362265785353,
+      "learning_rate": 9.356932216519588e-06,
+      "loss": 0.3763136625289917,
+      "num_input_tokens_seen": 38827464,
+      "step": 380,
+      "train_runtime": 5260.6931,
+      "train_tokens_per_second": 7380.675
+    },
+    {
+      "epoch": 0.33161068044788977,
+      "grad_norm": 0.6821166049126685,
+      "learning_rate": 9.340238585404787e-06,
+      "loss": 0.38970165252685546,
+      "num_input_tokens_seen": 39325920,
+      "step": 385,
+      "train_runtime": 5334.5963,
+      "train_tokens_per_second": 7371.864
+    },
+    {
+      "epoch": 0.3359173126614987,
+      "grad_norm": 0.7414418790229163,
+      "learning_rate": 9.32334633232659e-06,
+      "loss": 0.3639005184173584,
+      "num_input_tokens_seen": 39832488,
+      "step": 390,
+      "train_runtime": 5404.3058,
+      "train_tokens_per_second": 7370.51
+    },
+    {
+      "epoch": 0.34022394487510765,
+      "grad_norm": 0.6281453488939435,
+      "learning_rate": 9.306256230323714e-06,
+      "loss": 0.3740285873413086,
+      "num_input_tokens_seen": 40357880,
+      "step": 395,
+      "train_runtime": 5475.2685,
+      "train_tokens_per_second": 7370.941
+    },
+    {
+      "epoch": 0.34453057708871665,
+      "grad_norm": 0.6881409462313944,
+      "learning_rate": 9.28896906148902e-06,
+      "loss": 0.39269113540649414,
+      "num_input_tokens_seen": 40888088,
+      "step": 400,
+      "train_runtime": 5552.5547,
+      "train_tokens_per_second": 7363.833
+    },
+    {
+      "epoch": 0.3488372093023256,
+      "grad_norm": 0.7050381848889274,
+      "learning_rate": 9.271485616933725e-06,
+      "loss": 0.38478426933288573,
+      "num_input_tokens_seen": 41396320,
+      "step": 405,
+      "train_runtime": 5622.9597,
+      "train_tokens_per_second": 7362.016
+    },
+    {
+      "epoch": 0.35314384151593453,
+      "grad_norm": 0.6582656508034226,
+      "learning_rate": 9.253806696751185e-06,
+      "loss": 0.38239054679870604,
+      "num_input_tokens_seen": 41883152,
+      "step": 410,
+      "train_runtime": 5689.082,
+      "train_tokens_per_second": 7362.023
+    },
+    {
+      "epoch": 0.35745047372954347,
+      "grad_norm": 0.6625271537573424,
+      "learning_rate": 9.235933109980302e-06,
+      "loss": 0.39941422939300536,
+      "num_input_tokens_seen": 42386288,
+      "step": 415,
+      "train_runtime": 5763.4311,
+      "train_tokens_per_second": 7354.35
+    },
+    {
+      "epoch": 0.36175710594315247,
+      "grad_norm": 0.6020944056659068,
+      "learning_rate": 9.217865674568482e-06,
+      "loss": 0.3732919216156006,
+      "num_input_tokens_seen": 42921232,
+      "step": 420,
+      "train_runtime": 5830.9366,
+      "train_tokens_per_second": 7360.95
+    },
+    {
+      "epoch": 0.3660637381567614,
+      "grad_norm": 0.6612749680911271,
+      "learning_rate": 9.19960521733421e-06,
+      "loss": 0.36808338165283205,
+      "num_input_tokens_seen": 43431736,
+      "step": 425,
+      "train_runtime": 5899.5507,
+      "train_tokens_per_second": 7361.872
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.7052929015034174,
+      "learning_rate": 9.181152573929215e-06,
+      "loss": 0.3811118841171265,
+      "num_input_tokens_seen": 43944248,
+      "step": 430,
+      "train_runtime": 5967.9506,
+      "train_tokens_per_second": 7363.373
+    },
+    {
+      "epoch": 0.37467700258397935,
+      "grad_norm": 0.6443665940829485,
+      "learning_rate": 9.162508588800221e-06,
+      "loss": 0.40485553741455077,
+      "num_input_tokens_seen": 44470472,
+      "step": 435,
+      "train_runtime": 6044.9553,
+      "train_tokens_per_second": 7356.625
+    },
+    {
+      "epoch": 0.3789836347975883,
+      "grad_norm": 0.6692181607943731,
+      "learning_rate": 9.143674115150306e-06,
+      "loss": 0.3950349807739258,
+      "num_input_tokens_seen": 44997504,
+      "step": 440,
+      "train_runtime": 6115.4287,
+      "train_tokens_per_second": 7358.029
+    },
+    {
+      "epoch": 0.3832902670111972,
+      "grad_norm": 0.6656325377440188,
+      "learning_rate": 9.124650014899868e-06,
+      "loss": 0.3851844310760498,
+      "num_input_tokens_seen": 45490008,
+      "step": 445,
+      "train_runtime": 6183.897,
+      "train_tokens_per_second": 7356.204
+    },
+    {
+      "epoch": 0.3875968992248062,
+      "grad_norm": 0.6553358070624373,
+      "learning_rate": 9.105437158647154e-06,
+      "loss": 0.40310277938842776,
+      "num_input_tokens_seen": 46027896,
+      "step": 450,
+      "train_runtime": 6243.5546,
+      "train_tokens_per_second": 7372.066
+    },
+    {
+      "epoch": 0.39190353143841516,
+      "grad_norm": 0.6866680311340208,
+      "learning_rate": 9.086036425628453e-06,
+      "loss": 0.39217004776000974,
+      "num_input_tokens_seen": 46543128,
+      "step": 455,
+      "train_runtime": 6313.3771,
+      "train_tokens_per_second": 7372.144
+    },
+    {
+      "epoch": 0.3962101636520241,
+      "grad_norm": 0.642839268974144,
+      "learning_rate": 9.066448703677828e-06,
+      "loss": 0.3778968811035156,
+      "num_input_tokens_seen": 47048496,
+      "step": 460,
+      "train_runtime": 6382.8396,
+      "train_tokens_per_second": 7371.092
+    },
+    {
+      "epoch": 0.4005167958656331,
+      "grad_norm": 0.5917876909438503,
+      "learning_rate": 9.046674889186509e-06,
+      "loss": 0.3833005905151367,
+      "num_input_tokens_seen": 47582216,
+      "step": 465,
+      "train_runtime": 6459.5385,
+      "train_tokens_per_second": 7366.194
+    },
+    {
+      "epoch": 0.40482342807924204,
+      "grad_norm": 0.6249300807978534,
+      "learning_rate": 9.026715887061858e-06,
+      "loss": 0.38653390407562255,
+      "num_input_tokens_seen": 48103904,
+      "step": 470,
+      "train_runtime": 6531.4172,
+      "train_tokens_per_second": 7365.003
+    },
+    {
+      "epoch": 0.409130060292851,
+      "grad_norm": 0.6723447300315076,
+      "learning_rate": 9.006572610685969e-06,
+      "loss": 0.3771857738494873,
+      "num_input_tokens_seen": 48589720,
+      "step": 475,
+      "train_runtime": 6599.0101,
+      "train_tokens_per_second": 7363.183
+    },
+    {
+      "epoch": 0.4134366925064599,
+      "grad_norm": 0.6774904420934339,
+      "learning_rate": 8.986245981873854e-06,
+      "loss": 0.39016103744506836,
+      "num_input_tokens_seen": 49120400,
+      "step": 480,
+      "train_runtime": 6669.6779,
+      "train_tokens_per_second": 7364.733
+    },
+    {
+      "epoch": 0.4177433247200689,
+      "grad_norm": 0.6443265118405658,
+      "learning_rate": 8.965736930831272e-06,
+      "loss": 0.398733115196228,
+      "num_input_tokens_seen": 49651912,
+      "step": 485,
+      "train_runtime": 6740.852,
+      "train_tokens_per_second": 7365.821
+    },
+    {
+      "epoch": 0.42204995693367786,
+      "grad_norm": 0.6266168108513596,
+      "learning_rate": 8.945046396112158e-06,
+      "loss": 0.3702796697616577,
+      "num_input_tokens_seen": 50164680,
+      "step": 490,
+      "train_runtime": 6819.9627,
+      "train_tokens_per_second": 7355.565
+    },
+    {
+      "epoch": 0.4263565891472868,
+      "grad_norm": 0.6275664299170484,
+      "learning_rate": 8.92417532457566e-06,
+      "loss": 0.36701819896697996,
+      "num_input_tokens_seen": 50674248,
+      "step": 495,
+      "train_runtime": 6883.5614,
+      "train_tokens_per_second": 7361.632
+    },
+    {
+      "epoch": 0.4306632213608958,
+      "grad_norm": 0.7306000236240069,
+      "learning_rate": 8.903124671342822e-06,
+      "loss": 0.3911170959472656,
+      "num_input_tokens_seen": 51200088,
+      "step": 500,
+      "train_runtime": 6955.2058,
+      "train_tokens_per_second": 7361.405
+    },
+    {
+      "epoch": 0.43496985357450474,
+      "grad_norm": 0.7330422033270405,
+      "learning_rate": 8.881895399752873e-06,
+      "loss": 0.39371590614318847,
+      "num_input_tokens_seen": 51742280,
+      "step": 505,
+      "train_runtime": 7024.383,
+      "train_tokens_per_second": 7366.096
+    },
+    {
+      "epoch": 0.4392764857881137,
+      "grad_norm": 0.7263151417995986,
+      "learning_rate": 8.86048848131913e-06,
+      "loss": 0.3914726972579956,
+      "num_input_tokens_seen": 52294208,
+      "step": 510,
+      "train_runtime": 7092.1416,
+      "train_tokens_per_second": 7373.543
+    },
+    {
+      "epoch": 0.4435831180017227,
+      "grad_norm": 0.7435120437949877,
+      "learning_rate": 8.838904895684555e-06,
+      "loss": 0.39632158279418944,
+      "num_input_tokens_seen": 52831024,
+      "step": 515,
+      "train_runtime": 7167.5571,
+      "train_tokens_per_second": 7370.855
+    },
+    {
+      "epoch": 0.4478897502153316,
+      "grad_norm": 0.7138089969072026,
+      "learning_rate": 8.81714563057691e-06,
+      "loss": 0.38961071968078614,
+      "num_input_tokens_seen": 53372888,
+      "step": 520,
+      "train_runtime": 7238.2171,
+      "train_tokens_per_second": 7373.762
+    },
+    {
+      "epoch": 0.45219638242894056,
+      "grad_norm": 0.705868522029541,
+      "learning_rate": 8.795211681763565e-06,
+      "loss": 0.37157084941864016,
+      "num_input_tokens_seen": 53869408,
+      "step": 525,
+      "train_runtime": 7309.2034,
+      "train_tokens_per_second": 7370.079
+    },
+    {
+      "epoch": 0.45650301464254955,
+      "grad_norm": 0.6507976015541289,
+      "learning_rate": 8.773104053005928e-06,
+      "loss": 0.37537417411804197,
+      "num_input_tokens_seen": 54368528,
+      "step": 530,
+      "train_runtime": 7380.1139,
+      "train_tokens_per_second": 7366.896
+    },
+    {
+      "epoch": 0.4608096468561585,
+      "grad_norm": 0.6423561579091476,
+      "learning_rate": 8.750823756013498e-06,
+      "loss": 0.3898594379425049,
+      "num_input_tokens_seen": 54840552,
+      "step": 535,
+      "train_runtime": 7445.4319,
+      "train_tokens_per_second": 7365.664
+    },
+    {
+      "epoch": 0.46511627906976744,
+      "grad_norm": 0.6245224202252966,
+      "learning_rate": 8.728371810397586e-06,
+      "loss": 0.38840298652648925,
+      "num_input_tokens_seen": 55337456,
+      "step": 540,
+      "train_runtime": 7511.6912,
+      "train_tokens_per_second": 7366.844
+    },
+    {
+      "epoch": 0.4694229112833764,
+      "grad_norm": 0.6635902775699029,
+      "learning_rate": 8.705749243624635e-06,
+      "loss": 0.3719607353210449,
+      "num_input_tokens_seen": 55848016,
+      "step": 545,
+      "train_runtime": 7574.5817,
+      "train_tokens_per_second": 7373.083
+    },
+    {
+      "epoch": 0.4737295434969854,
+      "grad_norm": 0.6636142138451493,
+      "learning_rate": 8.68295709096922e-06,
+      "loss": 0.3457005262374878,
+      "num_input_tokens_seen": 56307448,
+      "step": 550,
+      "train_runtime": 7643.8778,
+      "train_tokens_per_second": 7366.346
+    },
+    {
+      "epoch": 0.4780361757105943,
+      "grad_norm": 0.6513434980018162,
+      "learning_rate": 8.659996395466648e-06,
+      "loss": 0.36336798667907716,
+      "num_input_tokens_seen": 56791296,
+      "step": 555,
+      "train_runtime": 7708.7709,
+      "train_tokens_per_second": 7367.101
+    },
+    {
+      "epoch": 0.48234280792420325,
+      "grad_norm": 0.655388134773357,
+      "learning_rate": 8.636868207865244e-06,
+      "loss": 0.3849674463272095,
+      "num_input_tokens_seen": 57286592,
+      "step": 560,
+      "train_runtime": 7776.3092,
+      "train_tokens_per_second": 7366.81
+    },
+    {
+      "epoch": 0.48664944013781225,
+      "grad_norm": 0.6158794330991001,
+      "learning_rate": 8.613573586578262e-06,
+      "loss": 0.3585472583770752,
+      "num_input_tokens_seen": 57779240,
+      "step": 565,
+      "train_runtime": 7846.3275,
+      "train_tokens_per_second": 7363.858
+    },
+    {
+      "epoch": 0.4909560723514212,
+      "grad_norm": 0.6282187190034991,
+      "learning_rate": 8.590113597635443e-06,
+      "loss": 0.3671374797821045,
+      "num_input_tokens_seen": 58296520,
+      "step": 570,
+      "train_runtime": 7902.9671,
+      "train_tokens_per_second": 7376.536
+    },
+    {
+      "epoch": 0.49526270456503013,
+      "grad_norm": 0.6820366351035635,
+      "learning_rate": 8.56648931463423e-06,
+      "loss": 0.3729721546173096,
+      "num_input_tokens_seen": 58787768,
+      "step": 575,
+      "train_runtime": 7976.9359,
+      "train_tokens_per_second": 7369.718
+    },
+    {
+      "epoch": 0.49956933677863913,
+      "grad_norm": 0.5953948188219655,
+      "learning_rate": 8.54270181869065e-06,
+      "loss": 0.37489094734191897,
+      "num_input_tokens_seen": 59253688,
+      "step": 580,
+      "train_runtime": 8050.0837,
+      "train_tokens_per_second": 7360.63
+    },
+    {
+      "epoch": 0.5038759689922481,
+      "grad_norm": 0.6536807624369786,
+      "learning_rate": 8.518752198389823e-06,
+      "loss": 0.3806663990020752,
+      "num_input_tokens_seen": 59777648,
+      "step": 585,
+      "train_runtime": 8117.638,
+      "train_tokens_per_second": 7363.921
+    },
+    {
+      "epoch": 0.508182601205857,
+      "grad_norm": 0.7077876190780552,
+      "learning_rate": 8.494641549736152e-06,
+      "loss": 0.35221052169799805,
+      "num_input_tokens_seen": 60267056,
+      "step": 590,
+      "train_runtime": 8186.7766,
+      "train_tokens_per_second": 7361.512
+    },
+    {
+      "epoch": 0.512489233419466,
+      "grad_norm": 0.6112186065184119,
+      "learning_rate": 8.470370976103171e-06,
+      "loss": 0.37081544399261473,
+      "num_input_tokens_seen": 60752944,
+      "step": 595,
+      "train_runtime": 8248.8743,
+      "train_tokens_per_second": 7364.998
+    },
+    {
+      "epoch": 0.5167958656330749,
+      "grad_norm": 0.6544431163583635,
+      "learning_rate": 8.445941588183042e-06,
+      "loss": 0.3763900279998779,
+      "num_input_tokens_seen": 61249152,
+      "step": 600,
+      "train_runtime": 8320.3633,
+      "train_tokens_per_second": 7361.356
+    },
+    {
+      "epoch": 0.5211024978466839,
+      "grad_norm": 0.6502245015430211,
+      "learning_rate": 8.421354503935733e-06,
+      "loss": 0.3733165740966797,
+      "num_input_tokens_seen": 61738896,
+      "step": 605,
+      "train_runtime": 8389.2584,
+      "train_tokens_per_second": 7359.279
+    },
+    {
+      "epoch": 0.5254091300602929,
+      "grad_norm": 0.5791570703545336,
+      "learning_rate": 8.396610848537858e-06,
+      "loss": 0.36464879512786863,
+      "num_input_tokens_seen": 62285720,
+      "step": 610,
+      "train_runtime": 8468.6352,
+      "train_tokens_per_second": 7354.871
+    },
+    {
+      "epoch": 0.5297157622739018,
+      "grad_norm": 0.6648330559819625,
+      "learning_rate": 8.371711754331181e-06,
+      "loss": 0.3548685073852539,
+      "num_input_tokens_seen": 62773392,
+      "step": 615,
+      "train_runtime": 8538.4686,
+      "train_tokens_per_second": 7351.833
+    },
+    {
+      "epoch": 0.5340223944875108,
+      "grad_norm": 0.7039435693401701,
+      "learning_rate": 8.346658360770798e-06,
+      "loss": 0.37767829895019533,
+      "num_input_tokens_seen": 63297240,
+      "step": 620,
+      "train_runtime": 8605.0669,
+      "train_tokens_per_second": 7355.81
+    },
+    {
+      "epoch": 0.5383290267011197,
+      "grad_norm": 0.7143763877497393,
+      "learning_rate": 8.321451814372998e-06,
+      "loss": 0.37348289489746095,
+      "num_input_tokens_seen": 63821872,
+      "step": 625,
+      "train_runtime": 8674.53,
+      "train_tokens_per_second": 7357.387
+    },
+    {
+      "epoch": 0.5426356589147286,
+      "grad_norm": 0.7179284363192735,
+      "learning_rate": 8.296093268662779e-06,
+      "loss": 0.3854733228683472,
+      "num_input_tokens_seen": 64361320,
+      "step": 630,
+      "train_runtime": 8743.8704,
+      "train_tokens_per_second": 7360.736
+    },
+    {
+      "epoch": 0.5469422911283376,
+      "grad_norm": 0.9509086516647082,
+      "learning_rate": 8.270583884121083e-06,
+      "loss": 0.3740884780883789,
+      "num_input_tokens_seen": 64877880,
+      "step": 635,
+      "train_runtime": 8818.0876,
+      "train_tokens_per_second": 7357.364
+    },
+    {
+      "epoch": 0.5512489233419466,
+      "grad_norm": 0.6546022698428935,
+      "learning_rate": 8.244924828131668e-06,
+      "loss": 0.3512680768966675,
+      "num_input_tokens_seen": 65380760,
+      "step": 640,
+      "train_runtime": 8884.2528,
+      "train_tokens_per_second": 7359.174
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.6415517399290378,
+      "learning_rate": 8.219117274927696e-06,
+      "loss": 0.36990017890930177,
+      "num_input_tokens_seen": 65889880,
+      "step": 645,
+      "train_runtime": 8944.8369,
+      "train_tokens_per_second": 7366.247
+    },
+    {
+      "epoch": 0.5598621877691645,
+      "grad_norm": 0.6441817326099615,
+      "learning_rate": 8.193162405537997e-06,
+      "loss": 0.3620732307434082,
+      "num_input_tokens_seen": 66397256,
+      "step": 650,
+      "train_runtime": 9013.6144,
+      "train_tokens_per_second": 7366.33
+    },
+    {
+      "epoch": 0.5641688199827735,
+      "grad_norm": 0.6686614531494821,
+      "learning_rate": 8.167061407733018e-06,
+      "loss": 0.37317404747009275,
+      "num_input_tokens_seen": 66907624,
+      "step": 655,
+      "train_runtime": 9082.5463,
+      "train_tokens_per_second": 7366.615
+    },
+    {
+      "epoch": 0.5684754521963824,
+      "grad_norm": 0.6944822176730631,
+      "learning_rate": 8.14081547597046e-06,
+      "loss": 0.36172118186950686,
+      "num_input_tokens_seen": 67392080,
+      "step": 660,
+      "train_runtime": 9153.9802,
+      "train_tokens_per_second": 7362.052
+    },
+    {
+      "epoch": 0.5727820844099913,
+      "grad_norm": 0.6266997505339615,
+      "learning_rate": 8.114425811340635e-06,
+      "loss": 0.36071798801422117,
+      "num_input_tokens_seen": 67880624,
+      "step": 665,
+      "train_runtime": 9227.7772,
+      "train_tokens_per_second": 7356.119
+    },
+    {
+      "epoch": 0.5770887166236004,
+      "grad_norm": 0.7282530987540681,
+      "learning_rate": 8.087893621511487e-06,
+      "loss": 0.39921371936798095,
+      "num_input_tokens_seen": 68390984,
+      "step": 670,
+      "train_runtime": 9296.9161,
+      "train_tokens_per_second": 7356.309
+    },
+    {
+      "epoch": 0.5813953488372093,
+      "grad_norm": 0.6426511038776365,
+      "learning_rate": 8.061220120673323e-06,
+      "loss": 0.38092870712280275,
+      "num_input_tokens_seen": 68896344,
+      "step": 675,
+      "train_runtime": 9363.7793,
+      "train_tokens_per_second": 7357.75
+    },
+    {
+      "epoch": 0.5857019810508183,
+      "grad_norm": 0.6872672773444454,
+      "learning_rate": 8.03440652948326e-06,
+      "loss": 0.35166177749633787,
+      "num_input_tokens_seen": 69390728,
+      "step": 680,
+      "train_runtime": 9436.3566,
+      "train_tokens_per_second": 7353.551
+    },
+    {
+      "epoch": 0.5900086132644272,
+      "grad_norm": 0.6663950335905633,
+      "learning_rate": 8.007454075009352e-06,
+      "loss": 0.36625821590423585,
+      "num_input_tokens_seen": 69913800,
+      "step": 685,
+      "train_runtime": 9512.1057,
+      "train_tokens_per_second": 7349.981
+    },
+    {
+      "epoch": 0.5943152454780362,
+      "grad_norm": 0.6365208018382441,
+      "learning_rate": 7.980363990674449e-06,
+      "loss": 0.3945136070251465,
+      "num_input_tokens_seen": 70444064,
+      "step": 690,
+      "train_runtime": 9585.1462,
+      "train_tokens_per_second": 7349.295
+    },
+    {
+      "epoch": 0.5986218776916451,
+      "grad_norm": 0.6592472029548292,
+      "learning_rate": 7.953137516199737e-06,
+      "loss": 0.36552479267120364,
+      "num_input_tokens_seen": 70941680,
+      "step": 695,
+      "train_runtime": 9653.5644,
+      "train_tokens_per_second": 7348.755
+    },
+    {
+      "epoch": 0.602928509905254,
+      "grad_norm": 0.6421651719984249,
+      "learning_rate": 7.925775897548013e-06,
+      "loss": 0.3570195198059082,
+      "num_input_tokens_seen": 71441992,
+      "step": 700,
+      "train_runtime": 9720.2583,
+      "train_tokens_per_second": 7349.804
+    },
+    {
+      "epoch": 0.6072351421188631,
+      "grad_norm": 0.6384747569278264,
+      "learning_rate": 7.898280386866673e-06,
+      "loss": 0.3764955043792725,
+      "num_input_tokens_seen": 71945280,
+      "step": 705,
+      "train_runtime": 9788.1349,
+      "train_tokens_per_second": 7350.254
+    },
+    {
+      "epoch": 0.611541774332472,
+      "grad_norm": 0.7114005381070436,
+      "learning_rate": 7.87065224243039e-06,
+      "loss": 0.38172245025634766,
+      "num_input_tokens_seen": 72420368,
+      "step": 710,
+      "train_runtime": 9855.659,
+      "train_tokens_per_second": 7348.1
+    },
+    {
+      "epoch": 0.615848406546081,
+      "grad_norm": 0.6215545093138584,
+      "learning_rate": 7.842892728583557e-06,
+      "loss": 0.3583367824554443,
+      "num_input_tokens_seen": 72937952,
+      "step": 715,
+      "train_runtime": 9916.7096,
+      "train_tokens_per_second": 7355.056
+    },
+    {
+      "epoch": 0.6201550387596899,
+      "grad_norm": 0.6932264510907867,
+      "learning_rate": 7.815003115682405e-06,
+      "loss": 0.3677781581878662,
+      "num_input_tokens_seen": 73453368,
+      "step": 720,
+      "train_runtime": 9987.9385,
+      "train_tokens_per_second": 7354.207
+    },
+    {
+      "epoch": 0.6244616709732989,
+      "grad_norm": 0.6755500942949265,
+      "learning_rate": 7.78698468003688e-06,
+      "loss": 0.36302263736724855,
+      "num_input_tokens_seen": 73959264,
+      "step": 725,
+      "train_runtime": 10054.9379,
+      "train_tokens_per_second": 7355.517
+    },
+    {
+      "epoch": 0.6287683031869078,
+      "grad_norm": 0.6952999463715118,
+      "learning_rate": 7.75883870385223e-06,
+      "loss": 0.37976875305175783,
+      "num_input_tokens_seen": 74444936,
+      "step": 730,
+      "train_runtime": 10131.0665,
+      "train_tokens_per_second": 7348.184
+    },
+    {
+      "epoch": 0.6330749354005168,
+      "grad_norm": 0.735886603640982,
+      "learning_rate": 7.730566475170334e-06,
+      "loss": 0.3730980157852173,
+      "num_input_tokens_seen": 74962952,
+      "step": 735,
+      "train_runtime": 10196.8779,
+      "train_tokens_per_second": 7351.559
+    },
+    {
+      "epoch": 0.6373815676141258,
+      "grad_norm": 0.6824970423272824,
+      "learning_rate": 7.702169287810751e-06,
+      "loss": 0.3567670345306396,
+      "num_input_tokens_seen": 75451216,
+      "step": 740,
+      "train_runtime": 10261.135,
+      "train_tokens_per_second": 7353.106
+    },
+    {
+      "epoch": 0.6416881998277347,
+      "grad_norm": 0.567176098283514,
+      "learning_rate": 7.67364844131151e-06,
+      "loss": 0.3866007328033447,
+      "num_input_tokens_seen": 76008248,
+      "step": 745,
+      "train_runtime": 10322.9684,
+      "train_tokens_per_second": 7363.022
+    },
+    {
+      "epoch": 0.6459948320413437,
+      "grad_norm": 0.6031679181986589,
+      "learning_rate": 7.645005240869644e-06,
+      "loss": 0.3532373905181885,
+      "num_input_tokens_seen": 76529328,
+      "step": 750,
+      "train_runtime": 10389.7584,
+      "train_tokens_per_second": 7365.843
+    },
+    {
+      "epoch": 0.6503014642549526,
+      "grad_norm": 0.7159571723969788,
+      "learning_rate": 7.616240997281465e-06,
+      "loss": 0.3541311502456665,
+      "num_input_tokens_seen": 77047776,
+      "step": 755,
+      "train_runtime": 10466.7096,
+      "train_tokens_per_second": 7361.222
+    },
+    {
+      "epoch": 0.6546080964685616,
+      "grad_norm": 0.634378436444587,
+      "learning_rate": 7.587357026882563e-06,
+      "loss": 0.36530122756958006,
+      "num_input_tokens_seen": 77585376,
+      "step": 760,
+      "train_runtime": 10541.6333,
+      "train_tokens_per_second": 7359.901
+    },
+    {
+      "epoch": 0.6589147286821705,
+      "grad_norm": 0.6298702094677439,
+      "learning_rate": 7.558354651487583e-06,
+      "loss": 0.3949885845184326,
+      "num_input_tokens_seen": 78094440,
+      "step": 765,
+      "train_runtime": 10611.005,
+      "train_tokens_per_second": 7359.759
+    },
+    {
+      "epoch": 0.6632213608957795,
+      "grad_norm": 0.6249157337085598,
+      "learning_rate": 7.529235198329723e-06,
+      "loss": 0.3596529006958008,
+      "num_input_tokens_seen": 78635856,
+      "step": 770,
+      "train_runtime": 10691.6063,
+      "train_tokens_per_second": 7354.915
+    },
+    {
+      "epoch": 0.6675279931093885,
+      "grad_norm": 0.6023145192971817,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.35566143989562987,
+      "num_input_tokens_seen": 79138600,
+      "step": 775,
+      "train_runtime": 10756.7977,
+      "train_tokens_per_second": 7357.078
+    },
+    {
+      "epoch": 0.6718346253229974,
+      "grad_norm": 0.6593097675747426,
+      "learning_rate": 7.470650394386274e-06,
+      "loss": 0.36934409141540525,
+      "num_input_tokens_seen": 79642424,
+      "step": 780,
+      "train_runtime": 10825.3842,
+      "train_tokens_per_second": 7357.007
+    },
+    {
+      "epoch": 0.6761412575366064,
+      "grad_norm": 0.7137202399672637,
+      "learning_rate": 7.441187724612007e-06,
+      "loss": 0.37495574951171873,
+      "num_input_tokens_seen": 80152744,
+      "step": 785,
+      "train_runtime": 10890.6152,
+      "train_tokens_per_second": 7359.8
+    },
+    {
+      "epoch": 0.6804478897502153,
+      "grad_norm": 0.605977280910932,
+      "learning_rate": 7.4116133389748115e-06,
+      "loss": 0.37909436225891113,
+      "num_input_tokens_seen": 80693400,
+      "step": 790,
+      "train_runtime": 10944.0447,
+      "train_tokens_per_second": 7373.27
+    },
+    {
+      "epoch": 0.6847545219638242,
+      "grad_norm": 0.7982064055258384,
+      "learning_rate": 7.381928590884741e-06,
+      "loss": 0.361489462852478,
+      "num_input_tokens_seen": 81180032,
+      "step": 795,
+      "train_runtime": 11009.7965,
+      "train_tokens_per_second": 7373.436
+    },
+    {
+      "epoch": 0.6890611541774333,
+      "grad_norm": 0.560346275445483,
+      "learning_rate": 7.35213483880236e-06,
+      "loss": 0.34208495616912843,
+      "num_input_tokens_seen": 81715960,
+      "step": 800,
+      "train_runtime": 11082.868,
+      "train_tokens_per_second": 7373.178
+    },
+    {
+      "epoch": 0.6933677863910422,
+      "grad_norm": 0.7576401916556533,
+      "learning_rate": 7.322233446176571e-06,
+      "loss": 0.3672114133834839,
+      "num_input_tokens_seen": 82195808,
+      "step": 805,
+      "train_runtime": 11153.6467,
+      "train_tokens_per_second": 7369.411
+    },
+    {
+      "epoch": 0.6976744186046512,
+      "grad_norm": 0.6525903599941298,
+      "learning_rate": 7.29222578138222e-06,
+      "loss": 0.3744704008102417,
+      "num_input_tokens_seen": 82680488,
+      "step": 810,
+      "train_runtime": 11221.591,
+      "train_tokens_per_second": 7367.983
+    },
+    {
+      "epoch": 0.7019810508182601,
+      "grad_norm": 0.5908244133789757,
+      "learning_rate": 7.2621132176574774e-06,
+      "loss": 0.373249888420105,
+      "num_input_tokens_seen": 83207016,
+      "step": 815,
+      "train_runtime": 11292.0868,
+      "train_tokens_per_second": 7368.613
+    },
+    {
+      "epoch": 0.7062876830318691,
+      "grad_norm": 0.8900954684782045,
+      "learning_rate": 7.231897133040997e-06,
+      "loss": 0.34656834602355957,
+      "num_input_tokens_seen": 83705464,
+      "step": 820,
+      "train_runtime": 11366.0296,
+      "train_tokens_per_second": 7364.53
+    },
+    {
+      "epoch": 0.710594315245478,
+      "grad_norm": 0.6599764009158128,
+      "learning_rate": 7.201578910308848e-06,
+      "loss": 0.36565725803375243,
+      "num_input_tokens_seen": 84223624,
+      "step": 825,
+      "train_runtime": 11439.0205,
+      "train_tokens_per_second": 7362.835
+    },
+    {
+      "epoch": 0.7149009474590869,
+      "grad_norm": 0.7671400370894249,
+      "learning_rate": 7.1711599369112385e-06,
+      "loss": 0.3640167474746704,
+      "num_input_tokens_seen": 84733008,
+      "step": 830,
+      "train_runtime": 11498.286,
+      "train_tokens_per_second": 7369.186
+    },
+    {
+      "epoch": 0.719207579672696,
+      "grad_norm": 0.6890917187634811,
+      "learning_rate": 7.14064160490902e-06,
+      "loss": 0.34748039245605467,
+      "num_input_tokens_seen": 85244448,
+      "step": 835,
+      "train_runtime": 11570.0337,
+      "train_tokens_per_second": 7367.692
+    },
+    {
+      "epoch": 0.7235142118863049,
+      "grad_norm": 0.5766158583478934,
+      "learning_rate": 7.110025310909981e-06,
+      "loss": 0.353003716468811,
+      "num_input_tokens_seen": 85762344,
+      "step": 840,
+      "train_runtime": 11638.7238,
+      "train_tokens_per_second": 7368.707
+    },
+    {
+      "epoch": 0.7278208440999139,
+      "grad_norm": 0.5823984856746932,
+      "learning_rate": 7.079312456004941e-06,
+      "loss": 0.37342329025268556,
+      "num_input_tokens_seen": 86289224,
+      "step": 845,
+      "train_runtime": 11716.1517,
+      "train_tokens_per_second": 7364.98
+    },
+    {
+      "epoch": 0.7321274763135228,
+      "grad_norm": 0.5927480726961869,
+      "learning_rate": 7.048504445703623e-06,
+      "loss": 0.3299787759780884,
+      "num_input_tokens_seen": 86796928,
+      "step": 850,
+      "train_runtime": 11785.4668,
+      "train_tokens_per_second": 7364.742
+    },
+    {
+      "epoch": 0.7364341085271318,
+      "grad_norm": 0.6321864557945756,
+      "learning_rate": 7.017602689870345e-06,
+      "loss": 0.36838767528533933,
+      "num_input_tokens_seen": 87320496,
+      "step": 855,
+      "train_runtime": 11850.93,
+      "train_tokens_per_second": 7368.24
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.6018531234880375,
+      "learning_rate": 6.986608602659486e-06,
+      "loss": 0.3541435718536377,
+      "num_input_tokens_seen": 87879104,
+      "step": 860,
+      "train_runtime": 11917.2964,
+      "train_tokens_per_second": 7374.081
+    },
+    {
+      "epoch": 0.7450473729543498,
+      "grad_norm": 0.6193130746098626,
+      "learning_rate": 6.95552360245078e-06,
+      "loss": 0.35468919277191163,
+      "num_input_tokens_seen": 88373816,
+      "step": 865,
+      "train_runtime": 11986.0817,
+      "train_tokens_per_second": 7373.036
+    },
+    {
+      "epoch": 0.7493540051679587,
+      "grad_norm": 0.5838704103103464,
+      "learning_rate": 6.92434911178441e-06,
+      "loss": 0.36195812225341795,
+      "num_input_tokens_seen": 88916672,
+      "step": 870,
+      "train_runtime": 12056.0128,
+      "train_tokens_per_second": 7375.297
+    },
+    {
+      "epoch": 0.7536606373815676,
+      "grad_norm": 0.6583966223351327,
+      "learning_rate": 6.893086557295896e-06,
+      "loss": 0.3709383726119995,
+      "num_input_tokens_seen": 89428224,
+      "step": 875,
+      "train_runtime": 12120.5201,
+      "train_tokens_per_second": 7378.25
+    },
+    {
+      "epoch": 0.7579672695951766,
+      "grad_norm": 0.579643853477019,
+      "learning_rate": 6.861737369650818e-06,
+      "loss": 0.362162446975708,
+      "num_input_tokens_seen": 89924488,
+      "step": 880,
+      "train_runtime": 12192.2773,
+      "train_tokens_per_second": 7375.528
+    },
+    {
+      "epoch": 0.7622739018087855,
+      "grad_norm": 0.5904164496483095,
+      "learning_rate": 6.830302983479344e-06,
+      "loss": 0.36125888824462893,
+      "num_input_tokens_seen": 90470424,
+      "step": 885,
+      "train_runtime": 12260.5076,
+      "train_tokens_per_second": 7379.011
+    },
+    {
+      "epoch": 0.7665805340223945,
+      "grad_norm": 0.6357648372036675,
+      "learning_rate": 6.7987848373105705e-06,
+      "loss": 0.35643205642700193,
+      "num_input_tokens_seen": 90996680,
+      "step": 890,
+      "train_runtime": 12336.8599,
+      "train_tokens_per_second": 7376.0
+    },
+    {
+      "epoch": 0.7708871662360034,
+      "grad_norm": 0.6943191292916018,
+      "learning_rate": 6.767184373506698e-06,
+      "loss": 0.34829187393188477,
+      "num_input_tokens_seen": 91480544,
+      "step": 895,
+      "train_runtime": 12387.5001,
+      "train_tokens_per_second": 7384.908
+    },
+    {
+      "epoch": 0.7751937984496124,
+      "grad_norm": 0.7220194168057159,
+      "learning_rate": 6.73550303819702e-06,
+      "loss": 0.3527660846710205,
+      "num_input_tokens_seen": 91969264,
+      "step": 900,
+      "train_runtime": 12458.5492,
+      "train_tokens_per_second": 7382.02
+    },
+    {
+      "epoch": 0.7795004306632214,
+      "grad_norm": 0.7375983296510873,
+      "learning_rate": 6.70374228121175e-06,
+      "loss": 0.35948958396911623,
+      "num_input_tokens_seen": 92456304,
+      "step": 905,
+      "train_runtime": 12529.4987,
+      "train_tokens_per_second": 7379.09
+    },
+    {
+      "epoch": 0.7838070628768303,
+      "grad_norm": 0.6425550494971753,
+      "learning_rate": 6.671903556015664e-06,
+      "loss": 0.35416080951690676,
+      "num_input_tokens_seen": 92982984,
+      "step": 910,
+      "train_runtime": 12602.0994,
+      "train_tokens_per_second": 7378.373
+    },
+    {
+      "epoch": 0.7881136950904393,
+      "grad_norm": 0.6001020126847847,
+      "learning_rate": 6.639988319641592e-06,
+      "loss": 0.34634552001953123,
+      "num_input_tokens_seen": 93461984,
+      "step": 915,
+      "train_runtime": 12671.0709,
+      "train_tokens_per_second": 7376.013
+    },
+    {
+      "epoch": 0.7924203273040482,
+      "grad_norm": 0.6439880804882693,
+      "learning_rate": 6.607998032623741e-06,
+      "loss": 0.37337236404418944,
+      "num_input_tokens_seen": 93979864,
+      "step": 920,
+      "train_runtime": 12735.015,
+      "train_tokens_per_second": 7379.643
+    },
+    {
+      "epoch": 0.7967269595176572,
+      "grad_norm": 0.6979094022106086,
+      "learning_rate": 6.57593415893085e-06,
+      "loss": 0.3476158380508423,
+      "num_input_tokens_seen": 94462232,
+      "step": 925,
+      "train_runtime": 12801.7577,
+      "train_tokens_per_second": 7378.849
+    },
+    {
+      "epoch": 0.8010335917312662,
+      "grad_norm": 0.765921272529139,
+      "learning_rate": 6.5437981658992e-06,
+      "loss": 0.35996217727661134,
+      "num_input_tokens_seen": 94992816,
+      "step": 930,
+      "train_runtime": 12865.11,
+      "train_tokens_per_second": 7383.755
+    },
+    {
+      "epoch": 0.8053402239448751,
+      "grad_norm": 0.599395651309773,
+      "learning_rate": 6.511591524165465e-06,
+      "loss": 0.36520819664001464,
+      "num_input_tokens_seen": 95488328,
+      "step": 935,
+      "train_runtime": 12937.4041,
+      "train_tokens_per_second": 7380.795
+    },
+    {
+      "epoch": 0.8096468561584841,
+      "grad_norm": 0.6451964755773082,
+      "learning_rate": 6.479315707599407e-06,
+      "loss": 0.36473860740661623,
+      "num_input_tokens_seen": 95954480,
+      "step": 940,
+      "train_runtime": 12996.4069,
+      "train_tokens_per_second": 7383.155
+    },
+    {
+      "epoch": 0.813953488372093,
+      "grad_norm": 0.6050461893812418,
+      "learning_rate": 6.446972193236433e-06,
+      "loss": 0.3522223472595215,
+      "num_input_tokens_seen": 96490896,
+      "step": 945,
+      "train_runtime": 13056.1311,
+      "train_tokens_per_second": 7390.466
+    },
+    {
+      "epoch": 0.818260120585702,
+      "grad_norm": 0.6648024183404038,
+      "learning_rate": 6.414562461209994e-06,
+      "loss": 0.3675665855407715,
+      "num_input_tokens_seen": 97032496,
+      "step": 950,
+      "train_runtime": 13126.6476,
+      "train_tokens_per_second": 7392.024
+    },
+    {
+      "epoch": 0.8225667527993109,
+      "grad_norm": 0.6311611856518416,
+      "learning_rate": 6.3820879946838585e-06,
+      "loss": 0.3679534435272217,
+      "num_input_tokens_seen": 97543080,
+      "step": 955,
+      "train_runtime": 13201.9771,
+      "train_tokens_per_second": 7388.521
+    },
+    {
+      "epoch": 0.8268733850129198,
+      "grad_norm": 0.7116714532190532,
+      "learning_rate": 6.349550279784231e-06,
+      "loss": 0.3499560594558716,
+      "num_input_tokens_seen": 98020152,
+      "step": 960,
+      "train_runtime": 13274.0305,
+      "train_tokens_per_second": 7384.355
+    },
+    {
+      "epoch": 0.8311800172265289,
+      "grad_norm": 0.615142461235978,
+      "learning_rate": 6.316950805531746e-06,
+      "loss": 0.3565502166748047,
+      "num_input_tokens_seen": 98498816,
+      "step": 965,
+      "train_runtime": 13343.5463,
+      "train_tokens_per_second": 7381.757
+    },
+    {
+      "epoch": 0.8354866494401378,
+      "grad_norm": 0.6531050926414822,
+      "learning_rate": 6.284291063773331e-06,
+      "loss": 0.36150004863739016,
+      "num_input_tokens_seen": 99045784,
+      "step": 970,
+      "train_runtime": 13419.4835,
+      "train_tokens_per_second": 7380.745
+    },
+    {
+      "epoch": 0.8397932816537468,
+      "grad_norm": 0.6368022854985766,
+      "learning_rate": 6.251572549113925e-06,
+      "loss": 0.36032235622406006,
+      "num_input_tokens_seen": 99579432,
+      "step": 975,
+      "train_runtime": 13487.1876,
+      "train_tokens_per_second": 7383.261
+    },
+    {
+      "epoch": 0.8440999138673557,
+      "grad_norm": 0.6146788889706721,
+      "learning_rate": 6.218796758848085e-06,
+      "loss": 0.3623358726501465,
+      "num_input_tokens_seen": 100089992,
+      "step": 980,
+      "train_runtime": 13556.551,
+      "train_tokens_per_second": 7383.146
+    },
+    {
+      "epoch": 0.8484065460809647,
+      "grad_norm": 0.6116609253098809,
+      "learning_rate": 6.185965192891472e-06,
+      "loss": 0.36282594203948976,
+      "num_input_tokens_seen": 100618600,
+      "step": 985,
+      "train_runtime": 13614.4768,
+      "train_tokens_per_second": 7390.559
+    },
+    {
+      "epoch": 0.8527131782945736,
+      "grad_norm": 0.6846681338486661,
+      "learning_rate": 6.153079353712201e-06,
+      "loss": 0.3642581939697266,
+      "num_input_tokens_seen": 101111776,
+      "step": 990,
+      "train_runtime": 13677.4451,
+      "train_tokens_per_second": 7392.592
+    },
+    {
+      "epoch": 0.8570198105081827,
+      "grad_norm": 0.6305569102019681,
+      "learning_rate": 6.120140746262091e-06,
+      "loss": 0.3563115119934082,
+      "num_input_tokens_seen": 101603528,
+      "step": 995,
+      "train_runtime": 13750.1142,
+      "train_tokens_per_second": 7389.286
+    },
+    {
+      "epoch": 0.8613264427217916,
+      "grad_norm": 0.6356575993962907,
+      "learning_rate": 6.087150877907786e-06,
+      "loss": 0.3616267442703247,
+      "num_input_tokens_seen": 102147912,
+      "step": 1000,
+      "train_runtime": 13817.3271,
+      "train_tokens_per_second": 7392.74
+    },
+    {
+      "epoch": 0.8656330749354005,
+      "grad_norm": 0.7059213261617521,
+      "learning_rate": 6.054111258361782e-06,
+      "loss": 0.3602945804595947,
+      "num_input_tokens_seen": 102659472,
+      "step": 1005,
+      "train_runtime": 13880.5359,
+      "train_tokens_per_second": 7395.93
+    },
+    {
+      "epoch": 0.8699397071490095,
+      "grad_norm": 0.6411294797248545,
+      "learning_rate": 6.021023399613335e-06,
+      "loss": 0.34914188385009765,
+      "num_input_tokens_seen": 103179440,
+      "step": 1010,
+      "train_runtime": 13951.8001,
+      "train_tokens_per_second": 7395.421
+    },
+    {
+      "epoch": 0.8742463393626184,
+      "grad_norm": 0.6072534268937941,
+      "learning_rate": 5.987888815859266e-06,
+      "loss": 0.3555158138275146,
+      "num_input_tokens_seen": 103690584,
+      "step": 1015,
+      "train_runtime": 14022.5116,
+      "train_tokens_per_second": 7394.58
+    },
+    {
+      "epoch": 0.8785529715762274,
+      "grad_norm": 0.6179975691240837,
+      "learning_rate": 5.954709023434669e-06,
+      "loss": 0.33635711669921875,
+      "num_input_tokens_seen": 104178112,
+      "step": 1020,
+      "train_runtime": 14085.2268,
+      "train_tokens_per_second": 7396.268
+    },
+    {
+      "epoch": 0.8828596037898363,
+      "grad_norm": 0.6614614352560472,
+      "learning_rate": 5.921485540743516e-06,
+      "loss": 0.35177910327911377,
+      "num_input_tokens_seen": 104699576,
+      "step": 1025,
+      "train_runtime": 14160.14,
+      "train_tokens_per_second": 7393.965
+    },
+    {
+      "epoch": 0.8871662360034454,
+      "grad_norm": 0.64245424991513,
+      "learning_rate": 5.888219888189176e-06,
+      "loss": 0.36098456382751465,
+      "num_input_tokens_seen": 105214336,
+      "step": 1030,
+      "train_runtime": 14213.0669,
+      "train_tokens_per_second": 7402.648
+    },
+    {
+      "epoch": 0.8914728682170543,
+      "grad_norm": 0.6765874092986255,
+      "learning_rate": 5.854913588104832e-06,
+      "loss": 0.37163600921630857,
+      "num_input_tokens_seen": 105712752,
+      "step": 1035,
+      "train_runtime": 14273.3512,
+      "train_tokens_per_second": 7406.302
+    },
+    {
+      "epoch": 0.8957795004306632,
+      "grad_norm": 0.6635023467337341,
+      "learning_rate": 5.82156816468382e-06,
+      "loss": 0.37590489387512205,
+      "num_input_tokens_seen": 106215336,
+      "step": 1040,
+      "train_runtime": 14341.6979,
+      "train_tokens_per_second": 7406.05
+    },
+    {
+      "epoch": 0.9000861326442722,
+      "grad_norm": 0.6210568645525162,
+      "learning_rate": 5.788185143909868e-06,
+      "loss": 0.35026397705078127,
+      "num_input_tokens_seen": 106737504,
+      "step": 1045,
+      "train_runtime": 14414.7002,
+      "train_tokens_per_second": 7404.768
+    },
+    {
+      "epoch": 0.9043927648578811,
+      "grad_norm": 0.6775230384008981,
+      "learning_rate": 5.754766053487274e-06,
+      "loss": 0.3652517080307007,
+      "num_input_tokens_seen": 107261064,
+      "step": 1050,
+      "train_runtime": 14484.8725,
+      "train_tokens_per_second": 7405.04
+    },
+    {
+      "epoch": 0.9086993970714901,
+      "grad_norm": 0.6034355872415312,
+      "learning_rate": 5.721312422770984e-06,
+      "loss": 0.3495129108428955,
+      "num_input_tokens_seen": 107787160,
+      "step": 1055,
+      "train_runtime": 14564.529,
+      "train_tokens_per_second": 7400.662
+    },
+    {
+      "epoch": 0.9130060292850991,
+      "grad_norm": 0.6266482958161476,
+      "learning_rate": 5.68782578269661e-06,
+      "loss": 0.34899396896362306,
+      "num_input_tokens_seen": 108278376,
+      "step": 1060,
+      "train_runtime": 14637.9348,
+      "train_tokens_per_second": 7397.107
+    },
+    {
+      "epoch": 0.917312661498708,
+      "grad_norm": 0.5978323602685123,
+      "learning_rate": 5.6543076657103705e-06,
+      "loss": 0.35709681510925295,
+      "num_input_tokens_seen": 108741080,
+      "step": 1065,
+      "train_runtime": 14703.9074,
+      "train_tokens_per_second": 7395.387
+    },
+    {
+      "epoch": 0.921619293712317,
+      "grad_norm": 0.5791503709903075,
+      "learning_rate": 5.620759605698953e-06,
+      "loss": 0.35440216064453123,
+      "num_input_tokens_seen": 109252016,
+      "step": 1070,
+      "train_runtime": 14783.6598,
+      "train_tokens_per_second": 7390.052
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 0.7142563752017141,
+      "learning_rate": 5.587183137919332e-06,
+      "loss": 0.3724498748779297,
+      "num_input_tokens_seen": 109763232,
+      "step": 1075,
+      "train_runtime": 14858.0697,
+      "train_tokens_per_second": 7387.449
+    },
+    {
+      "epoch": 0.9302325581395349,
+      "grad_norm": 0.609481394596669,
+      "learning_rate": 5.553579798928503e-06,
+      "loss": 0.35597658157348633,
+      "num_input_tokens_seen": 110258264,
+      "step": 1080,
+      "train_runtime": 14920.315,
+      "train_tokens_per_second": 7389.808
+    },
+    {
+      "epoch": 0.9345391903531438,
+      "grad_norm": 0.6901948885966874,
+      "learning_rate": 5.519951126513164e-06,
+      "loss": 0.366562032699585,
+      "num_input_tokens_seen": 110761536,
+      "step": 1085,
+      "train_runtime": 14989.6087,
+      "train_tokens_per_second": 7389.221
+    },
+    {
+      "epoch": 0.9388458225667528,
+      "grad_norm": 0.6917267032383838,
+      "learning_rate": 5.486298659619346e-06,
+      "loss": 0.3656296730041504,
+      "num_input_tokens_seen": 111228400,
+      "step": 1090,
+      "train_runtime": 15050.3018,
+      "train_tokens_per_second": 7390.443
+    },
+    {
+      "epoch": 0.9431524547803618,
+      "grad_norm": 0.6752126310495482,
+      "learning_rate": 5.452623938281983e-06,
+      "loss": 0.3466034412384033,
+      "num_input_tokens_seen": 111733504,
+      "step": 1095,
+      "train_runtime": 15125.111,
+      "train_tokens_per_second": 7387.285
+    },
+    {
+      "epoch": 0.9474590869939707,
+      "grad_norm": 0.6713559565663015,
+      "learning_rate": 5.418928503554436e-06,
+      "loss": 0.36214759349823,
+      "num_input_tokens_seen": 112225696,
+      "step": 1100,
+      "train_runtime": 15193.5378,
+      "train_tokens_per_second": 7386.41
+    },
+    {
+      "epoch": 0.9517657192075797,
+      "grad_norm": 0.632679462363844,
+      "learning_rate": 5.385213897437975e-06,
+      "loss": 0.3605997085571289,
+      "num_input_tokens_seen": 112749656,
+      "step": 1105,
+      "train_runtime": 15252.8865,
+      "train_tokens_per_second": 7392.021
+    },
+    {
+      "epoch": 0.9560723514211886,
+      "grad_norm": 0.6951338495408681,
+      "learning_rate": 5.351481662811208e-06,
+      "loss": 0.3977372407913208,
+      "num_input_tokens_seen": 113307680,
+      "step": 1110,
+      "train_runtime": 15315.8148,
+      "train_tokens_per_second": 7398.084
+    },
+    {
+      "epoch": 0.9603789836347976,
+      "grad_norm": 0.7046520621915809,
+      "learning_rate": 5.3177333433594734e-06,
+      "loss": 0.3589299201965332,
+      "num_input_tokens_seen": 113816672,
+      "step": 1115,
+      "train_runtime": 15386.2096,
+      "train_tokens_per_second": 7397.317
+    },
+    {
+      "epoch": 0.9646856158484065,
+      "grad_norm": 0.6263869344134583,
+      "learning_rate": 5.283970483504198e-06,
+      "loss": 0.35573978424072267,
+      "num_input_tokens_seen": 114328160,
+      "step": 1120,
+      "train_runtime": 15460.4356,
+      "train_tokens_per_second": 7394.886
+    },
+    {
+      "epoch": 0.9689922480620154,
+      "grad_norm": 0.5974542867890239,
+      "learning_rate": 5.2501946283322204e-06,
+      "loss": 0.3410700798034668,
+      "num_input_tokens_seen": 114790904,
+      "step": 1125,
+      "train_runtime": 15522.3975,
+      "train_tokens_per_second": 7395.179
+    },
+    {
+      "epoch": 0.9732988802756245,
+      "grad_norm": 0.7266994361124255,
+      "learning_rate": 5.216407323525083e-06,
+      "loss": 0.3575610876083374,
+      "num_input_tokens_seen": 115286400,
+      "step": 1130,
+      "train_runtime": 15589.5728,
+      "train_tokens_per_second": 7395.097
+    },
+    {
+      "epoch": 0.9776055124892334,
+      "grad_norm": 0.6603430551607196,
+      "learning_rate": 5.182610115288296e-06,
+      "loss": 0.35471277236938475,
+      "num_input_tokens_seen": 115819512,
+      "step": 1135,
+      "train_runtime": 15664.1723,
+      "train_tokens_per_second": 7393.912
+    },
+    {
+      "epoch": 0.9819121447028424,
+      "grad_norm": 0.6122815306318921,
+      "learning_rate": 5.148804550280578e-06,
+      "loss": 0.35717573165893557,
+      "num_input_tokens_seen": 116305168,
+      "step": 1140,
+      "train_runtime": 15727.3256,
+      "train_tokens_per_second": 7395.101
+    },
+    {
+      "epoch": 0.9862187769164513,
+      "grad_norm": 0.6081458803821285,
+      "learning_rate": 5.114992175543084e-06,
+      "loss": 0.346773099899292,
+      "num_input_tokens_seen": 116753064,
+      "step": 1145,
+      "train_runtime": 15789.1485,
+      "train_tokens_per_second": 7394.513
+    },
+    {
+      "epoch": 0.9905254091300603,
+      "grad_norm": 0.6199305616103742,
+      "learning_rate": 5.081174538428596e-06,
+      "loss": 0.36169114112854006,
+      "num_input_tokens_seen": 117272904,
+      "step": 1150,
+      "train_runtime": 15859.7259,
+      "train_tokens_per_second": 7394.384
+    },
+    {
+      "epoch": 0.9948320413436692,
+      "grad_norm": 0.5967979270790722,
+      "learning_rate": 5.047353186530718e-06,
+      "loss": 0.34629039764404296,
+      "num_input_tokens_seen": 117758136,
+      "step": 1155,
+      "train_runtime": 15925.2581,
+      "train_tokens_per_second": 7394.426
+    },
+    {
+      "epoch": 0.9991386735572783,
+      "grad_norm": 0.7520097407257402,
+      "learning_rate": 5.0135296676130576e-06,
+      "loss": 0.35693745613098143,
+      "num_input_tokens_seen": 118275664,
+      "step": 1160,
+      "train_runtime": 15997.7061,
+      "train_tokens_per_second": 7393.289
+    },
+    {
+      "epoch": 1.0034453057708872,
+      "grad_norm": 0.6519790289096927,
+      "learning_rate": 4.979705529538385e-06,
+      "loss": 0.3211682796478271,
+      "num_input_tokens_seen": 118772088,
+      "step": 1165,
+      "train_runtime": 16067.268,
+      "train_tokens_per_second": 7392.177
+    },
+    {
+      "epoch": 1.0077519379844961,
+      "grad_norm": 0.7176780843602725,
+      "learning_rate": 4.9458823201978124e-06,
+      "loss": 0.3132964611053467,
+      "num_input_tokens_seen": 119324520,
+      "step": 1170,
+      "train_runtime": 16146.0511,
+      "train_tokens_per_second": 7390.322
+    },
+    {
+      "epoch": 1.012058570198105,
+      "grad_norm": 0.7294575856519293,
+      "learning_rate": 4.912061587439944e-06,
+      "loss": 0.32307000160217286,
+      "num_input_tokens_seen": 119862352,
+      "step": 1175,
+      "train_runtime": 16220.7374,
+      "train_tokens_per_second": 7389.451
+    },
+    {
+      "epoch": 1.016365202411714,
+      "grad_norm": 0.6486717204872143,
+      "learning_rate": 4.8782448790000525e-06,
+      "loss": 0.3019747257232666,
+      "num_input_tokens_seen": 120353472,
+      "step": 1180,
+      "train_runtime": 16288.6517,
+      "train_tokens_per_second": 7388.793
+    },
+    {
+      "epoch": 1.020671834625323,
+      "grad_norm": 0.7028559369295986,
+      "learning_rate": 4.8444337424292445e-06,
+      "loss": 0.3043541669845581,
+      "num_input_tokens_seen": 120865584,
+      "step": 1185,
+      "train_runtime": 16369.8645,
+      "train_tokens_per_second": 7383.42
+    },
+    {
+      "epoch": 1.024978466838932,
+      "grad_norm": 0.7560138997334946,
+      "learning_rate": 4.81062972502364e-06,
+      "loss": 0.32289729118347166,
+      "num_input_tokens_seen": 121382112,
+      "step": 1190,
+      "train_runtime": 16445.2618,
+      "train_tokens_per_second": 7380.978
+    },
+    {
+      "epoch": 1.0292850990525408,
+      "grad_norm": 0.6855270470811776,
+      "learning_rate": 4.7768343737535694e-06,
+      "loss": 0.3106250524520874,
+      "num_input_tokens_seen": 121946272,
+      "step": 1195,
+      "train_runtime": 16520.8552,
+      "train_tokens_per_second": 7381.353
+    },
+    {
+      "epoch": 1.0335917312661498,
+      "grad_norm": 0.5866792589666393,
+      "learning_rate": 4.743049235192771e-06,
+      "loss": 0.30351016521453855,
+      "num_input_tokens_seen": 122456216,
+      "step": 1200,
+      "train_runtime": 16592.4847,
+      "train_tokens_per_second": 7380.222
+    },
+    {
+      "epoch": 1.0378983634797587,
+      "grad_norm": 0.5895059778955265,
+      "learning_rate": 4.7092758554476215e-06,
+      "loss": 0.30516183376312256,
+      "num_input_tokens_seen": 122997032,
+      "step": 1205,
+      "train_runtime": 16662.8806,
+      "train_tokens_per_second": 7381.499
+    },
+    {
+      "epoch": 1.0422049956933679,
+      "grad_norm": 0.7393826499994867,
+      "learning_rate": 4.6755157800863826e-06,
+      "loss": 0.30727853775024416,
+      "num_input_tokens_seen": 123504248,
+      "step": 1210,
+      "train_runtime": 16733.5532,
+      "train_tokens_per_second": 7380.635
+    },
+    {
+      "epoch": 1.0465116279069768,
+      "grad_norm": 0.657650517709326,
+      "learning_rate": 4.641770554068465e-06,
+      "loss": 0.2971274137496948,
+      "num_input_tokens_seen": 124021952,
+      "step": 1215,
+      "train_runtime": 16798.7638,
+      "train_tokens_per_second": 7382.802
+    },
+    {
+      "epoch": 1.0508182601205858,
+      "grad_norm": 0.6783125678934093,
+      "learning_rate": 4.608041721673731e-06,
+      "loss": 0.31333255767822266,
+      "num_input_tokens_seen": 124538536,
+      "step": 1220,
+      "train_runtime": 16866.0122,
+      "train_tokens_per_second": 7383.994
+    },
+    {
+      "epoch": 1.0551248923341947,
+      "grad_norm": 0.68062001198956,
+      "learning_rate": 4.574330826431822e-06,
+      "loss": 0.33772854804992675,
+      "num_input_tokens_seen": 125033152,
+      "step": 1225,
+      "train_runtime": 16937.5057,
+      "train_tokens_per_second": 7382.029
+    },
+    {
+      "epoch": 1.0594315245478036,
+      "grad_norm": 0.668586343087902,
+      "learning_rate": 4.540639411051528e-06,
+      "loss": 0.3157934665679932,
+      "num_input_tokens_seen": 125580048,
+      "step": 1230,
+      "train_runtime": 17000.2723,
+      "train_tokens_per_second": 7386.943
+    },
+    {
+      "epoch": 1.0637381567614126,
+      "grad_norm": 0.5699620351188704,
+      "learning_rate": 4.506969017350178e-06,
+      "loss": 0.304073166847229,
+      "num_input_tokens_seen": 126082048,
+      "step": 1235,
+      "train_runtime": 17068.426,
+      "train_tokens_per_second": 7386.859
+    },
+    {
+      "epoch": 1.0680447889750215,
+      "grad_norm": 0.6139561544523536,
+      "learning_rate": 4.473321186183091e-06,
+      "loss": 0.29829227924346924,
+      "num_input_tokens_seen": 126635088,
+      "step": 1240,
+      "train_runtime": 17133.6905,
+      "train_tokens_per_second": 7390.999
+    },
+    {
+      "epoch": 1.0723514211886305,
+      "grad_norm": 0.6987426072266698,
+      "learning_rate": 4.439697457373055e-06,
+      "loss": 0.2966867685317993,
+      "num_input_tokens_seen": 127113288,
+      "step": 1245,
+      "train_runtime": 17192.6486,
+      "train_tokens_per_second": 7393.467
+    },
+    {
+      "epoch": 1.0766580534022394,
+      "grad_norm": 0.663571335424868,
+      "learning_rate": 4.40609936963987e-06,
+      "loss": 0.30222883224487307,
+      "num_input_tokens_seen": 127635064,
+      "step": 1250,
+      "train_runtime": 17248.1503,
+      "train_tokens_per_second": 7399.928
+    },
+    {
+      "epoch": 1.0809646856158484,
+      "grad_norm": 0.6443165603778871,
+      "learning_rate": 4.37252846052992e-06,
+      "loss": 0.3315183877944946,
+      "num_input_tokens_seen": 128185744,
+      "step": 1255,
+      "train_runtime": 17313.1184,
+      "train_tokens_per_second": 7403.966
+    },
+    {
+      "epoch": 1.0852713178294573,
+      "grad_norm": 0.668552183385809,
+      "learning_rate": 4.338986266345819e-06,
+      "loss": 0.31603388786315917,
+      "num_input_tokens_seen": 128696104,
+      "step": 1260,
+      "train_runtime": 17374.9297,
+      "train_tokens_per_second": 7407.0
+    },
+    {
+      "epoch": 1.0895779500430662,
+      "grad_norm": 0.6529412865019517,
+      "learning_rate": 4.305474322076102e-06,
+      "loss": 0.29094948768615725,
+      "num_input_tokens_seen": 129191096,
+      "step": 1265,
+      "train_runtime": 17443.9341,
+      "train_tokens_per_second": 7406.076
+    },
+    {
+      "epoch": 1.0938845822566754,
+      "grad_norm": 0.7064731523627773,
+      "learning_rate": 4.271994161324977e-06,
+      "loss": 0.3133362293243408,
+      "num_input_tokens_seen": 129671408,
+      "step": 1270,
+      "train_runtime": 17512.3895,
+      "train_tokens_per_second": 7404.553
+    },
+    {
+      "epoch": 1.0981912144702843,
+      "grad_norm": 0.6668035883228985,
+      "learning_rate": 4.238547316242149e-06,
+      "loss": 0.3118987321853638,
+      "num_input_tokens_seen": 130159968,
+      "step": 1275,
+      "train_runtime": 17584.2318,
+      "train_tokens_per_second": 7402.084
+    },
+    {
+      "epoch": 1.1024978466838933,
+      "grad_norm": 0.6201182381091077,
+      "learning_rate": 4.205135317452701e-06,
+      "loss": 0.31302499771118164,
+      "num_input_tokens_seen": 130697176,
+      "step": 1280,
+      "train_runtime": 17654.8587,
+      "train_tokens_per_second": 7402.901
+    },
+    {
+      "epoch": 1.1068044788975022,
+      "grad_norm": 0.7787157406660087,
+      "learning_rate": 4.171759693987046e-06,
+      "loss": 0.3269864797592163,
+      "num_input_tokens_seen": 131192368,
+      "step": 1285,
+      "train_runtime": 17714.8718,
+      "train_tokens_per_second": 7405.776
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.6164244707522502,
+      "learning_rate": 4.1384219732109544e-06,
+      "loss": 0.3242049217224121,
+      "num_input_tokens_seen": 131711328,
+      "step": 1290,
+      "train_runtime": 17782.5318,
+      "train_tokens_per_second": 7406.782
+    },
+    {
+      "epoch": 1.11541774332472,
+      "grad_norm": 0.6247033321680014,
+      "learning_rate": 4.105123680755667e-06,
+      "loss": 0.30824267864227295,
+      "num_input_tokens_seen": 132247936,
+      "step": 1295,
+      "train_runtime": 17859.4257,
+      "train_tokens_per_second": 7404.938
+    },
+    {
+      "epoch": 1.119724375538329,
+      "grad_norm": 1.0131210118875211,
+      "learning_rate": 4.071866340448062e-06,
+      "loss": 0.3098948001861572,
+      "num_input_tokens_seen": 132771952,
+      "step": 1300,
+      "train_runtime": 17921.6225,
+      "train_tokens_per_second": 7408.478
+    },
+    {
+      "epoch": 1.124031007751938,
+      "grad_norm": 0.6160361056861662,
+      "learning_rate": 4.03865147424093e-06,
+      "loss": 0.312363338470459,
+      "num_input_tokens_seen": 133320696,
+      "step": 1305,
+      "train_runtime": 18000.4233,
+      "train_tokens_per_second": 7406.531
+    },
+    {
+      "epoch": 1.128337639965547,
+      "grad_norm": 0.6955496790672944,
+      "learning_rate": 4.0054806021433245e-06,
+      "loss": 0.3138570785522461,
+      "num_input_tokens_seen": 133811104,
+      "step": 1310,
+      "train_runtime": 18063.0877,
+      "train_tokens_per_second": 7407.986
+    },
+    {
+      "epoch": 1.1326442721791559,
+      "grad_norm": 0.6345676045747632,
+      "learning_rate": 3.9723552421509975e-06,
+      "loss": 0.2930002212524414,
+      "num_input_tokens_seen": 134319512,
+      "step": 1315,
+      "train_runtime": 18133.0035,
+      "train_tokens_per_second": 7407.461
+    },
+    {
+      "epoch": 1.1369509043927648,
+      "grad_norm": 0.6268160133086803,
+      "learning_rate": 3.939276910176939e-06,
+      "loss": 0.29963154792785646,
+      "num_input_tokens_seen": 134854032,
+      "step": 1320,
+      "train_runtime": 18213.0846,
+      "train_tokens_per_second": 7404.239
+    },
+    {
+      "epoch": 1.1412575366063737,
+      "grad_norm": 0.7417964459988887,
+      "learning_rate": 3.906247119981995e-06,
+      "loss": 0.31452503204345705,
+      "num_input_tokens_seen": 135369192,
+      "step": 1325,
+      "train_runtime": 18272.018,
+      "train_tokens_per_second": 7408.552
+    },
+    {
+      "epoch": 1.1455641688199827,
+      "grad_norm": 0.6724546111525185,
+      "learning_rate": 3.873267383105604e-06,
+      "loss": 0.3095171213150024,
+      "num_input_tokens_seen": 135915648,
+      "step": 1330,
+      "train_runtime": 18336.7572,
+      "train_tokens_per_second": 7412.197
+    },
+    {
+      "epoch": 1.1498708010335918,
+      "grad_norm": 0.6708401138495464,
+      "learning_rate": 3.840339208796611e-06,
+      "loss": 0.32448587417602537,
+      "num_input_tokens_seen": 136438696,
+      "step": 1335,
+      "train_runtime": 18410.0852,
+      "train_tokens_per_second": 7411.084
+    },
+    {
+      "epoch": 1.1541774332472006,
+      "grad_norm": 0.6395116916821753,
+      "learning_rate": 3.8074641039442138e-06,
+      "loss": 0.3084967613220215,
+      "num_input_tokens_seen": 136955600,
+      "step": 1340,
+      "train_runtime": 18478.1216,
+      "train_tokens_per_second": 7411.771
+    },
+    {
+      "epoch": 1.1584840654608097,
+      "grad_norm": 0.6729984961066279,
+      "learning_rate": 3.774643573008995e-06,
+      "loss": 0.28108384609222414,
+      "num_input_tokens_seen": 137454648,
+      "step": 1345,
+      "train_runtime": 18546.9342,
+      "train_tokens_per_second": 7411.179
+    },
+    {
+      "epoch": 1.1627906976744187,
+      "grad_norm": 0.6670203141700878,
+      "learning_rate": 3.741879117954077e-06,
+      "loss": 0.3180356502532959,
+      "num_input_tokens_seen": 137974504,
+      "step": 1350,
+      "train_runtime": 18616.3268,
+      "train_tokens_per_second": 7411.478
+    },
+    {
+      "epoch": 1.1670973298880276,
+      "grad_norm": 0.6874030328622756,
+      "learning_rate": 3.709172238176384e-06,
+      "loss": 0.2953655481338501,
+      "num_input_tokens_seen": 138476768,
+      "step": 1355,
+      "train_runtime": 18680.5568,
+      "train_tokens_per_second": 7412.882
+    },
+    {
+      "epoch": 1.1714039621016366,
+      "grad_norm": 0.6378347790279159,
+      "learning_rate": 3.6765244304380323e-06,
+      "loss": 0.31598858833312987,
+      "num_input_tokens_seen": 139004040,
+      "step": 1360,
+      "train_runtime": 18748.7212,
+      "train_tokens_per_second": 7414.054
+    },
+    {
+      "epoch": 1.1757105943152455,
+      "grad_norm": 0.6324830387437662,
+      "learning_rate": 3.643937188797826e-06,
+      "loss": 0.31764779090881345,
+      "num_input_tokens_seen": 139527672,
+      "step": 1365,
+      "train_runtime": 18813.059,
+      "train_tokens_per_second": 7416.533
+    },
+    {
+      "epoch": 1.1800172265288544,
+      "grad_norm": 0.6068454280272533,
+      "learning_rate": 3.611412004542887e-06,
+      "loss": 0.30363523960113525,
+      "num_input_tokens_seen": 140051536,
+      "step": 1370,
+      "train_runtime": 18882.0382,
+      "train_tokens_per_second": 7417.183
+    },
+    {
+      "epoch": 1.1843238587424634,
+      "grad_norm": 0.7822567824694513,
+      "learning_rate": 3.578950366120414e-06,
+      "loss": 0.3027554750442505,
+      "num_input_tokens_seen": 140523744,
+      "step": 1375,
+      "train_runtime": 18939.4105,
+      "train_tokens_per_second": 7419.647
+    },
+    {
+      "epoch": 1.1886304909560723,
+      "grad_norm": 0.7142925613608084,
+      "learning_rate": 3.54655375906956e-06,
+      "loss": 0.27700581550598147,
+      "num_input_tokens_seen": 140991200,
+      "step": 1380,
+      "train_runtime": 19015.4433,
+      "train_tokens_per_second": 7414.563
+    },
+    {
+      "epoch": 1.1929371231696813,
+      "grad_norm": 0.736983710804409,
+      "learning_rate": 3.514223665953455e-06,
+      "loss": 0.3062134742736816,
+      "num_input_tokens_seen": 141521152,
+      "step": 1385,
+      "train_runtime": 19085.2553,
+      "train_tokens_per_second": 7415.209
+    },
+    {
+      "epoch": 1.1972437553832902,
+      "grad_norm": 0.6552057330865397,
+      "learning_rate": 3.481961566291358e-06,
+      "loss": 0.2952254772186279,
+      "num_input_tokens_seen": 142052768,
+      "step": 1390,
+      "train_runtime": 19153.6423,
+      "train_tokens_per_second": 7416.489
+    },
+    {
+      "epoch": 1.2015503875968991,
+      "grad_norm": 0.6794659598243468,
+      "learning_rate": 3.4497689364909483e-06,
+      "loss": 0.30109152793884275,
+      "num_input_tokens_seen": 142577656,
+      "step": 1395,
+      "train_runtime": 19233.2813,
+      "train_tokens_per_second": 7413.07
+    },
+    {
+      "epoch": 1.2058570198105083,
+      "grad_norm": 0.5949831747224361,
+      "learning_rate": 3.417647249780761e-06,
+      "loss": 0.30313336849212646,
+      "num_input_tokens_seen": 143091504,
+      "step": 1400,
+      "train_runtime": 19289.193,
+      "train_tokens_per_second": 7418.221
+    },
+    {
+      "epoch": 1.210163652024117,
+      "grad_norm": 0.5687006661433506,
+      "learning_rate": 3.3855979761427705e-06,
+      "loss": 0.30561926364898684,
+      "num_input_tokens_seen": 143579880,
+      "step": 1405,
+      "train_runtime": 19352.506,
+      "train_tokens_per_second": 7419.188
+    },
+    {
+      "epoch": 1.2144702842377262,
+      "grad_norm": 0.5856929019081937,
+      "learning_rate": 3.3536225822451164e-06,
+      "loss": 0.3229517459869385,
+      "num_input_tokens_seen": 144116304,
+      "step": 1410,
+      "train_runtime": 19419.2134,
+      "train_tokens_per_second": 7421.326
+    },
+    {
+      "epoch": 1.2187769164513351,
+      "grad_norm": 0.7023895767304632,
+      "learning_rate": 3.321722531374988e-06,
+      "loss": 0.3148001432418823,
+      "num_input_tokens_seen": 144609456,
+      "step": 1415,
+      "train_runtime": 19497.6013,
+      "train_tokens_per_second": 7416.782
+    },
+    {
+      "epoch": 1.223083548664944,
+      "grad_norm": 0.7334327168318343,
+      "learning_rate": 3.289899283371657e-06,
+      "loss": 0.3218817710876465,
+      "num_input_tokens_seen": 145088128,
+      "step": 1420,
+      "train_runtime": 19562.664,
+      "train_tokens_per_second": 7416.583
+    },
+    {
+      "epoch": 1.227390180878553,
+      "grad_norm": 0.6787852523222232,
+      "learning_rate": 3.258154294559671e-06,
+      "loss": 0.3102770566940308,
+      "num_input_tokens_seen": 145589104,
+      "step": 1425,
+      "train_runtime": 19633.9236,
+      "train_tokens_per_second": 7415.181
+    },
+    {
+      "epoch": 1.231696813092162,
+      "grad_norm": 0.5985537130657856,
+      "learning_rate": 3.2264890176822137e-06,
+      "loss": 0.28916454315185547,
+      "num_input_tokens_seen": 146084280,
+      "step": 1430,
+      "train_runtime": 19704.3502,
+      "train_tokens_per_second": 7413.809
+    },
+    {
+      "epoch": 1.2360034453057709,
+      "grad_norm": 0.6605779036797204,
+      "learning_rate": 3.194904901834613e-06,
+      "loss": 0.2934903860092163,
+      "num_input_tokens_seen": 146581160,
+      "step": 1435,
+      "train_runtime": 19766.1983,
+      "train_tokens_per_second": 7415.749
+    },
+    {
+      "epoch": 1.2403100775193798,
+      "grad_norm": 0.6981230112849379,
+      "learning_rate": 3.1634033923980336e-06,
+      "loss": 0.2953122854232788,
+      "num_input_tokens_seen": 147076232,
+      "step": 1440,
+      "train_runtime": 19829.6949,
+      "train_tokens_per_second": 7416.969
+    },
+    {
+      "epoch": 1.2446167097329888,
+      "grad_norm": 0.6739980887459467,
+      "learning_rate": 3.131985930973329e-06,
+      "loss": 0.3041204929351807,
+      "num_input_tokens_seen": 147584128,
+      "step": 1445,
+      "train_runtime": 19894.1589,
+      "train_tokens_per_second": 7418.465
+    },
+    {
+      "epoch": 1.2489233419465977,
+      "grad_norm": 0.5697654101198504,
+      "learning_rate": 3.1006539553150727e-06,
+      "loss": 0.3094919204711914,
+      "num_input_tokens_seen": 148127248,
+      "step": 1450,
+      "train_runtime": 19971.1142,
+      "train_tokens_per_second": 7417.075
+    },
+    {
+      "epoch": 1.2532299741602067,
+      "grad_norm": 0.7121831368767915,
+      "learning_rate": 3.0694088992657617e-06,
+      "loss": 0.31122531890869143,
+      "num_input_tokens_seen": 148624776,
+      "step": 1455,
+      "train_runtime": 20037.7999,
+      "train_tokens_per_second": 7417.22
+    },
+    {
+      "epoch": 1.2575366063738156,
+      "grad_norm": 0.6188148109575211,
+      "learning_rate": 3.0382521926901946e-06,
+      "loss": 0.30587000846862794,
+      "num_input_tokens_seen": 149084784,
+      "step": 1460,
+      "train_runtime": 20101.6364,
+      "train_tokens_per_second": 7416.55
+    },
+    {
+      "epoch": 1.2618432385874248,
+      "grad_norm": 0.6011511034119341,
+      "learning_rate": 3.0071852614100427e-06,
+      "loss": 0.3310825347900391,
+      "num_input_tokens_seen": 149611248,
+      "step": 1465,
+      "train_runtime": 20175.4197,
+      "train_tokens_per_second": 7415.521
+    },
+    {
+      "epoch": 1.2661498708010335,
+      "grad_norm": 0.7576333400833488,
+      "learning_rate": 2.9762095271385995e-06,
+      "loss": 0.31157741546630857,
+      "num_input_tokens_seen": 150113280,
+      "step": 1470,
+      "train_runtime": 20245.8465,
+      "train_tokens_per_second": 7414.522
+    },
+    {
+      "epoch": 1.2704565030146426,
+      "grad_norm": 0.6244530019532303,
+      "learning_rate": 2.9453264074157134e-06,
+      "loss": 0.29090151786804197,
+      "num_input_tokens_seen": 150580240,
+      "step": 1475,
+      "train_runtime": 20317.2584,
+      "train_tokens_per_second": 7411.445
+    },
+    {
+      "epoch": 1.2747631352282516,
+      "grad_norm": 0.6748375953775086,
+      "learning_rate": 2.9145373155429263e-06,
+      "loss": 0.3190771102905273,
+      "num_input_tokens_seen": 151097168,
+      "step": 1480,
+      "train_runtime": 20380.9093,
+      "train_tokens_per_second": 7413.662
+    },
+    {
+      "epoch": 1.2790697674418605,
+      "grad_norm": 0.6010186642646648,
+      "learning_rate": 2.883843660518787e-06,
+      "loss": 0.31575181484222414,
+      "num_input_tokens_seen": 151620032,
+      "step": 1485,
+      "train_runtime": 20445.4483,
+      "train_tokens_per_second": 7415.833
+    },
+    {
+      "epoch": 1.2833763996554695,
+      "grad_norm": 0.6452504243127156,
+      "learning_rate": 2.853246846974382e-06,
+      "loss": 0.31672372817993166,
+      "num_input_tokens_seen": 152160432,
+      "step": 1490,
+      "train_runtime": 20522.0963,
+      "train_tokens_per_second": 7414.468
+    },
+    {
+      "epoch": 1.2876830318690784,
+      "grad_norm": 0.6803092051264894,
+      "learning_rate": 2.8227482751090445e-06,
+      "loss": 0.3057991027832031,
+      "num_input_tokens_seen": 152629192,
+      "step": 1495,
+      "train_runtime": 20576.0761,
+      "train_tokens_per_second": 7417.799
+    },
+    {
+      "epoch": 1.2919896640826873,
+      "grad_norm": 0.6639737361974689,
+      "learning_rate": 2.7923493406262836e-06,
+      "loss": 0.3172436237335205,
+      "num_input_tokens_seen": 153140224,
+      "step": 1500,
+      "train_runtime": 20645.2633,
+      "train_tokens_per_second": 7417.693
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.6227705768459294,
+      "learning_rate": 2.7620514346699103e-06,
+      "loss": 0.2983067512512207,
+      "num_input_tokens_seen": 153656144,
+      "step": 1505,
+      "train_runtime": 20719.2912,
+      "train_tokens_per_second": 7416.091
+    },
+    {
+      "epoch": 1.3006029285099052,
+      "grad_norm": 0.7307371833087898,
+      "learning_rate": 2.73185594376038e-06,
+      "loss": 0.31225831508636476,
+      "num_input_tokens_seen": 154153816,
+      "step": 1510,
+      "train_runtime": 20786.1098,
+      "train_tokens_per_second": 7416.194
+    },
+    {
+      "epoch": 1.3049095607235142,
+      "grad_norm": 0.6309746385583832,
+      "learning_rate": 2.7017642497313324e-06,
+      "loss": 0.302036190032959,
+      "num_input_tokens_seen": 154670920,
+      "step": 1515,
+      "train_runtime": 20854.4481,
+      "train_tokens_per_second": 7416.687
+    },
+    {
+      "epoch": 1.309216192937123,
+      "grad_norm": 0.6416098423112634,
+      "learning_rate": 2.671777729666363e-06,
+      "loss": 0.3167644262313843,
+      "num_input_tokens_seen": 155192016,
+      "step": 1520,
+      "train_runtime": 20930.5858,
+      "train_tokens_per_second": 7414.605
+    },
+    {
+      "epoch": 1.313522825150732,
+      "grad_norm": 0.6604523059153524,
+      "learning_rate": 2.641897755835997e-06,
+      "loss": 0.30224175453186036,
+      "num_input_tokens_seen": 155707640,
+      "step": 1525,
+      "train_runtime": 21004.7702,
+      "train_tokens_per_second": 7412.966
+    },
+    {
+      "epoch": 1.3178294573643412,
+      "grad_norm": 0.6284223784527019,
+      "learning_rate": 2.6121256956348966e-06,
+      "loss": 0.31644363403320314,
+      "num_input_tokens_seen": 156250728,
+      "step": 1530,
+      "train_runtime": 21079.0736,
+      "train_tokens_per_second": 7412.599
+    },
+    {
+      "epoch": 1.32213608957795,
+      "grad_norm": 0.6796758501164194,
+      "learning_rate": 2.58246291151928e-06,
+      "loss": 0.3076641082763672,
+      "num_input_tokens_seen": 156749112,
+      "step": 1535,
+      "train_runtime": 21151.8237,
+      "train_tokens_per_second": 7410.667
+    },
+    {
+      "epoch": 1.326442721791559,
+      "grad_norm": 0.6121495038952994,
+      "learning_rate": 2.5529107609445737e-06,
+      "loss": 0.3091968774795532,
+      "num_input_tokens_seen": 157258624,
+      "step": 1540,
+      "train_runtime": 21223.059,
+      "train_tokens_per_second": 7409.8
+    },
+    {
+      "epoch": 1.330749354005168,
+      "grad_norm": 0.6749697046548327,
+      "learning_rate": 2.5234705963032917e-06,
+      "loss": 0.31270341873168944,
+      "num_input_tokens_seen": 157737392,
+      "step": 1545,
+      "train_runtime": 21293.431,
+      "train_tokens_per_second": 7407.796
+    },
+    {
+      "epoch": 1.335055986218777,
+      "grad_norm": 0.7232023270557488,
+      "learning_rate": 2.494143764863142e-06,
+      "loss": 0.29509544372558594,
+      "num_input_tokens_seen": 158223688,
+      "step": 1550,
+      "train_runtime": 21357.9775,
+      "train_tokens_per_second": 7408.177
+    },
+    {
+      "epoch": 1.339362618432386,
+      "grad_norm": 0.6512809834688581,
+      "learning_rate": 2.464931608705384e-06,
+      "loss": 0.28979308605194093,
+      "num_input_tokens_seen": 158688096,
+      "step": 1555,
+      "train_runtime": 21415.5307,
+      "train_tokens_per_second": 7409.954
+    },
+    {
+      "epoch": 1.3436692506459949,
+      "grad_norm": 0.6519868412863893,
+      "learning_rate": 2.4358354646633915e-06,
+      "loss": 0.3088102102279663,
+      "num_input_tokens_seen": 159187424,
+      "step": 1560,
+      "train_runtime": 21483.2722,
+      "train_tokens_per_second": 7409.831
+    },
+    {
+      "epoch": 1.3479758828596038,
+      "grad_norm": 0.611925227867202,
+      "learning_rate": 2.4068566642614923e-06,
+      "loss": 0.29939565658569334,
+      "num_input_tokens_seen": 159669688,
+      "step": 1565,
+      "train_runtime": 21550.1291,
+      "train_tokens_per_second": 7409.222
+    },
+    {
+      "epoch": 1.3522825150732127,
+      "grad_norm": 0.6563878117514377,
+      "learning_rate": 2.3779965336540237e-06,
+      "loss": 0.3042983770370483,
+      "num_input_tokens_seen": 160172312,
+      "step": 1570,
+      "train_runtime": 21611.6367,
+      "train_tokens_per_second": 7411.392
+    },
+    {
+      "epoch": 1.3565891472868217,
+      "grad_norm": 0.6120474875361092,
+      "learning_rate": 2.3492563935646493e-06,
+      "loss": 0.2935898542404175,
+      "num_input_tokens_seen": 160708952,
+      "step": 1575,
+      "train_runtime": 21688.22,
+      "train_tokens_per_second": 7409.965
+    },
+    {
+      "epoch": 1.3608957795004306,
+      "grad_norm": 0.607518756395221,
+      "learning_rate": 2.3206375592259205e-06,
+      "loss": 0.30011003017425536,
+      "num_input_tokens_seen": 161217216,
+      "step": 1580,
+      "train_runtime": 21761.1281,
+      "train_tokens_per_second": 7408.495
+    },
+    {
+      "epoch": 1.3652024117140396,
+      "grad_norm": 0.6300850271418421,
+      "learning_rate": 2.2921413403190774e-06,
+      "loss": 0.3024036645889282,
+      "num_input_tokens_seen": 161705312,
+      "step": 1585,
+      "train_runtime": 21836.3225,
+      "train_tokens_per_second": 7405.336
+    },
+    {
+      "epoch": 1.3695090439276485,
+      "grad_norm": 0.6230690301238715,
+      "learning_rate": 2.263769040914125e-06,
+      "loss": 0.30027027130126954,
+      "num_input_tokens_seen": 162163072,
+      "step": 1590,
+      "train_runtime": 21896.5811,
+      "train_tokens_per_second": 7405.863
+    },
+    {
+      "epoch": 1.3738156761412577,
+      "grad_norm": 0.6928798019826016,
+      "learning_rate": 2.2355219594101483e-06,
+      "loss": 0.3255805492401123,
+      "num_input_tokens_seen": 162678312,
+      "step": 1595,
+      "train_runtime": 21966.5196,
+      "train_tokens_per_second": 7405.739
+    },
+    {
+      "epoch": 1.3781223083548664,
+      "grad_norm": 0.6363368596470697,
+      "learning_rate": 2.2074013884758993e-06,
+      "loss": 0.31427993774414065,
+      "num_input_tokens_seen": 163224632,
+      "step": 1600,
+      "train_runtime": 22032.9938,
+      "train_tokens_per_second": 7408.191
+    },
+    {
+      "epoch": 1.3824289405684755,
+      "grad_norm": 0.6419634556123296,
+      "learning_rate": 2.179408614990635e-06,
+      "loss": 0.3022661924362183,
+      "num_input_tokens_seen": 163699352,
+      "step": 1605,
+      "train_runtime": 22099.9328,
+      "train_tokens_per_second": 7407.233
+    },
+    {
+      "epoch": 1.3867355727820845,
+      "grad_norm": 0.5818964338969693,
+      "learning_rate": 2.1515449199852307e-06,
+      "loss": 0.31579549312591554,
+      "num_input_tokens_seen": 164236200,
+      "step": 1610,
+      "train_runtime": 22170.2317,
+      "train_tokens_per_second": 7407.96
+    },
+    {
+      "epoch": 1.3910422049956934,
+      "grad_norm": 0.6207480539135273,
+      "learning_rate": 2.1238115785835512e-06,
+      "loss": 0.31688437461853025,
+      "num_input_tokens_seen": 164733432,
+      "step": 1615,
+      "train_runtime": 22241.2909,
+      "train_tokens_per_second": 7406.649
+    },
+    {
+      "epoch": 1.3953488372093024,
+      "grad_norm": 0.6195171977696273,
+      "learning_rate": 2.0962098599441065e-06,
+      "loss": 0.3063546657562256,
+      "num_input_tokens_seen": 165233040,
+      "step": 1620,
+      "train_runtime": 22304.129,
+      "train_tokens_per_second": 7408.182
+    },
+    {
+      "epoch": 1.3996554694229113,
+      "grad_norm": 0.6192636575391592,
+      "learning_rate": 2.068741027201961e-06,
+      "loss": 0.28777213096618653,
+      "num_input_tokens_seen": 165704392,
+      "step": 1625,
+      "train_runtime": 22371.6308,
+      "train_tokens_per_second": 7406.898
+    },
+    {
+      "epoch": 1.4039621016365202,
+      "grad_norm": 0.5840180544779381,
+      "learning_rate": 2.0414063374109326e-06,
+      "loss": 0.3152973413467407,
+      "num_input_tokens_seen": 166213024,
+      "step": 1630,
+      "train_runtime": 22439.0022,
+      "train_tokens_per_second": 7407.327
+    },
+    {
+      "epoch": 1.4082687338501292,
+      "grad_norm": 0.6182614190184378,
+      "learning_rate": 2.0142070414860704e-06,
+      "loss": 0.29006712436676024,
+      "num_input_tokens_seen": 166683888,
+      "step": 1635,
+      "train_runtime": 22498.0177,
+      "train_tokens_per_second": 7408.826
+    },
+    {
+      "epoch": 1.4125753660637381,
+      "grad_norm": 0.6297841421388773,
+      "learning_rate": 1.9871443841464004e-06,
+      "loss": 0.3009150505065918,
+      "num_input_tokens_seen": 167208664,
+      "step": 1640,
+      "train_runtime": 22570.3637,
+      "train_tokens_per_second": 7408.328
+    },
+    {
+      "epoch": 1.416881998277347,
+      "grad_norm": 0.6550334755642108,
+      "learning_rate": 1.9602196038579774e-06,
+      "loss": 0.308090877532959,
+      "num_input_tokens_seen": 167711896,
+      "step": 1645,
+      "train_runtime": 22636.4806,
+      "train_tokens_per_second": 7408.921
+    },
+    {
+      "epoch": 1.421188630490956,
+      "grad_norm": 0.6656593254119286,
+      "learning_rate": 1.933433932777193e-06,
+      "loss": 0.31527588367462156,
+      "num_input_tokens_seen": 168253824,
+      "step": 1650,
+      "train_runtime": 22710.0595,
+      "train_tokens_per_second": 7408.78
+    },
+    {
+      "epoch": 1.425495262704565,
+      "grad_norm": 0.6870898791567921,
+      "learning_rate": 1.9067885966943983e-06,
+      "loss": 0.2953861474990845,
+      "num_input_tokens_seen": 168753216,
+      "step": 1655,
+      "train_runtime": 22778.2895,
+      "train_tokens_per_second": 7408.511
+    },
+    {
+      "epoch": 1.429801894918174,
+      "grad_norm": 0.6715846070836667,
+      "learning_rate": 1.880284814977807e-06,
+      "loss": 0.3124719142913818,
+      "num_input_tokens_seen": 169237536,
+      "step": 1660,
+      "train_runtime": 22844.2856,
+      "train_tokens_per_second": 7408.309
+    },
+    {
+      "epoch": 1.4341085271317828,
+      "grad_norm": 0.5885167014186917,
+      "learning_rate": 1.8539238005176912e-06,
+      "loss": 0.28069252967834474,
+      "num_input_tokens_seen": 169742192,
+      "step": 1665,
+      "train_runtime": 22917.8396,
+      "train_tokens_per_second": 7406.553
+    },
+    {
+      "epoch": 1.438415159345392,
+      "grad_norm": 0.6406321645578682,
+      "learning_rate": 1.8277067596708793e-06,
+      "loss": 0.30810205936431884,
+      "num_input_tokens_seen": 170221760,
+      "step": 1670,
+      "train_runtime": 22987.4305,
+      "train_tokens_per_second": 7404.993
+    },
+    {
+      "epoch": 1.442721791559001,
+      "grad_norm": 0.6177250258017325,
+      "learning_rate": 1.8016348922055448e-06,
+      "loss": 0.3010256767272949,
+      "num_input_tokens_seen": 170720016,
+      "step": 1675,
+      "train_runtime": 23056.7807,
+      "train_tokens_per_second": 7404.33
+    },
+    {
+      "epoch": 1.4470284237726099,
+      "grad_norm": 0.6542576135864537,
+      "learning_rate": 1.7757093912463058e-06,
+      "loss": 0.30887162685394287,
+      "num_input_tokens_seen": 171232760,
+      "step": 1680,
+      "train_runtime": 23123.5146,
+      "train_tokens_per_second": 7405.136
+    },
+    {
+      "epoch": 1.4513350559862188,
+      "grad_norm": 0.6205881235437254,
+      "learning_rate": 1.7499314432196257e-06,
+      "loss": 0.3016911506652832,
+      "num_input_tokens_seen": 171731496,
+      "step": 1685,
+      "train_runtime": 23196.0481,
+      "train_tokens_per_second": 7403.481
+    },
+    {
+      "epoch": 1.4556416881998278,
+      "grad_norm": 0.6593735329024515,
+      "learning_rate": 1.7243022277995109e-06,
+      "loss": 0.3081092357635498,
+      "num_input_tokens_seen": 172245912,
+      "step": 1690,
+      "train_runtime": 23270.0505,
+      "train_tokens_per_second": 7402.043
+    },
+    {
+      "epoch": 1.4599483204134367,
+      "grad_norm": 0.6421196878908372,
+      "learning_rate": 1.698822917853532e-06,
+      "loss": 0.3349776268005371,
+      "num_input_tokens_seen": 172780776,
+      "step": 1695,
+      "train_runtime": 23338.4767,
+      "train_tokens_per_second": 7403.259
+    },
+    {
+      "epoch": 1.4642549526270456,
+      "grad_norm": 0.6393067681048293,
+      "learning_rate": 1.6734946793891498e-06,
+      "loss": 0.2948931694030762,
+      "num_input_tokens_seen": 173260968,
+      "step": 1700,
+      "train_runtime": 23393.7968,
+      "train_tokens_per_second": 7406.278
+    },
+    {
+      "epoch": 1.4685615848406546,
+      "grad_norm": 0.626157385220703,
+      "learning_rate": 1.6483186715003523e-06,
+      "loss": 0.3255448818206787,
+      "num_input_tokens_seen": 173774184,
+      "step": 1705,
+      "train_runtime": 23453.7592,
+      "train_tokens_per_second": 7409.225
+    },
+    {
+      "epoch": 1.4728682170542635,
+      "grad_norm": 0.6000516868104079,
+      "learning_rate": 1.623296046314612e-06,
+      "loss": 0.3152692556381226,
+      "num_input_tokens_seen": 174299296,
+      "step": 1710,
+      "train_runtime": 23525.525,
+      "train_tokens_per_second": 7408.944
+    },
+    {
+      "epoch": 1.4771748492678725,
+      "grad_norm": 0.6223596745571496,
+      "learning_rate": 1.5984279489401655e-06,
+      "loss": 0.3075227737426758,
+      "num_input_tokens_seen": 174801056,
+      "step": 1715,
+      "train_runtime": 23599.8686,
+      "train_tokens_per_second": 7406.866
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.6218964277556629,
+      "learning_rate": 1.5737155174136042e-06,
+      "loss": 0.2994909524917603,
+      "num_input_tokens_seen": 175312064,
+      "step": 1720,
+      "train_runtime": 23669.1828,
+      "train_tokens_per_second": 7406.765
+    },
+    {
+      "epoch": 1.4857881136950906,
+      "grad_norm": 0.6927663365147764,
+      "learning_rate": 1.5491598826477967e-06,
+      "loss": 0.2942391872406006,
+      "num_input_tokens_seen": 175813120,
+      "step": 1725,
+      "train_runtime": 23735.9044,
+      "train_tokens_per_second": 7407.054
+    },
+    {
+      "epoch": 1.4900947459086993,
+      "grad_norm": 0.6770702789982874,
+      "learning_rate": 1.5247621683801344e-06,
+      "loss": 0.30359663963317873,
+      "num_input_tokens_seen": 176311208,
+      "step": 1730,
+      "train_runtime": 23813.1099,
+      "train_tokens_per_second": 7403.956
+    },
+    {
+      "epoch": 1.4944013781223084,
+      "grad_norm": 0.6876682910616844,
+      "learning_rate": 1.500523491121108e-06,
+      "loss": 0.30945577621459963,
+      "num_input_tokens_seen": 176837256,
+      "step": 1735,
+      "train_runtime": 23875.8608,
+      "train_tokens_per_second": 7406.529
+    },
+    {
+      "epoch": 1.4987080103359174,
+      "grad_norm": 0.6409833412888383,
+      "learning_rate": 1.476444960103211e-06,
+      "loss": 0.3017910957336426,
+      "num_input_tokens_seen": 177342432,
+      "step": 1740,
+      "train_runtime": 23950.1948,
+      "train_tokens_per_second": 7404.634
+    },
+    {
+      "epoch": 1.5030146425495263,
+      "grad_norm": 0.6943474108994365,
+      "learning_rate": 1.4525276772301761e-06,
+      "loss": 0.2982076644897461,
+      "num_input_tokens_seen": 177837496,
+      "step": 1745,
+      "train_runtime": 24022.3232,
+      "train_tokens_per_second": 7403.01
+    },
+    {
+      "epoch": 1.5073212747631353,
+      "grad_norm": 0.5799829679643863,
+      "learning_rate": 1.4287727370265558e-06,
+      "loss": 0.305989408493042,
+      "num_input_tokens_seen": 178367400,
+      "step": 1750,
+      "train_runtime": 24098.0102,
+      "train_tokens_per_second": 7401.748
+    },
+    {
+      "epoch": 1.5116279069767442,
+      "grad_norm": 0.6398941855178756,
+      "learning_rate": 1.4051812265876257e-06,
+      "loss": 0.31029796600341797,
+      "num_input_tokens_seen": 178879808,
+      "step": 1755,
+      "train_runtime": 24168.3597,
+      "train_tokens_per_second": 7401.405
+    },
+    {
+      "epoch": 1.5159345391903531,
+      "grad_norm": 0.6325656951057348,
+      "learning_rate": 1.381754225529639e-06,
+      "loss": 0.29440352916717527,
+      "num_input_tokens_seen": 179399168,
+      "step": 1760,
+      "train_runtime": 24235.603,
+      "train_tokens_per_second": 7402.299
+    },
+    {
+      "epoch": 1.520241171403962,
+      "grad_norm": 0.7760734610328013,
+      "learning_rate": 1.3584928059404207e-06,
+      "loss": 0.3081679344177246,
+      "num_input_tokens_seen": 179913984,
+      "step": 1765,
+      "train_runtime": 24301.9048,
+      "train_tokens_per_second": 7403.287
+    },
+    {
+      "epoch": 1.524547803617571,
+      "grad_norm": 0.6020927671705726,
+      "learning_rate": 1.3353980323303067e-06,
+      "loss": 0.28179948329925536,
+      "num_input_tokens_seen": 180380992,
+      "step": 1770,
+      "train_runtime": 24364.2331,
+      "train_tokens_per_second": 7403.516
+    },
+    {
+      "epoch": 1.52885443583118,
+      "grad_norm": 0.6775155388063839,
+      "learning_rate": 1.3124709615834263e-06,
+      "loss": 0.3199465274810791,
+      "num_input_tokens_seen": 180895408,
+      "step": 1775,
+      "train_runtime": 24428.2069,
+      "train_tokens_per_second": 7405.186
+    },
+    {
+      "epoch": 1.5331610680447891,
+      "grad_norm": 0.5909440385527984,
+      "learning_rate": 1.2897126429093354e-06,
+      "loss": 0.31137895584106445,
+      "num_input_tokens_seen": 181405216,
+      "step": 1780,
+      "train_runtime": 24501.3699,
+      "train_tokens_per_second": 7403.881
+    },
+    {
+      "epoch": 1.5374677002583979,
+      "grad_norm": 0.6389531630672336,
+      "learning_rate": 1.2671241177950078e-06,
+      "loss": 0.2975735902786255,
+      "num_input_tokens_seen": 181929944,
+      "step": 1785,
+      "train_runtime": 24574.289,
+      "train_tokens_per_second": 7403.264
+    },
+    {
+      "epoch": 1.541774332472007,
+      "grad_norm": 0.6814051618467906,
+      "learning_rate": 1.2447064199571657e-06,
+      "loss": 0.30163776874542236,
+      "num_input_tokens_seen": 182464632,
+      "step": 1790,
+      "train_runtime": 24647.5158,
+      "train_tokens_per_second": 7402.962
+    },
+    {
+      "epoch": 1.5460809646856157,
+      "grad_norm": 0.7477112566892565,
+      "learning_rate": 1.2224605752949786e-06,
+      "loss": 0.2984232664108276,
+      "num_input_tokens_seen": 182958712,
+      "step": 1795,
+      "train_runtime": 24708.0115,
+      "train_tokens_per_second": 7404.834
+    },
+    {
+      "epoch": 1.550387596899225,
+      "grad_norm": 0.6529963125920666,
+      "learning_rate": 1.2003876018431133e-06,
+      "loss": 0.29271659851074217,
+      "num_input_tokens_seen": 183435376,
+      "step": 1800,
+      "train_runtime": 24760.234,
+      "train_tokens_per_second": 7408.467
+    },
+    {
+      "epoch": 1.5546942291128336,
+      "grad_norm": 0.5835418021024796,
+      "learning_rate": 1.1784885097251474e-06,
+      "loss": 0.2989810466766357,
+      "num_input_tokens_seen": 183939080,
+      "step": 1805,
+      "train_runtime": 24816.2993,
+      "train_tokens_per_second": 7412.027
+    },
+    {
+      "epoch": 1.5590008613264428,
+      "grad_norm": 0.6843482118934102,
+      "learning_rate": 1.1567643011073393e-06,
+      "loss": 0.31319284439086914,
+      "num_input_tokens_seen": 184492224,
+      "step": 1810,
+      "train_runtime": 24887.0045,
+      "train_tokens_per_second": 7413.195
+    },
+    {
+      "epoch": 1.5633074935400517,
+      "grad_norm": 0.6189022938446993,
+      "learning_rate": 1.1352159701527743e-06,
+      "loss": 0.3045823574066162,
+      "num_input_tokens_seen": 185017248,
+      "step": 1815,
+      "train_runtime": 24965.7799,
+      "train_tokens_per_second": 7410.834
+    },
+    {
+      "epoch": 1.5676141257536607,
+      "grad_norm": 0.6412913917213614,
+      "learning_rate": 1.1138445029758594e-06,
+      "loss": 0.29555253982543944,
+      "num_input_tokens_seen": 185525392,
+      "step": 1820,
+      "train_runtime": 25039.0119,
+      "train_tokens_per_second": 7409.453
+    },
+    {
+      "epoch": 1.5719207579672696,
+      "grad_norm": 0.6383974503891945,
+      "learning_rate": 1.0926508775971995e-06,
+      "loss": 0.29001967906951903,
+      "num_input_tokens_seen": 186004920,
+      "step": 1825,
+      "train_runtime": 25108.8304,
+      "train_tokens_per_second": 7407.948
+    },
+    {
+      "epoch": 1.5762273901808785,
+      "grad_norm": 0.5896126682457972,
+      "learning_rate": 1.0716360638988416e-06,
+      "loss": 0.2898190259933472,
+      "num_input_tokens_seen": 186528776,
+      "step": 1830,
+      "train_runtime": 25173.4594,
+      "train_tokens_per_second": 7409.739
+    },
+    {
+      "epoch": 1.5805340223944875,
+      "grad_norm": 0.6250987792113757,
+      "learning_rate": 1.0508010235798904e-06,
+      "loss": 0.3022593975067139,
+      "num_input_tokens_seen": 187030096,
+      "step": 1835,
+      "train_runtime": 25237.5065,
+      "train_tokens_per_second": 7410.799
+    },
+    {
+      "epoch": 1.5848406546080964,
+      "grad_norm": 0.6978498804363219,
+      "learning_rate": 1.0301467101124956e-06,
+      "loss": 0.29773921966552735,
+      "num_input_tokens_seen": 187530704,
+      "step": 1840,
+      "train_runtime": 25300.156,
+      "train_tokens_per_second": 7412.235
+    },
+    {
+      "epoch": 1.5891472868217056,
+      "grad_norm": 0.733936599658224,
+      "learning_rate": 1.0096740686982192e-06,
+      "loss": 0.2982914447784424,
+      "num_input_tokens_seen": 187997312,
+      "step": 1845,
+      "train_runtime": 25374.7743,
+      "train_tokens_per_second": 7408.827
+    },
+    {
+      "epoch": 1.5934539190353143,
+      "grad_norm": 0.6380738781500264,
+      "learning_rate": 9.893840362247809e-07,
+      "loss": 0.292753005027771,
+      "num_input_tokens_seen": 188486784,
+      "step": 1850,
+      "train_runtime": 25443.3247,
+      "train_tokens_per_second": 7408.104
+    },
+    {
+      "epoch": 1.5977605512489235,
+      "grad_norm": 0.6343838191463229,
+      "learning_rate": 9.692775412231863e-07,
+      "loss": 0.3059796571731567,
+      "num_input_tokens_seen": 189005680,
+      "step": 1855,
+      "train_runtime": 25518.9616,
+      "train_tokens_per_second": 7406.48
+    },
+    {
+      "epoch": 1.6020671834625322,
+      "grad_norm": 0.673292925225458,
+      "learning_rate": 9.493555038252273e-07,
+      "loss": 0.29838857650756834,
+      "num_input_tokens_seen": 189510720,
+      "step": 1860,
+      "train_runtime": 25583.3652,
+      "train_tokens_per_second": 7407.576
+    },
+    {
+      "epoch": 1.6063738156761413,
+      "grad_norm": 0.7066024445607715,
+      "learning_rate": 9.296188357213804e-07,
+      "loss": 0.3135526180267334,
+      "num_input_tokens_seen": 190005232,
+      "step": 1865,
+      "train_runtime": 25655.2288,
+      "train_tokens_per_second": 7406.102
+    },
+    {
+      "epoch": 1.61068044788975,
+      "grad_norm": 0.6451799213539555,
+      "learning_rate": 9.100684401190829e-07,
+      "loss": 0.2981297492980957,
+      "num_input_tokens_seen": 190519568,
+      "step": 1870,
+      "train_runtime": 25729.8902,
+      "train_tokens_per_second": 7404.601
+    },
+    {
+      "epoch": 1.6149870801033592,
+      "grad_norm": 0.6386914239692977,
+      "learning_rate": 8.907052117013981e-07,
+      "loss": 0.3049411058425903,
+      "num_input_tokens_seen": 191051376,
+      "step": 1875,
+      "train_runtime": 25804.946,
+      "train_tokens_per_second": 7403.673
+    },
+    {
+      "epoch": 1.6192937123169682,
+      "grad_norm": 0.7923656421587169,
+      "learning_rate": 8.715300365860784e-07,
+      "loss": 0.2883660316467285,
+      "num_input_tokens_seen": 191555640,
+      "step": 1880,
+      "train_runtime": 25872.9743,
+      "train_tokens_per_second": 7403.696
+    },
+    {
+      "epoch": 1.623600344530577,
+      "grad_norm": 0.5861974629399523,
+      "learning_rate": 8.525437922850033e-07,
+      "loss": 0.2918602228164673,
+      "num_input_tokens_seen": 192040232,
+      "step": 1885,
+      "train_runtime": 25940.9831,
+      "train_tokens_per_second": 7402.967
+    },
+    {
+      "epoch": 1.627906976744186,
+      "grad_norm": 0.6536777995059396,
+      "learning_rate": 8.337473476640318e-07,
+      "loss": 0.296920108795166,
+      "num_input_tokens_seen": 192555328,
+      "step": 1890,
+      "train_runtime": 25999.4311,
+      "train_tokens_per_second": 7406.136
+    },
+    {
+      "epoch": 1.632213608957795,
+      "grad_norm": 0.709106460699193,
+      "learning_rate": 8.151415629032338e-07,
+      "loss": 0.3084681749343872,
+      "num_input_tokens_seen": 193047904,
+      "step": 1895,
+      "train_runtime": 26071.7257,
+      "train_tokens_per_second": 7404.493
+    },
+    {
+      "epoch": 1.636520241171404,
+      "grad_norm": 0.6265073195440819,
+      "learning_rate": 7.967272894575312e-07,
+      "loss": 0.3049050807952881,
+      "num_input_tokens_seen": 193578488,
+      "step": 1900,
+      "train_runtime": 26142.2043,
+      "train_tokens_per_second": 7404.827
+    },
+    {
+      "epoch": 1.6408268733850129,
+      "grad_norm": 0.5901143607833178,
+      "learning_rate": 7.785053700177275e-07,
+      "loss": 0.31264266967773435,
+      "num_input_tokens_seen": 194106056,
+      "step": 1905,
+      "train_runtime": 26207.0856,
+      "train_tokens_per_second": 7406.625
+    },
+    {
+      "epoch": 1.645133505598622,
+      "grad_norm": 0.6070076296317176,
+      "learning_rate": 7.604766384719487e-07,
+      "loss": 0.32028186321258545,
+      "num_input_tokens_seen": 194637600,
+      "step": 1910,
+      "train_runtime": 26271.2516,
+      "train_tokens_per_second": 7408.768
+    },
+    {
+      "epoch": 1.6494401378122308,
+      "grad_norm": 0.6800282511844107,
+      "learning_rate": 7.426419198674773e-07,
+      "loss": 0.29713547229766846,
+      "num_input_tokens_seen": 195135888,
+      "step": 1915,
+      "train_runtime": 26344.3031,
+      "train_tokens_per_second": 7407.138
+    },
+    {
+      "epoch": 1.65374677002584,
+      "grad_norm": 0.648137387161232,
+      "learning_rate": 7.250020303730015e-07,
+      "loss": 0.318635368347168,
+      "num_input_tokens_seen": 195658344,
+      "step": 1920,
+      "train_runtime": 26410.9309,
+      "train_tokens_per_second": 7408.234
+    },
+    {
+      "epoch": 1.6580534022394486,
+      "grad_norm": 0.7101813432427155,
+      "learning_rate": 7.075577772412607e-07,
+      "loss": 0.29038522243499754,
+      "num_input_tokens_seen": 196159872,
+      "step": 1925,
+      "train_runtime": 26475.9547,
+      "train_tokens_per_second": 7408.982
+    },
+    {
+      "epoch": 1.6623600344530578,
+      "grad_norm": 0.5997169278780675,
+      "learning_rate": 6.903099587721024e-07,
+      "loss": 0.3107494831085205,
+      "num_input_tokens_seen": 196617976,
+      "step": 1930,
+      "train_runtime": 26543.6519,
+      "train_tokens_per_second": 7407.345
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.5757077293663654,
+      "learning_rate": 6.732593642759533e-07,
+      "loss": 0.29712848663330077,
+      "num_input_tokens_seen": 197100472,
+      "step": 1935,
+      "train_runtime": 26613.1626,
+      "train_tokens_per_second": 7406.127
+    },
+    {
+      "epoch": 1.6709732988802757,
+      "grad_norm": 0.6672790463782956,
+      "learning_rate": 6.564067740376951e-07,
+      "loss": 0.29893629550933837,
+      "num_input_tokens_seen": 197626872,
+      "step": 1940,
+      "train_runtime": 26679.8578,
+      "train_tokens_per_second": 7407.344
+    },
+    {
+      "epoch": 1.6752799310938846,
+      "grad_norm": 0.6497288298918169,
+      "learning_rate": 6.397529592809615e-07,
+      "loss": 0.31168351173400877,
+      "num_input_tokens_seen": 198123984,
+      "step": 1945,
+      "train_runtime": 26746.7628,
+      "train_tokens_per_second": 7407.4
+    },
+    {
+      "epoch": 1.6795865633074936,
+      "grad_norm": 0.8599577731412777,
+      "learning_rate": 6.232986821328374e-07,
+      "loss": 0.29562864303588865,
+      "num_input_tokens_seen": 198655680,
+      "step": 1950,
+      "train_runtime": 26823.6042,
+      "train_tokens_per_second": 7406.003
+    },
+    {
+      "epoch": 1.6838931955211025,
+      "grad_norm": 0.650247687298536,
+      "learning_rate": 6.070446955889853e-07,
+      "loss": 0.2916395664215088,
+      "num_input_tokens_seen": 199167544,
+      "step": 1955,
+      "train_runtime": 26896.5949,
+      "train_tokens_per_second": 7404.935
+    },
+    {
+      "epoch": 1.6881998277347114,
+      "grad_norm": 0.642603913967143,
+      "learning_rate": 5.909917434791884e-07,
+      "loss": 0.30132670402526857,
+      "num_input_tokens_seen": 199673096,
+      "step": 1960,
+      "train_runtime": 26963.0457,
+      "train_tokens_per_second": 7405.436
+    },
+    {
+      "epoch": 1.6925064599483204,
+      "grad_norm": 0.6050771009316471,
+      "learning_rate": 5.75140560433306e-07,
+      "loss": 0.3065808296203613,
+      "num_input_tokens_seen": 200183744,
+      "step": 1965,
+      "train_runtime": 27028.667,
+      "train_tokens_per_second": 7406.349
+    },
+    {
+      "epoch": 1.6968130921619293,
+      "grad_norm": 0.7019989661295893,
+      "learning_rate": 5.594918718476589e-07,
+      "loss": 0.30456061363220216,
+      "num_input_tokens_seen": 200697696,
+      "step": 1970,
+      "train_runtime": 27101.6197,
+      "train_tokens_per_second": 7405.376
+    },
+    {
+      "epoch": 1.7011197243755385,
+      "grad_norm": 0.6625920188017488,
+      "learning_rate": 5.440463938518304e-07,
+      "loss": 0.3014059066772461,
+      "num_input_tokens_seen": 201206880,
+      "step": 1975,
+      "train_runtime": 27172.9204,
+      "train_tokens_per_second": 7404.684
+    },
+    {
+      "epoch": 1.7054263565891472,
+      "grad_norm": 0.6748217547745429,
+      "learning_rate": 5.288048332758961e-07,
+      "loss": 0.2906249284744263,
+      "num_input_tokens_seen": 201692448,
+      "step": 1980,
+      "train_runtime": 27245.0858,
+      "train_tokens_per_second": 7402.893
+    },
+    {
+      "epoch": 1.7097329888027564,
+      "grad_norm": 0.6493936382897342,
+      "learning_rate": 5.137678876180746e-07,
+      "loss": 0.3080711841583252,
+      "num_input_tokens_seen": 202206200,
+      "step": 1985,
+      "train_runtime": 27302.2262,
+      "train_tokens_per_second": 7406.217
+    },
+    {
+      "epoch": 1.714039621016365,
+      "grad_norm": 0.6134604818431901,
+      "learning_rate": 4.989362450128133e-07,
+      "loss": 0.3220229148864746,
+      "num_input_tokens_seen": 202716392,
+      "step": 1990,
+      "train_runtime": 27369.6446,
+      "train_tokens_per_second": 7406.614
+    },
+    {
+      "epoch": 1.7183462532299743,
+      "grad_norm": 0.646962297611449,
+      "learning_rate": 4.843105841992895e-07,
+      "loss": 0.31665678024291993,
+      "num_input_tokens_seen": 203222616,
+      "step": 1995,
+      "train_runtime": 27428.8755,
+      "train_tokens_per_second": 7409.076
+    },
+    {
+      "epoch": 1.722652885443583,
+      "grad_norm": 0.6371911776149713,
+      "learning_rate": 4.698915744903559e-07,
+      "loss": 0.2909515380859375,
+      "num_input_tokens_seen": 203712976,
+      "step": 2000,
+      "train_runtime": 27503.273,
+      "train_tokens_per_second": 7406.863
+    },
+    {
+      "epoch": 1.7269595176571921,
+      "grad_norm": 0.6718397087570519,
+      "learning_rate": 4.5567987574190677e-07,
+      "loss": 0.299733304977417,
+      "num_input_tokens_seen": 204220640,
+      "step": 2005,
+      "train_runtime": 27582.7926,
+      "train_tokens_per_second": 7403.915
+    },
+    {
+      "epoch": 1.731266149870801,
+      "grad_norm": 0.5969000639721187,
+      "learning_rate": 4.416761383226864e-07,
+      "loss": 0.2933858156204224,
+      "num_input_tokens_seen": 204740576,
+      "step": 2010,
+      "train_runtime": 27655.0348,
+      "train_tokens_per_second": 7403.374
+    },
+    {
+      "epoch": 1.73557278208441,
+      "grad_norm": 0.5803238112892645,
+      "learning_rate": 4.278810030845193e-07,
+      "loss": 0.2882672309875488,
+      "num_input_tokens_seen": 205261256,
+      "step": 2015,
+      "train_runtime": 27722.013,
+      "train_tokens_per_second": 7404.27
+    },
+    {
+      "epoch": 1.739879414298019,
+      "grad_norm": 0.5931274387562315,
+      "learning_rate": 4.1429510133298714e-07,
+      "loss": 0.300402307510376,
+      "num_input_tokens_seen": 205763456,
+      "step": 2020,
+      "train_runtime": 27786.5499,
+      "train_tokens_per_second": 7405.146
+    },
+    {
+      "epoch": 1.744186046511628,
+      "grad_norm": 0.6224395945864603,
+      "learning_rate": 4.0091905479853865e-07,
+      "loss": 0.30848591327667235,
+      "num_input_tokens_seen": 206254632,
+      "step": 2025,
+      "train_runtime": 27854.9195,
+      "train_tokens_per_second": 7404.603
+    },
+    {
+      "epoch": 1.7484926787252368,
+      "grad_norm": 0.7690853275407137,
+      "learning_rate": 3.877534756080353e-07,
+      "loss": 0.30381178855895996,
+      "num_input_tokens_seen": 206761200,
+      "step": 2030,
+      "train_runtime": 27925.9518,
+      "train_tokens_per_second": 7403.909
+    },
+    {
+      "epoch": 1.7527993109388458,
+      "grad_norm": 0.588121336370943,
+      "learning_rate": 3.747989662567403e-07,
+      "loss": 0.30392751693725584,
+      "num_input_tokens_seen": 207254168,
+      "step": 2035,
+      "train_runtime": 27994.4378,
+      "train_tokens_per_second": 7403.405
+    },
+    {
+      "epoch": 1.757105943152455,
+      "grad_norm": 0.6086100201180733,
+      "learning_rate": 3.620561195807454e-07,
+      "loss": 0.2981263637542725,
+      "num_input_tokens_seen": 207724624,
+      "step": 2040,
+      "train_runtime": 28064.886,
+      "train_tokens_per_second": 7401.584
+    },
+    {
+      "epoch": 1.7614125753660637,
+      "grad_norm": 0.5837314444107731,
+      "learning_rate": 3.4952551872984295e-07,
+      "loss": 0.2957895278930664,
+      "num_input_tokens_seen": 208230432,
+      "step": 2045,
+      "train_runtime": 28140.0297,
+      "train_tokens_per_second": 7399.794
+    },
+    {
+      "epoch": 1.7657192075796728,
+      "grad_norm": 0.7083602081207774,
+      "learning_rate": 3.372077371408361e-07,
+      "loss": 0.31419479846954346,
+      "num_input_tokens_seen": 208756176,
+      "step": 2050,
+      "train_runtime": 28203.3818,
+      "train_tokens_per_second": 7401.814
+    },
+    {
+      "epoch": 1.7700258397932815,
+      "grad_norm": 0.6007868998583191,
+      "learning_rate": 3.2510333851129895e-07,
+      "loss": 0.29912557601928713,
+      "num_input_tokens_seen": 209283048,
+      "step": 2055,
+      "train_runtime": 28276.6448,
+      "train_tokens_per_second": 7401.269
+    },
+    {
+      "epoch": 1.7743324720068907,
+      "grad_norm": 0.6111214072804273,
+      "learning_rate": 3.132128767737824e-07,
+      "loss": 0.2877781391143799,
+      "num_input_tokens_seen": 209776608,
+      "step": 2060,
+      "train_runtime": 28339.5361,
+      "train_tokens_per_second": 7402.26
+    },
+    {
+      "epoch": 1.7786391042204994,
+      "grad_norm": 0.6104558696682314,
+      "learning_rate": 3.015368960704584e-07,
+      "loss": 0.3083071708679199,
+      "num_input_tokens_seen": 210290720,
+      "step": 2065,
+      "train_runtime": 28413.252,
+      "train_tokens_per_second": 7401.149
+    },
+    {
+      "epoch": 1.7829457364341086,
+      "grad_norm": 0.5836728591190887,
+      "learning_rate": 2.9007593072822246e-07,
+      "loss": 0.30655722618103026,
+      "num_input_tokens_seen": 210795064,
+      "step": 2070,
+      "train_runtime": 28480.1482,
+      "train_tokens_per_second": 7401.474
+    },
+    {
+      "epoch": 1.7872523686477175,
+      "grad_norm": 0.5843802582004601,
+      "learning_rate": 2.7883050523424214e-07,
+      "loss": 0.2855481386184692,
+      "num_input_tokens_seen": 211307816,
+      "step": 2075,
+      "train_runtime": 28550.624,
+      "train_tokens_per_second": 7401.163
+    },
+    {
+      "epoch": 1.7915590008613265,
+      "grad_norm": 0.5970608151039677,
+      "learning_rate": 2.67801134211953e-07,
+      "loss": 0.3143181800842285,
+      "num_input_tokens_seen": 211819656,
+      "step": 2080,
+      "train_runtime": 28622.1835,
+      "train_tokens_per_second": 7400.541
+    },
+    {
+      "epoch": 1.7958656330749354,
+      "grad_norm": 0.631118629714427,
+      "learning_rate": 2.569883223975078e-07,
+      "loss": 0.3136578559875488,
+      "num_input_tokens_seen": 212342848,
+      "step": 2085,
+      "train_runtime": 28690.2587,
+      "train_tokens_per_second": 7401.218
+    },
+    {
+      "epoch": 1.8001722652885443,
+      "grad_norm": 0.6050935590065106,
+      "learning_rate": 2.463925646166793e-07,
+      "loss": 0.3073784351348877,
+      "num_input_tokens_seen": 212865944,
+      "step": 2090,
+      "train_runtime": 28764.8887,
+      "train_tokens_per_second": 7400.201
+    },
+    {
+      "epoch": 1.8044788975021533,
+      "grad_norm": 0.7462730804752461,
+      "learning_rate": 2.3601434576221548e-07,
+      "loss": 0.2910205364227295,
+      "num_input_tokens_seen": 213312424,
+      "step": 2095,
+      "train_runtime": 28826.7975,
+      "train_tokens_per_second": 7399.796
+    },
+    {
+      "epoch": 1.8087855297157622,
+      "grad_norm": 0.6890875242639481,
+      "learning_rate": 2.2585414077164914e-07,
+      "loss": 0.3124124050140381,
+      "num_input_tokens_seen": 213861568,
+      "step": 2100,
+      "train_runtime": 28886.3363,
+      "train_tokens_per_second": 7403.555
+    },
+    {
+      "epoch": 1.8130921619293714,
+      "grad_norm": 0.6451854669090437,
+      "learning_rate": 2.1591241460556355e-07,
+      "loss": 0.3035388469696045,
+      "num_input_tokens_seen": 214400864,
+      "step": 2105,
+      "train_runtime": 28962.1347,
+      "train_tokens_per_second": 7402.799
+    },
+    {
+      "epoch": 1.8173987941429801,
+      "grad_norm": 0.5932564357300775,
+      "learning_rate": 2.0618962222631434e-07,
+      "loss": 0.30562429428100585,
+      "num_input_tokens_seen": 214933176,
+      "step": 2110,
+      "train_runtime": 29019.6315,
+      "train_tokens_per_second": 7406.475
+    },
+    {
+      "epoch": 1.8217054263565893,
+      "grad_norm": 0.7273993480230436,
+      "learning_rate": 1.9668620857720865e-07,
+      "loss": 0.3033386945724487,
+      "num_input_tokens_seen": 215387240,
+      "step": 2115,
+      "train_runtime": 29081.5024,
+      "train_tokens_per_second": 7406.331
+    },
+    {
+      "epoch": 1.826012058570198,
+      "grad_norm": 0.5794672931344478,
+      "learning_rate": 1.8740260856214388e-07,
+      "loss": 0.30868115425109866,
+      "num_input_tokens_seen": 215936824,
+      "step": 2120,
+      "train_runtime": 29150.3054,
+      "train_tokens_per_second": 7407.704
+    },
+    {
+      "epoch": 1.8303186907838072,
+      "grad_norm": 0.644533623475115,
+      "learning_rate": 1.7833924702570725e-07,
+      "loss": 0.30875744819641116,
+      "num_input_tokens_seen": 216480984,
+      "step": 2125,
+      "train_runtime": 29226.4831,
+      "train_tokens_per_second": 7407.015
+    },
+    {
+      "epoch": 1.8346253229974159,
+      "grad_norm": 0.6156843271663307,
+      "learning_rate": 1.6949653873372828e-07,
+      "loss": 0.2976231098175049,
+      "num_input_tokens_seen": 216985056,
+      "step": 2130,
+      "train_runtime": 29303.3245,
+      "train_tokens_per_second": 7404.793
+    },
+    {
+      "epoch": 1.838931955211025,
+      "grad_norm": 0.5929671577727561,
+      "learning_rate": 1.6087488835430208e-07,
+      "loss": 0.3137772560119629,
+      "num_input_tokens_seen": 217567072,
+      "step": 2135,
+      "train_runtime": 29380.4773,
+      "train_tokens_per_second": 7405.158
+    },
+    {
+      "epoch": 1.843238587424634,
+      "grad_norm": 0.6832951069959544,
+      "learning_rate": 1.5247469043927153e-07,
+      "loss": 0.304307746887207,
+      "num_input_tokens_seen": 218066976,
+      "step": 2140,
+      "train_runtime": 29445.7965,
+      "train_tokens_per_second": 7405.708
+    },
+    {
+      "epoch": 1.847545219638243,
+      "grad_norm": 0.6130447989309619,
+      "learning_rate": 1.4429632940616721e-07,
+      "loss": 0.31688547134399414,
+      "num_input_tokens_seen": 218547776,
+      "step": 2145,
+      "train_runtime": 29519.0458,
+      "train_tokens_per_second": 7403.619
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 0.5920284879807376,
+      "learning_rate": 1.3634017952061872e-07,
+      "loss": 0.2871793270111084,
+      "num_input_tokens_seen": 219097408,
+      "step": 2150,
+      "train_runtime": 29593.4565,
+      "train_tokens_per_second": 7403.576
+    },
+    {
+      "epoch": 1.8561584840654608,
+      "grad_norm": 0.5989232424009299,
+      "learning_rate": 1.2860660487922616e-07,
+      "loss": 0.3016232967376709,
+      "num_input_tokens_seen": 219617128,
+      "step": 2155,
+      "train_runtime": 29659.0858,
+      "train_tokens_per_second": 7404.717
+    },
+    {
+      "epoch": 1.8604651162790697,
+      "grad_norm": 0.6274792319418983,
+      "learning_rate": 1.2109595939289676e-07,
+      "loss": 0.3069549560546875,
+      "num_input_tokens_seen": 220118992,
+      "step": 2160,
+      "train_runtime": 29729.4276,
+      "train_tokens_per_second": 7404.078
+    },
+    {
+      "epoch": 1.8647717484926787,
+      "grad_norm": 0.639624885028358,
+      "learning_rate": 1.1380858677065177e-07,
+      "loss": 0.29118402004241944,
+      "num_input_tokens_seen": 220620384,
+      "step": 2165,
+      "train_runtime": 29800.9726,
+      "train_tokens_per_second": 7403.127
+    },
+    {
+      "epoch": 1.8690783807062878,
+      "grad_norm": 0.6263759770644098,
+      "learning_rate": 1.0674482050389457e-07,
+      "loss": 0.3211177349090576,
+      "num_input_tokens_seen": 221104624,
+      "step": 2170,
+      "train_runtime": 29867.1222,
+      "train_tokens_per_second": 7402.944
+    },
+    {
+      "epoch": 1.8733850129198966,
+      "grad_norm": 0.5850849374408646,
+      "learning_rate": 9.990498385115066e-08,
+      "loss": 0.2935049533843994,
+      "num_input_tokens_seen": 221664440,
+      "step": 2175,
+      "train_runtime": 29933.121,
+      "train_tokens_per_second": 7405.323
+    },
+    {
+      "epoch": 1.8776916451335057,
+      "grad_norm": 0.6187947000777021,
+      "learning_rate": 9.328938982327396e-08,
+      "loss": 0.2967803478240967,
+      "num_input_tokens_seen": 222166280,
+      "step": 2180,
+      "train_runtime": 29994.8442,
+      "train_tokens_per_second": 7406.816
+    },
+    {
+      "epoch": 1.8819982773471144,
+      "grad_norm": 0.6117271333511343,
+      "learning_rate": 8.68983411691221e-08,
+      "loss": 0.3003566265106201,
+      "num_input_tokens_seen": 222663120,
+      "step": 2185,
+      "train_runtime": 30054.1743,
+      "train_tokens_per_second": 7408.725
+    },
+    {
+      "epoch": 1.8863049095607236,
+      "grad_norm": 0.6274877192256639,
+      "learning_rate": 8.073213036170258e-08,
+      "loss": 0.30220422744750974,
+      "num_input_tokens_seen": 223174608,
+      "step": 2190,
+      "train_runtime": 30117.6604,
+      "train_tokens_per_second": 7410.091
+    },
+    {
+      "epoch": 1.8906115417743323,
+      "grad_norm": 0.5663763479699041,
+      "learning_rate": 7.479103958478783e-08,
+      "loss": 0.3004784107208252,
+      "num_input_tokens_seen": 223724416,
+      "step": 2195,
+      "train_runtime": 30178.196,
+      "train_tokens_per_second": 7413.446
+    },
+    {
+      "epoch": 1.8949181739879415,
+      "grad_norm": 0.6615913855044907,
+      "learning_rate": 6.907534072000177e-08,
+      "loss": 0.30257329940795896,
+      "num_input_tokens_seen": 224257552,
+      "step": 2200,
+      "train_runtime": 30257.9783,
+      "train_tokens_per_second": 7411.518
+    },
+    {
+      "epoch": 1.8992248062015504,
+      "grad_norm": 0.554939875230265,
+      "learning_rate": 6.358529533437796e-08,
+      "loss": 0.2995903491973877,
+      "num_input_tokens_seen": 224783568,
+      "step": 2205,
+      "train_runtime": 30331.9171,
+      "train_tokens_per_second": 7410.793
+    },
+    {
+      "epoch": 1.9035314384151594,
+      "grad_norm": 0.6083628550269634,
+      "learning_rate": 5.832115466838872e-08,
+      "loss": 0.29779205322265623,
+      "num_input_tokens_seen": 225281672,
+      "step": 2210,
+      "train_runtime": 30407.547,
+      "train_tokens_per_second": 7408.742
+    },
+    {
+      "epoch": 1.9078380706287683,
+      "grad_norm": 0.6279252265728321,
+      "learning_rate": 5.3283159624448745e-08,
+      "loss": 0.3170241117477417,
+      "num_input_tokens_seen": 225814656,
+      "step": 2215,
+      "train_runtime": 30466.2688,
+      "train_tokens_per_second": 7411.956
+    },
+    {
+      "epoch": 1.9121447028423773,
+      "grad_norm": 0.7185275451441703,
+      "learning_rate": 4.847154075589e-08,
+      "loss": 0.3137502670288086,
+      "num_input_tokens_seen": 226330440,
+      "step": 2220,
+      "train_runtime": 30533.9136,
+      "train_tokens_per_second": 7412.428
+    },
+    {
+      "epoch": 1.9164513350559862,
+      "grad_norm": 0.5936044355682394,
+      "learning_rate": 4.3886518256411325e-08,
+      "loss": 0.3071382999420166,
+      "num_input_tokens_seen": 226837032,
+      "step": 2225,
+      "train_runtime": 30605.5094,
+      "train_tokens_per_second": 7411.64
+    },
+    {
+      "epoch": 1.9207579672695951,
+      "grad_norm": 0.6318773911159581,
+      "learning_rate": 3.9528301950000345e-08,
+      "loss": 0.30357215404510496,
+      "num_input_tokens_seen": 227345616,
+      "step": 2230,
+      "train_runtime": 30681.0349,
+      "train_tokens_per_second": 7409.972
+    },
+    {
+      "epoch": 1.9250645994832043,
+      "grad_norm": 0.5951318413675115,
+      "learning_rate": 3.539709128133395e-08,
+      "loss": 0.3153378963470459,
+      "num_input_tokens_seen": 227840184,
+      "step": 2235,
+      "train_runtime": 30751.0125,
+      "train_tokens_per_second": 7409.193
+    },
+    {
+      "epoch": 1.929371231696813,
+      "grad_norm": 0.6707938307520876,
+      "learning_rate": 3.149307530664891e-08,
+      "loss": 0.30716452598571775,
+      "num_input_tokens_seen": 228355216,
+      "step": 2240,
+      "train_runtime": 30817.8728,
+      "train_tokens_per_second": 7409.831
+    },
+    {
+      "epoch": 1.9336778639104222,
+      "grad_norm": 0.6127756791716347,
+      "learning_rate": 2.7816432685091598e-08,
+      "loss": 0.2936999320983887,
+      "num_input_tokens_seen": 228882600,
+      "step": 2245,
+      "train_runtime": 30882.7718,
+      "train_tokens_per_second": 7411.336
+    },
+    {
+      "epoch": 1.937984496124031,
+      "grad_norm": 0.551559122734471,
+      "learning_rate": 2.4367331670539506e-08,
+      "loss": 0.3067632675170898,
+      "num_input_tokens_seen": 229400664,
+      "step": 2250,
+      "train_runtime": 30954.957,
+      "train_tokens_per_second": 7410.789
+    },
+    {
+      "epoch": 1.94229112833764,
+      "grad_norm": 0.6350936400123717,
+      "learning_rate": 2.1145930103904645e-08,
+      "loss": 0.3079858303070068,
+      "num_input_tokens_seen": 229912008,
+      "step": 2255,
+      "train_runtime": 31022.3371,
+      "train_tokens_per_second": 7411.176
+    },
+    {
+      "epoch": 1.9465977605512488,
+      "grad_norm": 0.6737908112781935,
+      "learning_rate": 1.8152375405909305e-08,
+      "loss": 0.3009981632232666,
+      "num_input_tokens_seen": 230446552,
+      "step": 2260,
+      "train_runtime": 31098.5518,
+      "train_tokens_per_second": 7410.202
+    },
+    {
+      "epoch": 1.950904392764858,
+      "grad_norm": 0.7144758716926696,
+      "learning_rate": 1.538680457033814e-08,
+      "loss": 0.3050046920776367,
+      "num_input_tokens_seen": 230974000,
+      "step": 2265,
+      "train_runtime": 31173.146,
+      "train_tokens_per_second": 7409.39
+    },
+    {
+      "epoch": 1.9552110249784669,
+      "grad_norm": 0.6702413804814763,
+      "learning_rate": 1.284934415776984e-08,
+      "loss": 0.3118910789489746,
+      "num_input_tokens_seen": 231510712,
+      "step": 2270,
+      "train_runtime": 31249.4408,
+      "train_tokens_per_second": 7408.475
+    },
+    {
+      "epoch": 1.9595176571920758,
+      "grad_norm": 0.6610446821494167,
+      "learning_rate": 1.0540110289786742e-08,
+      "loss": 0.28809175491333006,
+      "num_input_tokens_seen": 231960472,
+      "step": 2275,
+      "train_runtime": 31314.9022,
+      "train_tokens_per_second": 7407.351
+    },
+    {
+      "epoch": 1.9638242894056848,
+      "grad_norm": 0.6292947995618373,
+      "learning_rate": 8.459208643659122e-09,
+      "loss": 0.3013895511627197,
+      "num_input_tokens_seen": 232466016,
+      "step": 2280,
+      "train_runtime": 31390.756,
+      "train_tokens_per_second": 7405.556
+    },
+    {
+      "epoch": 1.9681309216192937,
+      "grad_norm": 0.6388156769135576,
+      "learning_rate": 6.606734447507923e-09,
+      "loss": 0.2956396579742432,
+      "num_input_tokens_seen": 233001544,
+      "step": 2285,
+      "train_runtime": 31455.4291,
+      "train_tokens_per_second": 7407.355
+    },
+    {
+      "epoch": 1.9724375538329026,
+      "grad_norm": 0.7338122921014575,
+      "learning_rate": 4.982772475951026e-09,
+      "loss": 0.3004451751708984,
+      "num_input_tokens_seen": 233486880,
+      "step": 2290,
+      "train_runtime": 31524.8943,
+      "train_tokens_per_second": 7406.429
+    },
+    {
+      "epoch": 1.9767441860465116,
+      "grad_norm": 0.6784313338377094,
+      "learning_rate": 3.587397046219132e-09,
+      "loss": 0.3195258617401123,
+      "num_input_tokens_seen": 233992944,
+      "step": 2295,
+      "train_runtime": 31592.025,
+      "train_tokens_per_second": 7406.709
+    },
+    {
+      "epoch": 1.9810508182601207,
+      "grad_norm": 0.5996807721646643,
+      "learning_rate": 2.4206720147573748e-09,
+      "loss": 0.3012960433959961,
+      "num_input_tokens_seen": 234505824,
+      "step": 2300,
+      "train_runtime": 31662.022,
+      "train_tokens_per_second": 7406.533
+    },
+    {
+      "epoch": 1.9853574504737295,
+      "grad_norm": 0.612165146123787,
+      "learning_rate": 1.4826507743032071e-09,
+      "loss": 0.2931736707687378,
+      "num_input_tokens_seen": 234988896,
+      "step": 2305,
+      "train_runtime": 31727.6848,
+      "train_tokens_per_second": 7406.431
+    },
+    {
+      "epoch": 1.9896640826873386,
+      "grad_norm": 0.5687329803521689,
+      "learning_rate": 7.733762514411381e-10,
+      "loss": 0.3052358865737915,
+      "num_input_tokens_seen": 235490216,
+      "step": 2310,
+      "train_runtime": 31802.994,
+      "train_tokens_per_second": 7404.656
+    },
+    {
+      "epoch": 1.9939707149009473,
+      "grad_norm": 0.6676211550092361,
+      "learning_rate": 2.928809046398584e-10,
+      "loss": 0.29867162704467776,
+      "num_input_tokens_seen": 236010080,
+      "step": 2315,
+      "train_runtime": 31879.2306,
+      "train_tokens_per_second": 7403.255
+    },
+    {
+      "epoch": 1.9982773471145565,
+      "grad_norm": 0.5980690602806559,
+      "learning_rate": 4.118672276620661e-11,
+      "loss": 0.3062272548675537,
+      "num_input_tokens_seen": 236545192,
+      "step": 2320,
+      "train_runtime": 31958.4063,
+      "train_tokens_per_second": 7401.658
+    },
+    {
+      "epoch": 2.0,
+      "num_input_tokens_seen": 236749552,
+      "step": 2322,
+      "total_flos": 437831363461120.0,
+      "train_loss": 0.3453142216989031,
+      "train_runtime": 32003.0179,
+      "train_samples_per_second": 0.58,
+      "train_steps_per_second": 0.073
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 2322,
+  "num_input_tokens_seen": 236749552,
+  "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": 437831363461120.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}