diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17125 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 14.990723562152134,
+  "eval_steps": 500,
+  "global_step": 12120,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0012368583797155227,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.6501650165016504e-07,
+      "loss": 2.8095,
+      "step": 1
+    },
+    {
+      "epoch": 0.006184291898577613,
+      "grad_norm": 3.1875,
+      "learning_rate": 8.25082508250825e-07,
+      "loss": 2.8204,
+      "step": 5
+    },
+    {
+      "epoch": 0.012368583797155226,
+      "grad_norm": 1.9140625,
+      "learning_rate": 1.65016501650165e-06,
+      "loss": 2.7934,
+      "step": 10
+    },
+    {
+      "epoch": 0.01855287569573284,
+      "grad_norm": 2.171875,
+      "learning_rate": 2.4752475247524753e-06,
+      "loss": 2.7932,
+      "step": 15
+    },
+    {
+      "epoch": 0.024737167594310452,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3.3003300330033e-06,
+      "loss": 2.842,
+      "step": 20
+    },
+    {
+      "epoch": 0.030921459492888066,
+      "grad_norm": 1.96875,
+      "learning_rate": 4.125412541254126e-06,
+      "loss": 2.8041,
+      "step": 25
+    },
+    {
+      "epoch": 0.03710575139146568,
+      "grad_norm": 27.5,
+      "learning_rate": 4.950495049504951e-06,
+      "loss": 2.7951,
+      "step": 30
+    },
+    {
+      "epoch": 0.04329004329004329,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.775577557755775e-06,
+      "loss": 2.8127,
+      "step": 35
+    },
+    {
+      "epoch": 0.049474335188620905,
+      "grad_norm": 1.8671875,
+      "learning_rate": 6.6006600660066e-06,
+      "loss": 2.8001,
+      "step": 40
+    },
+    {
+      "epoch": 0.055658627087198514,
+      "grad_norm": 1.625,
+      "learning_rate": 7.4257425742574256e-06,
+      "loss": 2.76,
+      "step": 45
+    },
+    {
+      "epoch": 0.06184291898577613,
+      "grad_norm": 1.5546875,
+      "learning_rate": 8.250825082508252e-06,
+      "loss": 2.7348,
+      "step": 50
+    },
+    {
+      "epoch": 0.06802721088435375,
+      "grad_norm": 2.765625,
+      "learning_rate": 9.075907590759077e-06,
+      "loss": 2.7171,
+      "step": 55
+    },
+    {
+      "epoch": 0.07421150278293136,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.900990099009901e-06,
+      "loss": 2.6948,
+      "step": 60
+    },
+    {
+      "epoch": 0.08039579468150897,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0726072607260726e-05,
+      "loss": 2.6657,
+      "step": 65
+    },
+    {
+      "epoch": 0.08658008658008658,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.155115511551155e-05,
+      "loss": 2.6411,
+      "step": 70
+    },
+    {
+      "epoch": 0.09276437847866419,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.2376237623762377e-05,
+      "loss": 2.5744,
+      "step": 75
+    },
+    {
+      "epoch": 0.09894867037724181,
+      "grad_norm": 3.0,
+      "learning_rate": 1.32013201320132e-05,
+      "loss": 2.5252,
+      "step": 80
+    },
+    {
+      "epoch": 0.10513296227581942,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.4026402640264028e-05,
+      "loss": 2.4702,
+      "step": 85
+    },
+    {
+      "epoch": 0.11131725417439703,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.4851485148514851e-05,
+      "loss": 2.3871,
+      "step": 90
+    },
+    {
+      "epoch": 0.11750154607297464,
+      "grad_norm": 2.84375,
+      "learning_rate": 1.567656765676568e-05,
+      "loss": 2.3574,
+      "step": 95
+    },
+    {
+      "epoch": 0.12368583797155226,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.6501650165016504e-05,
+      "loss": 2.3075,
+      "step": 100
+    },
+    {
+      "epoch": 0.12987012987012986,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.7326732673267325e-05,
+      "loss": 2.2358,
+      "step": 105
+    },
+    {
+      "epoch": 0.1360544217687075,
+      "grad_norm": 2.25,
+      "learning_rate": 1.8151815181518153e-05,
+      "loss": 2.1817,
+      "step": 110
+    },
+    {
+      "epoch": 0.1422387136672851,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.8976897689768978e-05,
+      "loss": 2.1457,
+      "step": 115
+    },
+    {
+      "epoch": 0.14842300556586271,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.9801980198019803e-05,
+      "loss": 2.1015,
+      "step": 120
+    },
+    {
+      "epoch": 0.15460729746444032,
+      "grad_norm": 5.375,
+      "learning_rate": 2.0627062706270627e-05,
+      "loss": 2.0488,
+      "step": 125
+    },
+    {
+      "epoch": 0.16079158936301793,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.1452145214521452e-05,
+      "loss": 1.9967,
+      "step": 130
+    },
+    {
+      "epoch": 0.16697588126159554,
+      "grad_norm": 1.8359375,
+      "learning_rate": 2.227722772277228e-05,
+      "loss": 1.9499,
+      "step": 135
+    },
+    {
+      "epoch": 0.17316017316017315,
+      "grad_norm": 5.3125,
+      "learning_rate": 2.31023102310231e-05,
+      "loss": 1.9143,
+      "step": 140
+    },
+    {
+      "epoch": 0.17934446505875076,
+      "grad_norm": 5.625,
+      "learning_rate": 2.392739273927393e-05,
+      "loss": 1.8633,
+      "step": 145
+    },
+    {
+      "epoch": 0.18552875695732837,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.4752475247524754e-05,
+      "loss": 1.8392,
+      "step": 150
+    },
+    {
+      "epoch": 0.191713048855906,
+      "grad_norm": 1.859375,
+      "learning_rate": 2.557755775577558e-05,
+      "loss": 1.7943,
+      "step": 155
+    },
+    {
+      "epoch": 0.19789734075448362,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.64026402640264e-05,
+      "loss": 1.7733,
+      "step": 160
+    },
+    {
+      "epoch": 0.20408163265306123,
+      "grad_norm": 1.5,
+      "learning_rate": 2.722772277227723e-05,
+      "loss": 1.7403,
+      "step": 165
+    },
+    {
+      "epoch": 0.21026592455163884,
+      "grad_norm": 1.53125,
+      "learning_rate": 2.8052805280528056e-05,
+      "loss": 1.6989,
+      "step": 170
+    },
+    {
+      "epoch": 0.21645021645021645,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.8877887788778878e-05,
+      "loss": 1.6679,
+      "step": 175
+    },
+    {
+      "epoch": 0.22263450834879406,
+      "grad_norm": 1.59375,
+      "learning_rate": 2.9702970297029702e-05,
+      "loss": 1.629,
+      "step": 180
+    },
+    {
+      "epoch": 0.22881880024737167,
+      "grad_norm": 0.8671875,
+      "learning_rate": 3.052805280528053e-05,
+      "loss": 1.6264,
+      "step": 185
+    },
+    {
+      "epoch": 0.23500309214594928,
+      "grad_norm": 1.2109375,
+      "learning_rate": 3.135313531353136e-05,
+      "loss": 1.581,
+      "step": 190
+    },
+    {
+      "epoch": 0.24118738404452691,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.217821782178218e-05,
+      "loss": 1.5598,
+      "step": 195
+    },
+    {
+      "epoch": 0.24737167594310452,
+      "grad_norm": 0.8828125,
+      "learning_rate": 3.300330033003301e-05,
+      "loss": 1.5259,
+      "step": 200
+    },
+    {
+      "epoch": 0.2535559678416821,
+      "grad_norm": 0.38671875,
+      "learning_rate": 3.382838283828383e-05,
+      "loss": 1.4943,
+      "step": 205
+    },
+    {
+      "epoch": 0.2597402597402597,
+      "grad_norm": 0.59375,
+      "learning_rate": 3.465346534653465e-05,
+      "loss": 1.4984,
+      "step": 210
+    },
+    {
+      "epoch": 0.2659245516388373,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.5478547854785485e-05,
+      "loss": 1.4686,
+      "step": 215
+    },
+    {
+      "epoch": 0.272108843537415,
+      "grad_norm": 0.6484375,
+      "learning_rate": 3.6303630363036307e-05,
+      "loss": 1.4699,
+      "step": 220
+    },
+    {
+      "epoch": 0.2782931354359926,
+      "grad_norm": 1.1484375,
+      "learning_rate": 3.712871287128713e-05,
+      "loss": 1.4291,
+      "step": 225
+    },
+    {
+      "epoch": 0.2844774273345702,
+      "grad_norm": 0.4609375,
+      "learning_rate": 3.7953795379537956e-05,
+      "loss": 1.4235,
+      "step": 230
+    },
+    {
+      "epoch": 0.2906617192331478,
+      "grad_norm": 0.48828125,
+      "learning_rate": 3.877887788778878e-05,
+      "loss": 1.4041,
+      "step": 235
+    },
+    {
+      "epoch": 0.29684601113172543,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.9603960396039605e-05,
+      "loss": 1.3848,
+      "step": 240
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 0.90234375,
+      "learning_rate": 4.042904290429043e-05,
+      "loss": 1.3611,
+      "step": 245
+    },
+    {
+      "epoch": 0.30921459492888065,
+      "grad_norm": 0.314453125,
+      "learning_rate": 4.1254125412541255e-05,
+      "loss": 1.3558,
+      "step": 250
+    },
+    {
+      "epoch": 0.31539888682745826,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.207920792079208e-05,
+      "loss": 1.3451,
+      "step": 255
+    },
+    {
+      "epoch": 0.32158317872603587,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.2904290429042904e-05,
+      "loss": 1.3326,
+      "step": 260
+    },
+    {
+      "epoch": 0.3277674706246135,
+      "grad_norm": 0.62890625,
+      "learning_rate": 4.372937293729373e-05,
+      "loss": 1.3084,
+      "step": 265
+    },
+    {
+      "epoch": 0.3339517625231911,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.455445544554456e-05,
+      "loss": 1.3091,
+      "step": 270
+    },
+    {
+      "epoch": 0.3401360544217687,
+      "grad_norm": 0.33984375,
+      "learning_rate": 4.537953795379538e-05,
+      "loss": 1.3078,
+      "step": 275
+    },
+    {
+      "epoch": 0.3463203463203463,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.62046204620462e-05,
+      "loss": 1.2944,
+      "step": 280
+    },
+    {
+      "epoch": 0.3525046382189239,
+      "grad_norm": 0.447265625,
+      "learning_rate": 4.702970297029703e-05,
+      "loss": 1.2689,
+      "step": 285
+    },
+    {
+      "epoch": 0.3586889301175015,
+      "grad_norm": 0.52734375,
+      "learning_rate": 4.785478547854786e-05,
+      "loss": 1.2907,
+      "step": 290
+    },
+    {
+      "epoch": 0.36487322201607914,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.867986798679868e-05,
+      "loss": 1.2734,
+      "step": 295
+    },
+    {
+      "epoch": 0.37105751391465674,
+      "grad_norm": 0.408203125,
+      "learning_rate": 4.950495049504951e-05,
+      "loss": 1.2533,
+      "step": 300
+    },
+    {
+      "epoch": 0.3772418058132344,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.0330033003300336e-05,
+      "loss": 1.2568,
+      "step": 305
+    },
+    {
+      "epoch": 0.383426097711812,
+      "grad_norm": 0.48828125,
+      "learning_rate": 5.115511551155116e-05,
+      "loss": 1.2476,
+      "step": 310
+    },
+    {
+      "epoch": 0.38961038961038963,
+      "grad_norm": 0.5078125,
+      "learning_rate": 5.1980198019801986e-05,
+      "loss": 1.2506,
+      "step": 315
+    },
+    {
+      "epoch": 0.39579468150896724,
+      "grad_norm": 0.337890625,
+      "learning_rate": 5.28052805280528e-05,
+      "loss": 1.2311,
+      "step": 320
+    },
+    {
+      "epoch": 0.40197897340754485,
+      "grad_norm": 0.357421875,
+      "learning_rate": 5.3630363036303635e-05,
+      "loss": 1.2302,
+      "step": 325
+    },
+    {
+      "epoch": 0.40816326530612246,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.445544554455446e-05,
+      "loss": 1.2182,
+      "step": 330
+    },
+    {
+      "epoch": 0.41434755720470007,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.528052805280528e-05,
+      "loss": 1.2322,
+      "step": 335
+    },
+    {
+      "epoch": 0.4205318491032777,
+      "grad_norm": 0.466796875,
+      "learning_rate": 5.610561056105611e-05,
+      "loss": 1.2181,
+      "step": 340
+    },
+    {
+      "epoch": 0.4267161410018553,
+      "grad_norm": 0.55078125,
+      "learning_rate": 5.693069306930693e-05,
+      "loss": 1.2136,
+      "step": 345
+    },
+    {
+      "epoch": 0.4329004329004329,
+      "grad_norm": 0.546875,
+      "learning_rate": 5.7755775577557755e-05,
+      "loss": 1.2089,
+      "step": 350
+    },
+    {
+      "epoch": 0.4390847247990105,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.858085808580859e-05,
+      "loss": 1.197,
+      "step": 355
+    },
+    {
+      "epoch": 0.4452690166975881,
+      "grad_norm": 0.55859375,
+      "learning_rate": 5.9405940594059404e-05,
+      "loss": 1.1929,
+      "step": 360
+    },
+    {
+      "epoch": 0.4514533085961657,
+      "grad_norm": 0.55859375,
+      "learning_rate": 6.023102310231023e-05,
+      "loss": 1.1859,
+      "step": 365
+    },
+    {
+      "epoch": 0.45763760049474334,
+      "grad_norm": 0.443359375,
+      "learning_rate": 6.105610561056106e-05,
+      "loss": 1.1969,
+      "step": 370
+    },
+    {
+      "epoch": 0.46382189239332094,
+      "grad_norm": 0.46484375,
+      "learning_rate": 6.188118811881188e-05,
+      "loss": 1.1898,
+      "step": 375
+    },
+    {
+      "epoch": 0.47000618429189855,
+      "grad_norm": 0.3828125,
+      "learning_rate": 6.270627062706272e-05,
+      "loss": 1.1736,
+      "step": 380
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.353135313531354e-05,
+      "loss": 1.1827,
+      "step": 385
+    },
+    {
+      "epoch": 0.48237476808905383,
+      "grad_norm": 0.4921875,
+      "learning_rate": 6.435643564356436e-05,
+      "loss": 1.1792,
+      "step": 390
+    },
+    {
+      "epoch": 0.48855905998763144,
+      "grad_norm": 0.375,
+      "learning_rate": 6.518151815181518e-05,
+      "loss": 1.1674,
+      "step": 395
+    },
+    {
+      "epoch": 0.49474335188620905,
+      "grad_norm": 0.349609375,
+      "learning_rate": 6.600660066006602e-05,
+      "loss": 1.1585,
+      "step": 400
+    },
+    {
+      "epoch": 0.5009276437847866,
+      "grad_norm": 0.333984375,
+      "learning_rate": 6.683168316831684e-05,
+      "loss": 1.1601,
+      "step": 405
+    },
+    {
+      "epoch": 0.5071119356833642,
+      "grad_norm": 0.400390625,
+      "learning_rate": 6.765676567656766e-05,
+      "loss": 1.1643,
+      "step": 410
+    },
+    {
+      "epoch": 0.5132962275819418,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.848184818481849e-05,
+      "loss": 1.1668,
+      "step": 415
+    },
+    {
+      "epoch": 0.5194805194805194,
+      "grad_norm": 0.4609375,
+      "learning_rate": 6.93069306930693e-05,
+      "loss": 1.1485,
+      "step": 420
+    },
+    {
+      "epoch": 0.525664811379097,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.013201320132014e-05,
+      "loss": 1.1561,
+      "step": 425
+    },
+    {
+      "epoch": 0.5318491032776747,
+      "grad_norm": 0.40625,
+      "learning_rate": 7.095709570957097e-05,
+      "loss": 1.1572,
+      "step": 430
+    },
+    {
+      "epoch": 0.5380333951762524,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.178217821782178e-05,
+      "loss": 1.1527,
+      "step": 435
+    },
+    {
+      "epoch": 0.54421768707483,
+      "grad_norm": 0.419921875,
+      "learning_rate": 7.260726072607261e-05,
+      "loss": 1.1559,
+      "step": 440
+    },
+    {
+      "epoch": 0.5504019789734076,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.343234323432343e-05,
+      "loss": 1.1587,
+      "step": 445
+    },
+    {
+      "epoch": 0.5565862708719852,
+      "grad_norm": 0.46484375,
+      "learning_rate": 7.425742574257426e-05,
+      "loss": 1.1394,
+      "step": 450
+    },
+    {
+      "epoch": 0.5627705627705628,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.508250825082509e-05,
+      "loss": 1.1451,
+      "step": 455
+    },
+    {
+      "epoch": 0.5689548546691404,
+      "grad_norm": 0.40625,
+      "learning_rate": 7.590759075907591e-05,
+      "loss": 1.1405,
+      "step": 460
+    },
+    {
+      "epoch": 0.575139146567718,
+      "grad_norm": 0.55859375,
+      "learning_rate": 7.673267326732673e-05,
+      "loss": 1.1397,
+      "step": 465
+    },
+    {
+      "epoch": 0.5813234384662956,
+      "grad_norm": 0.49609375,
+      "learning_rate": 7.755775577557755e-05,
+      "loss": 1.1213,
+      "step": 470
+    },
+    {
+      "epoch": 0.5875077303648732,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.838283828382839e-05,
+      "loss": 1.1313,
+      "step": 475
+    },
+    {
+      "epoch": 0.5936920222634509,
+      "grad_norm": 0.52734375,
+      "learning_rate": 7.920792079207921e-05,
+      "loss": 1.1268,
+      "step": 480
+    },
+    {
+      "epoch": 0.5998763141620285,
+      "grad_norm": 0.39453125,
+      "learning_rate": 8.003300330033003e-05,
+      "loss": 1.1137,
+      "step": 485
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.578125,
+      "learning_rate": 8.085808580858087e-05,
+      "loss": 1.1352,
+      "step": 490
+    },
+    {
+      "epoch": 0.6122448979591837,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.168316831683169e-05,
+      "loss": 1.1321,
+      "step": 495
+    },
+    {
+      "epoch": 0.6184291898577613,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.250825082508251e-05,
+      "loss": 1.116,
+      "step": 500
+    },
+    {
+      "epoch": 0.6246134817563389,
+      "grad_norm": 0.5703125,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 1.1233,
+      "step": 505
+    },
+    {
+      "epoch": 0.6307977736549165,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.415841584158417e-05,
+      "loss": 1.1291,
+      "step": 510
+    },
+    {
+      "epoch": 0.6369820655534941,
+      "grad_norm": 0.69140625,
+      "learning_rate": 8.498349834983499e-05,
+      "loss": 1.1307,
+      "step": 515
+    },
+    {
+      "epoch": 0.6431663574520717,
+      "grad_norm": 0.54296875,
+      "learning_rate": 8.580858085808581e-05,
+      "loss": 1.1188,
+      "step": 520
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 0.490234375,
+      "learning_rate": 8.663366336633664e-05,
+      "loss": 1.116,
+      "step": 525
+    },
+    {
+      "epoch": 0.655534941249227,
+      "grad_norm": 0.51953125,
+      "learning_rate": 8.745874587458746e-05,
+      "loss": 1.1128,
+      "step": 530
+    },
+    {
+      "epoch": 0.6617192331478046,
+      "grad_norm": 0.58203125,
+      "learning_rate": 8.828382838283829e-05,
+      "loss": 1.1046,
+      "step": 535
+    },
+    {
+      "epoch": 0.6679035250463822,
+      "grad_norm": 0.69921875,
+      "learning_rate": 8.910891089108912e-05,
+      "loss": 1.1184,
+      "step": 540
+    },
+    {
+      "epoch": 0.6740878169449598,
+      "grad_norm": 0.55859375,
+      "learning_rate": 8.993399339933993e-05,
+      "loss": 1.1039,
+      "step": 545
+    },
+    {
+      "epoch": 0.6802721088435374,
+      "grad_norm": 0.65234375,
+      "learning_rate": 9.075907590759076e-05,
+      "loss": 1.1097,
+      "step": 550
+    },
+    {
+      "epoch": 0.686456400742115,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.15841584158416e-05,
+      "loss": 1.1088,
+      "step": 555
+    },
+    {
+      "epoch": 0.6926406926406926,
+      "grad_norm": 0.85546875,
+      "learning_rate": 9.24092409240924e-05,
+      "loss": 1.0955,
+      "step": 560
+    },
+    {
+      "epoch": 0.6988249845392702,
+      "grad_norm": 0.57421875,
+      "learning_rate": 9.323432343234324e-05,
+      "loss": 1.1105,
+      "step": 565
+    },
+    {
+      "epoch": 0.7050092764378478,
+      "grad_norm": 0.6953125,
+      "learning_rate": 9.405940594059406e-05,
+      "loss": 1.1024,
+      "step": 570
+    },
+    {
+      "epoch": 0.7111935683364254,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.488448844884488e-05,
+      "loss": 1.0922,
+      "step": 575
+    },
+    {
+      "epoch": 0.717377860235003,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.570957095709572e-05,
+      "loss": 1.0985,
+      "step": 580
+    },
+    {
+      "epoch": 0.7235621521335807,
+      "grad_norm": 0.5390625,
+      "learning_rate": 9.653465346534654e-05,
+      "loss": 1.0908,
+      "step": 585
+    },
+    {
+      "epoch": 0.7297464440321583,
+      "grad_norm": 0.60546875,
+      "learning_rate": 9.735973597359736e-05,
+      "loss": 1.0882,
+      "step": 590
+    },
+    {
+      "epoch": 0.7359307359307359,
+      "grad_norm": 0.451171875,
+      "learning_rate": 9.818481848184818e-05,
+      "loss": 1.102,
+      "step": 595
+    },
+    {
+      "epoch": 0.7421150278293135,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.900990099009902e-05,
+      "loss": 1.0799,
+      "step": 600
+    },
+    {
+      "epoch": 0.7482993197278912,
+      "grad_norm": 0.57421875,
+      "learning_rate": 9.983498349834984e-05,
+      "loss": 1.0933,
+      "step": 605
+    },
+    {
+      "epoch": 0.7544836116264688,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00010066006600660067,
+      "loss": 1.0881,
+      "step": 610
+    },
+    {
+      "epoch": 0.7606679035250464,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001014851485148515,
+      "loss": 1.0926,
+      "step": 615
+    },
+    {
+      "epoch": 0.766852195423624,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00010231023102310232,
+      "loss": 1.0931,
+      "step": 620
+    },
+    {
+      "epoch": 0.7730364873222016,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00010313531353135315,
+      "loss": 1.0923,
+      "step": 625
+    },
+    {
+      "epoch": 0.7792207792207793,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00010396039603960397,
+      "loss": 1.0854,
+      "step": 630
+    },
+    {
+      "epoch": 0.7854050711193569,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00010478547854785479,
+      "loss": 1.0864,
+      "step": 635
+    },
+    {
+      "epoch": 0.7915893630179345,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001056105610561056,
+      "loss": 1.0863,
+      "step": 640
+    },
+    {
+      "epoch": 0.7977736549165121,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00010643564356435645,
+      "loss": 1.0843,
+      "step": 645
+    },
+    {
+      "epoch": 0.8039579468150897,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00010726072607260727,
+      "loss": 1.0822,
+      "step": 650
+    },
+    {
+      "epoch": 0.8101422387136673,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00010808580858085808,
+      "loss": 1.0802,
+      "step": 655
+    },
+    {
+      "epoch": 0.8163265306122449,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00010891089108910893,
+      "loss": 1.0822,
+      "step": 660
+    },
+    {
+      "epoch": 0.8225108225108225,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00010973597359735975,
+      "loss": 1.076,
+      "step": 665
+    },
+    {
+      "epoch": 0.8286951144094001,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00011056105610561056,
+      "loss": 1.0688,
+      "step": 670
+    },
+    {
+      "epoch": 0.8348794063079777,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001113861386138614,
+      "loss": 1.0839,
+      "step": 675
+    },
+    {
+      "epoch": 0.8410636982065554,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00011221122112211223,
+      "loss": 1.0802,
+      "step": 680
+    },
+    {
+      "epoch": 0.847247990105133,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00011303630363036303,
+      "loss": 1.0699,
+      "step": 685
+    },
+    {
+      "epoch": 0.8534322820037106,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00011386138613861385,
+      "loss": 1.0692,
+      "step": 690
+    },
+    {
+      "epoch": 0.8596165739022882,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001146864686468647,
+      "loss": 1.0756,
+      "step": 695
+    },
+    {
+      "epoch": 0.8658008658008658,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00011551155115511551,
+      "loss": 1.0819,
+      "step": 700
+    },
+    {
+      "epoch": 0.8719851576994434,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00011633663366336633,
+      "loss": 1.071,
+      "step": 705
+    },
+    {
+      "epoch": 0.878169449598021,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00011716171617161718,
+      "loss": 1.069,
+      "step": 710
+    },
+    {
+      "epoch": 0.8843537414965986,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00011798679867986799,
+      "loss": 1.0676,
+      "step": 715
+    },
+    {
+      "epoch": 0.8905380333951762,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00011881188118811881,
+      "loss": 1.0562,
+      "step": 720
+    },
+    {
+      "epoch": 0.8967223252937538,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00011963696369636966,
+      "loss": 1.0725,
+      "step": 725
+    },
+    {
+      "epoch": 0.9029066171923315,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00012046204620462047,
+      "loss": 1.0705,
+      "step": 730
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00012128712871287129,
+      "loss": 1.071,
+      "step": 735
+    },
+    {
+      "epoch": 0.9152752009894867,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00012211221122112212,
+      "loss": 1.0615,
+      "step": 740
+    },
+    {
+      "epoch": 0.9214594928880643,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00012293729372937296,
+      "loss": 1.0725,
+      "step": 745
+    },
+    {
+      "epoch": 0.9276437847866419,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00012376237623762376,
+      "loss": 1.0583,
+      "step": 750
+    },
+    {
+      "epoch": 0.9338280766852195,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0001245874587458746,
+      "loss": 1.059,
+      "step": 755
+    },
+    {
+      "epoch": 0.9400123685837971,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00012541254125412543,
+      "loss": 1.0561,
+      "step": 760
+    },
+    {
+      "epoch": 0.9461966604823747,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00012623762376237624,
+      "loss": 1.0576,
+      "step": 765
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00012706270627062708,
+      "loss": 1.056,
+      "step": 770
+    },
+    {
+      "epoch": 0.95856524427953,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001278877887788779,
+      "loss": 1.0583,
+      "step": 775
+    },
+    {
+      "epoch": 0.9647495361781077,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00012871287128712872,
+      "loss": 1.0501,
+      "step": 780
+    },
+    {
+      "epoch": 0.9709338280766853,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00012953795379537955,
+      "loss": 1.057,
+      "step": 785
+    },
+    {
+      "epoch": 0.9771181199752629,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00013036303630363036,
+      "loss": 1.0529,
+      "step": 790
+    },
+    {
+      "epoch": 0.9833024118738405,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001311881188118812,
+      "loss": 1.0481,
+      "step": 795
+    },
+    {
+      "epoch": 0.9894867037724181,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00013201320132013203,
+      "loss": 1.0671,
+      "step": 800
+    },
+    {
+      "epoch": 0.9956709956709957,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00013283828382838284,
+      "loss": 1.0494,
+      "step": 805
+    },
+    {
+      "epoch": 0.9993815708101422,
+      "eval_loss": 2.4662458896636963,
+      "eval_runtime": 0.6569,
+      "eval_samples_per_second": 15.223,
+      "eval_steps_per_second": 1.522,
+      "step": 808
+    },
+    {
+      "epoch": 1.0018552875695732,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00013366336633663367,
+      "loss": 1.0467,
+      "step": 810
+    },
+    {
+      "epoch": 1.008039579468151,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001344884488448845,
+      "loss": 1.0493,
+      "step": 815
+    },
+    {
+      "epoch": 1.0142238713667284,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00013531353135313532,
+      "loss": 1.0417,
+      "step": 820
+    },
+    {
+      "epoch": 1.0204081632653061,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00013613861386138615,
+      "loss": 1.0413,
+      "step": 825
+    },
+    {
+      "epoch": 1.0265924551638836,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00013696369636963699,
+      "loss": 1.07,
+      "step": 830
+    },
+    {
+      "epoch": 1.0327767470624614,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0001377887788778878,
+      "loss": 1.0425,
+      "step": 835
+    },
+    {
+      "epoch": 1.0389610389610389,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001386138613861386,
+      "loss": 1.0537,
+      "step": 840
+    },
+    {
+      "epoch": 1.0451453308596166,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00013943894389438946,
+      "loss": 1.0425,
+      "step": 845
+    },
+    {
+      "epoch": 1.051329622758194,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00014026402640264027,
+      "loss": 1.0478,
+      "step": 850
+    },
+    {
+      "epoch": 1.0575139146567718,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00014108910891089108,
+      "loss": 1.0427,
+      "step": 855
+    },
+    {
+      "epoch": 1.0636982065553493,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00014191419141914194,
+      "loss": 1.0406,
+      "step": 860
+    },
+    {
+      "epoch": 1.069882498453927,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00014273927392739275,
+      "loss": 1.0384,
+      "step": 865
+    },
+    {
+      "epoch": 1.0760667903525047,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00014356435643564356,
+      "loss": 1.031,
+      "step": 870
+    },
+    {
+      "epoch": 1.0822510822510822,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00014438943894389442,
+      "loss": 1.0371,
+      "step": 875
+    },
+    {
+      "epoch": 1.08843537414966,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00014521452145214523,
+      "loss": 1.0256,
+      "step": 880
+    },
+    {
+      "epoch": 1.0946196660482375,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00014603960396039603,
+      "loss": 1.049,
+      "step": 885
+    },
+    {
+      "epoch": 1.1008039579468152,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00014686468646864687,
+      "loss": 1.0373,
+      "step": 890
+    },
+    {
+      "epoch": 1.1069882498453927,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0001476897689768977,
+      "loss": 1.0404,
+      "step": 895
+    },
+    {
+      "epoch": 1.1131725417439704,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001485148514851485,
+      "loss": 1.0224,
+      "step": 900
+    },
+    {
+      "epoch": 1.119356833642548,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00014933993399339935,
+      "loss": 1.035,
+      "step": 905
+    },
+    {
+      "epoch": 1.1255411255411256,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00015016501650165018,
+      "loss": 1.0376,
+      "step": 910
+    },
+    {
+      "epoch": 1.1317254174397031,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.000150990099009901,
+      "loss": 1.0351,
+      "step": 915
+    },
+    {
+      "epoch": 1.1379097093382808,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00015181518151815182,
+      "loss": 1.041,
+      "step": 920
+    },
+    {
+      "epoch": 1.1440940012368583,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00015264026402640266,
+      "loss": 1.0341,
+      "step": 925
+    },
+    {
+      "epoch": 1.150278293135436,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00015346534653465347,
+      "loss": 1.0292,
+      "step": 930
+    },
+    {
+      "epoch": 1.1564625850340136,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001542904290429043,
+      "loss": 1.0407,
+      "step": 935
+    },
+    {
+      "epoch": 1.1626468769325913,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0001551155115511551,
+      "loss": 1.0252,
+      "step": 940
+    },
+    {
+      "epoch": 1.1688311688311688,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00015594059405940594,
+      "loss": 1.0396,
+      "step": 945
+    },
+    {
+      "epoch": 1.1750154607297465,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00015676567656765678,
+      "loss": 1.0375,
+      "step": 950
+    },
+    {
+      "epoch": 1.181199752628324,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00015759075907590759,
+      "loss": 1.0245,
+      "step": 955
+    },
+    {
+      "epoch": 1.1873840445269017,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00015841584158415842,
+      "loss": 1.0366,
+      "step": 960
+    },
+    {
+      "epoch": 1.1935683364254792,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00015924092409240926,
+      "loss": 1.0336,
+      "step": 965
+    },
+    {
+      "epoch": 1.199752628324057,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00016006600660066006,
+      "loss": 1.0353,
+      "step": 970
+    },
+    {
+      "epoch": 1.2059369202226344,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001608910891089109,
+      "loss": 1.0294,
+      "step": 975
+    },
+    {
+      "epoch": 1.2121212121212122,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00016171617161716173,
+      "loss": 1.0297,
+      "step": 980
+    },
+    {
+      "epoch": 1.2183055040197897,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00016254125412541254,
+      "loss": 1.0302,
+      "step": 985
+    },
+    {
+      "epoch": 1.2244897959183674,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00016336633663366338,
+      "loss": 1.0242,
+      "step": 990
+    },
+    {
+      "epoch": 1.2306740878169449,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0001641914191419142,
+      "loss": 1.0314,
+      "step": 995
+    },
+    {
+      "epoch": 1.2368583797155226,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00016501650165016502,
+      "loss": 1.0194,
+      "step": 1000
+    },
+    {
+      "epoch": 1.2430426716141,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00016584158415841585,
+      "loss": 1.0178,
+      "step": 1005
+    },
+    {
+      "epoch": 1.2492269635126778,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 1.0241,
+      "step": 1010
+    },
+    {
+      "epoch": 1.2554112554112553,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001674917491749175,
+      "loss": 1.0271,
+      "step": 1015
+    },
+    {
+      "epoch": 1.261595547309833,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00016831683168316833,
+      "loss": 1.0307,
+      "step": 1020
+    },
+    {
+      "epoch": 1.2677798392084108,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00016914191419141917,
+      "loss": 1.0366,
+      "step": 1025
+    },
+    {
+      "epoch": 1.2739641311069883,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00016996699669966997,
+      "loss": 1.0276,
+      "step": 1030
+    },
+    {
+      "epoch": 1.2801484230055657,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0001707920792079208,
+      "loss": 1.0177,
+      "step": 1035
+    },
+    {
+      "epoch": 1.2863327149041435,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00017161716171617162,
+      "loss": 1.0301,
+      "step": 1040
+    },
+    {
+      "epoch": 1.2925170068027212,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00017244224422442245,
+      "loss": 1.0193,
+      "step": 1045
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00017326732673267329,
+      "loss": 1.0295,
+      "step": 1050
+    },
+    {
+      "epoch": 1.3048855905998762,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001740924092409241,
+      "loss": 1.0248,
+      "step": 1055
+    },
+    {
+      "epoch": 1.311069882498454,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00017491749174917493,
+      "loss": 1.0218,
+      "step": 1060
+    },
+    {
+      "epoch": 1.3172541743970316,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00017574257425742576,
+      "loss": 1.0141,
+      "step": 1065
+    },
+    {
+      "epoch": 1.3234384662956091,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00017656765676567657,
+      "loss": 1.0232,
+      "step": 1070
+    },
+    {
+      "epoch": 1.3296227581941866,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001773927392739274,
+      "loss": 1.0198,
+      "step": 1075
+    },
+    {
+      "epoch": 1.3358070500927643,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00017821782178217824,
+      "loss": 1.0236,
+      "step": 1080
+    },
+    {
+      "epoch": 1.341991341991342,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00017904290429042905,
+      "loss": 1.031,
+      "step": 1085
+    },
+    {
+      "epoch": 1.3481756338899196,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00017986798679867986,
+      "loss": 1.0143,
+      "step": 1090
+    },
+    {
+      "epoch": 1.3543599257884973,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00018069306930693072,
+      "loss": 1.0273,
+      "step": 1095
+    },
+    {
+      "epoch": 1.3605442176870748,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00018151815181518153,
+      "loss": 1.0228,
+      "step": 1100
+    },
+    {
+      "epoch": 1.3667285095856525,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00018234323432343233,
+      "loss": 1.0276,
+      "step": 1105
+    },
+    {
+      "epoch": 1.37291280148423,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001831683168316832,
+      "loss": 1.0196,
+      "step": 1110
+    },
+    {
+      "epoch": 1.3790970933828077,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.000183993399339934,
+      "loss": 1.0158,
+      "step": 1115
+    },
+    {
+      "epoch": 1.3852813852813852,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0001848184818481848,
+      "loss": 1.0063,
+      "step": 1120
+    },
+    {
+      "epoch": 1.391465677179963,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00018564356435643567,
+      "loss": 1.0192,
+      "step": 1125
+    },
+    {
+      "epoch": 1.3976499690785404,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00018646864686468648,
+      "loss": 1.0265,
+      "step": 1130
+    },
+    {
+      "epoch": 1.4038342609771182,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001872937293729373,
+      "loss": 1.0139,
+      "step": 1135
+    },
+    {
+      "epoch": 1.4100185528756957,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00018811881188118812,
+      "loss": 1.0196,
+      "step": 1140
+    },
+    {
+      "epoch": 1.4162028447742734,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00018894389438943896,
+      "loss": 1.0174,
+      "step": 1145
+    },
+    {
+      "epoch": 1.4223871366728509,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00018976897689768977,
+      "loss": 1.0146,
+      "step": 1150
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001905940594059406,
+      "loss": 1.0182,
+      "step": 1155
+    },
+    {
+      "epoch": 1.434755720470006,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019141914191419144,
+      "loss": 1.0137,
+      "step": 1160
+    },
+    {
+      "epoch": 1.4409400123685838,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019224422442244224,
+      "loss": 1.0134,
+      "step": 1165
+    },
+    {
+      "epoch": 1.4471243042671613,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00019306930693069308,
+      "loss": 1.0123,
+      "step": 1170
+    },
+    {
+      "epoch": 1.453308596165739,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001938943894389439,
+      "loss": 1.011,
+      "step": 1175
+    },
+    {
+      "epoch": 1.4594928880643168,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019471947194719472,
+      "loss": 0.9988,
+      "step": 1180
+    },
+    {
+      "epoch": 1.4656771799628943,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019554455445544556,
+      "loss": 1.0111,
+      "step": 1185
+    },
+    {
+      "epoch": 1.4718614718614718,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019636963696369636,
+      "loss": 1.0151,
+      "step": 1190
+    },
+    {
+      "epoch": 1.4780457637600495,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001971947194719472,
+      "loss": 1.0119,
+      "step": 1195
+    },
+    {
+      "epoch": 1.4842300556586272,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019801980198019803,
+      "loss": 1.0085,
+      "step": 1200
+    },
+    {
+      "epoch": 1.4904143475572047,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00019884488448844884,
+      "loss": 1.0071,
+      "step": 1205
+    },
+    {
+      "epoch": 1.4965986394557822,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019966996699669968,
+      "loss": 1.0071,
+      "step": 1210
+    },
+    {
+      "epoch": 1.50278293135436,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001999999626730957,
+      "loss": 1.0041,
+      "step": 1215
+    },
+    {
+      "epoch": 1.5089672232529376,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019999973456433681,
+      "loss": 1.0051,
+      "step": 1220
+    },
+    {
+      "epoch": 1.5151515151515151,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019999929908446061,
+      "loss": 1.0011,
+      "step": 1225
+    },
+    {
+      "epoch": 1.5213358070500926,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019999865623437013,
+      "loss": 1.007,
+      "step": 1230
+    },
+    {
+      "epoch": 1.5275200989486704,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.00019999780601539848,
+      "loss": 0.9903,
+      "step": 1235
+    },
+    {
+      "epoch": 1.533704390847248,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00019999674842930876,
+      "loss": 1.0055,
+      "step": 1240
+    },
+    {
+      "epoch": 1.5398886827458256,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0001999954834782941,
+      "loss": 0.9958,
+      "step": 1245
+    },
+    {
+      "epoch": 1.546072974644403,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019999401116497763,
+      "loss": 1.0027,
+      "step": 1250
+    },
+    {
+      "epoch": 1.5522572665429808,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019999233149241253,
+      "loss": 1.0078,
+      "step": 1255
+    },
+    {
+      "epoch": 1.5584415584415585,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.000199990444464082,
+      "loss": 1.0108,
+      "step": 1260
+    },
+    {
+      "epoch": 1.564625850340136,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001999883500838992,
+      "loss": 1.0173,
+      "step": 1265
+    },
+    {
+      "epoch": 1.5708101422387135,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019998604835620717,
+      "loss": 1.0167,
+      "step": 1270
+    },
+    {
+      "epoch": 1.5769944341372912,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019998353928577919,
+      "loss": 1.0032,
+      "step": 1275
+    },
+    {
+      "epoch": 1.583178726035869,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019998082287781826,
+      "loss": 1.0133,
+      "step": 1280
+    },
+    {
+      "epoch": 1.5893630179344465,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00019997789913795747,
+      "loss": 0.9938,
+      "step": 1285
+    },
+    {
+      "epoch": 1.595547309833024,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019997476807225985,
+      "loss": 0.9967,
+      "step": 1290
+    },
+    {
+      "epoch": 1.601731601731602,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019997142968721833,
+      "loss": 1.0054,
+      "step": 1295
+    },
+    {
+      "epoch": 1.6079158936301794,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019996788398975578,
+      "loss": 1.0033,
+      "step": 1300
+    },
+    {
+      "epoch": 1.614100185528757,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019996413098722493,
+      "loss": 0.9963,
+      "step": 1305
+    },
+    {
+      "epoch": 1.6202844774273346,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0001999601706874085,
+      "loss": 1.0035,
+      "step": 1310
+    },
+    {
+      "epoch": 1.6264687693259123,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.000199956003098519,
+      "loss": 0.9968,
+      "step": 1315
+    },
+    {
+      "epoch": 1.6326530612244898,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019995162822919883,
+      "loss": 0.9884,
+      "step": 1320
+    },
+    {
+      "epoch": 1.6388373531230673,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00019994704608852022,
+      "loss": 1.0013,
+      "step": 1325
+    },
+    {
+      "epoch": 1.645021645021645,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.00019994225668598526,
+      "loss": 1.0153,
+      "step": 1330
+    },
+    {
+      "epoch": 1.6512059369202228,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019993726003152582,
+      "loss": 1.0002,
+      "step": 1335
+    },
+    {
+      "epoch": 1.6573902288188003,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001999320561355035,
+      "loss": 1.0107,
+      "step": 1340
+    },
+    {
+      "epoch": 1.6635745207173778,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019992664500870976,
+      "loss": 1.0183,
+      "step": 1345
+    },
+    {
+      "epoch": 1.6697588126159555,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019992102666236566,
+      "loss": 1.0067,
+      "step": 1350
+    },
+    {
+      "epoch": 1.6759431045145332,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019991520110812215,
+      "loss": 1.0032,
+      "step": 1355
+    },
+    {
+      "epoch": 1.6821273964131107,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019990916835805974,
+      "loss": 0.994,
+      "step": 1360
+    },
+    {
+      "epoch": 1.6883116883116882,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00019990292842468868,
+      "loss": 0.9995,
+      "step": 1365
+    },
+    {
+      "epoch": 1.694495980210266,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019989648132094873,
+      "loss": 0.9878,
+      "step": 1370
+    },
+    {
+      "epoch": 1.7006802721088436,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019988982706020946,
+      "loss": 1.002,
+      "step": 1375
+    },
+    {
+      "epoch": 1.7068645640074211,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019988296565626987,
+      "loss": 0.9913,
+      "step": 1380
+    },
+    {
+      "epoch": 1.7130488559059986,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019987589712335856,
+      "loss": 1.0095,
+      "step": 1385
+    },
+    {
+      "epoch": 1.7192331478045764,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0001998686214761337,
+      "loss": 0.9962,
+      "step": 1390
+    },
+    {
+      "epoch": 1.725417439703154,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001998611387296829,
+      "loss": 0.9979,
+      "step": 1395
+    },
+    {
+      "epoch": 1.7316017316017316,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019985344889952327,
+      "loss": 1.0057,
+      "step": 1400
+    },
+    {
+      "epoch": 1.737786023500309,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019984555200160128,
+      "loss": 0.9932,
+      "step": 1405
+    },
+    {
+      "epoch": 1.7439703153988868,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019983744805229296,
+      "loss": 0.9973,
+      "step": 1410
+    },
+    {
+      "epoch": 1.7501546072974645,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019982913706840353,
+      "loss": 0.9919,
+      "step": 1415
+    },
+    {
+      "epoch": 1.756338899196042,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00019982061906716764,
+      "loss": 0.9875,
+      "step": 1420
+    },
+    {
+      "epoch": 1.7625231910946195,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019981189406624922,
+      "loss": 1.0013,
+      "step": 1425
+    },
+    {
+      "epoch": 1.7687074829931972,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019980296208374143,
+      "loss": 0.9961,
+      "step": 1430
+    },
+    {
+      "epoch": 1.774891774891775,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019979382313816668,
+      "loss": 0.9928,
+      "step": 1435
+    },
+    {
+      "epoch": 1.7810760667903525,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019978447724847652,
+      "loss": 0.9939,
+      "step": 1440
+    },
+    {
+      "epoch": 1.78726035868893,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019977492443405174,
+      "loss": 0.987,
+      "step": 1445
+    },
+    {
+      "epoch": 1.7934446505875077,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019976516471470216,
+      "loss": 1.0033,
+      "step": 1450
+    },
+    {
+      "epoch": 1.7996289424860854,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019975519811066663,
+      "loss": 0.9884,
+      "step": 1455
+    },
+    {
+      "epoch": 1.805813234384663,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0001997450246426131,
+      "loss": 0.9823,
+      "step": 1460
+    },
+    {
+      "epoch": 1.8119975262832406,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019973464433163844,
+      "loss": 0.9838,
+      "step": 1465
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0001997240571992685,
+      "loss": 0.993,
+      "step": 1470
+    },
+    {
+      "epoch": 1.8243661100803958,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019971326326745793,
+      "loss": 0.9835,
+      "step": 1475
+    },
+    {
+      "epoch": 1.8305504019789733,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00019970226255859038,
+      "loss": 0.99,
+      "step": 1480
+    },
+    {
+      "epoch": 1.836734693877551,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019969105509547812,
+      "loss": 0.9923,
+      "step": 1485
+    },
+    {
+      "epoch": 1.8429189857761288,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001996796409013623,
+      "loss": 0.998,
+      "step": 1490
+    },
+    {
+      "epoch": 1.8491032776747063,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001996680199999127,
+      "loss": 0.9902,
+      "step": 1495
+    },
+    {
+      "epoch": 1.8552875695732838,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001996561924152278,
+      "loss": 0.9845,
+      "step": 1500
+    },
+    {
+      "epoch": 1.8614718614718615,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0001996441581718347,
+      "loss": 0.9722,
+      "step": 1505
+    },
+    {
+      "epoch": 1.8676561533704392,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019963191729468888,
+      "loss": 0.9983,
+      "step": 1510
+    },
+    {
+      "epoch": 1.8738404452690167,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019961946980917456,
+      "loss": 0.9817,
+      "step": 1515
+    },
+    {
+      "epoch": 1.8800247371675942,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019960681574110426,
+      "loss": 0.9803,
+      "step": 1520
+    },
+    {
+      "epoch": 1.886209029066172,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00019959395511671898,
+      "loss": 0.9839,
+      "step": 1525
+    },
+    {
+      "epoch": 1.8923933209647497,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019958088796268793,
+      "loss": 0.9842,
+      "step": 1530
+    },
+    {
+      "epoch": 1.8985776128633272,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019956761430610874,
+      "loss": 0.9782,
+      "step": 1535
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001995541341745072,
+      "loss": 0.9888,
+      "step": 1540
+    },
+    {
+      "epoch": 1.9109461966604824,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001995404475958373,
+      "loss": 0.9859,
+      "step": 1545
+    },
+    {
+      "epoch": 1.91713048855906,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001995265545984811,
+      "loss": 0.9986,
+      "step": 1550
+    },
+    {
+      "epoch": 1.9233147804576376,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019951245521124874,
+      "loss": 0.9737,
+      "step": 1555
+    },
+    {
+      "epoch": 1.929499072356215,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019949814946337838,
+      "loss": 0.977,
+      "step": 1560
+    },
+    {
+      "epoch": 1.9356833642547928,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019948363738453607,
+      "loss": 0.9834,
+      "step": 1565
+    },
+    {
+      "epoch": 1.9418676561533705,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019946891900481578,
+      "loss": 0.9903,
+      "step": 1570
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00019945399435473922,
+      "loss": 0.9813,
+      "step": 1575
+    },
+    {
+      "epoch": 1.9542362399505255,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001994388634652559,
+      "loss": 0.9774,
+      "step": 1580
+    },
+    {
+      "epoch": 1.9604205318491033,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019942352636774296,
+      "loss": 0.9785,
+      "step": 1585
+    },
+    {
+      "epoch": 1.966604823747681,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00019940798309400526,
+      "loss": 0.9799,
+      "step": 1590
+    },
+    {
+      "epoch": 1.9727891156462585,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0001993922336762751,
+      "loss": 0.969,
+      "step": 1595
+    },
+    {
+      "epoch": 1.978973407544836,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019937627814721237,
+      "loss": 0.9758,
+      "step": 1600
+    },
+    {
+      "epoch": 1.9851576994434137,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019936011653990426,
+      "loss": 0.9819,
+      "step": 1605
+    },
+    {
+      "epoch": 1.9913419913419914,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00019934374888786537,
+      "loss": 0.9826,
+      "step": 1610
+    },
+    {
+      "epoch": 1.997526283240569,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001993271752250376,
+      "loss": 0.9661,
+      "step": 1615
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 2.4507155418395996,
+      "eval_runtime": 0.5376,
+      "eval_samples_per_second": 18.6,
+      "eval_steps_per_second": 1.86,
+      "step": 1617
+    },
+    {
+      "epoch": 2.0037105751391464,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00019931039558578997,
+      "loss": 0.9843,
+      "step": 1620
+    },
+    {
+      "epoch": 2.0098948670377244,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019929341000491876,
+      "loss": 0.9676,
+      "step": 1625
+    },
+    {
+      "epoch": 2.016079158936302,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00019927621851764725,
+      "loss": 0.9727,
+      "step": 1630
+    },
+    {
+      "epoch": 2.0222634508348794,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00019925882115962568,
+      "loss": 0.9764,
+      "step": 1635
+    },
+    {
+      "epoch": 2.028447742733457,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019924121796693127,
+      "loss": 0.9708,
+      "step": 1640
+    },
+    {
+      "epoch": 2.034632034632035,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019922340897606805,
+      "loss": 0.9728,
+      "step": 1645
+    },
+    {
+      "epoch": 2.0408163265306123,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.0001992053942239668,
+      "loss": 0.9724,
+      "step": 1650
+    },
+    {
+      "epoch": 2.04700061842919,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019918717374798502,
+      "loss": 0.9669,
+      "step": 1655
+    },
+    {
+      "epoch": 2.0531849103277673,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019916874758590684,
+      "loss": 0.9634,
+      "step": 1660
+    },
+    {
+      "epoch": 2.0593692022263452,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019915011577594286,
+      "loss": 0.9761,
+      "step": 1665
+    },
+    {
+      "epoch": 2.0655534941249227,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019913127835673023,
+      "loss": 0.9667,
+      "step": 1670
+    },
+    {
+      "epoch": 2.0717377860235002,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019911223536733235,
+      "loss": 0.9747,
+      "step": 1675
+    },
+    {
+      "epoch": 2.0779220779220777,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019909298684723904,
+      "loss": 0.9682,
+      "step": 1680
+    },
+    {
+      "epoch": 2.0841063698206557,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019907353283636628,
+      "loss": 0.9606,
+      "step": 1685
+    },
+    {
+      "epoch": 2.090290661719233,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019905387337505612,
+      "loss": 0.9635,
+      "step": 1690
+    },
+    {
+      "epoch": 2.0964749536178107,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019903400850407676,
+      "loss": 0.9612,
+      "step": 1695
+    },
+    {
+      "epoch": 2.102659245516388,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001990139382646223,
+      "loss": 0.9682,
+      "step": 1700
+    },
+    {
+      "epoch": 2.108843537414966,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019899366269831274,
+      "loss": 0.9599,
+      "step": 1705
+    },
+    {
+      "epoch": 2.1150278293135436,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00019897318184719385,
+      "loss": 0.9769,
+      "step": 1710
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019895249575373712,
+      "loss": 0.9683,
+      "step": 1715
+    },
+    {
+      "epoch": 2.1273964131106986,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00019893160446083963,
+      "loss": 0.9577,
+      "step": 1720
+    },
+    {
+      "epoch": 2.1335807050092765,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.000198910508011824,
+      "loss": 0.9543,
+      "step": 1725
+    },
+    {
+      "epoch": 2.139764996907854,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00019888920645043831,
+      "loss": 0.9643,
+      "step": 1730
+    },
+    {
+      "epoch": 2.1459492888064315,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019886769982085597,
+      "loss": 0.9651,
+      "step": 1735
+    },
+    {
+      "epoch": 2.1521335807050095,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00019884598816767563,
+      "loss": 0.9451,
+      "step": 1740
+    },
+    {
+      "epoch": 2.158317872603587,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019882407153592107,
+      "loss": 0.9691,
+      "step": 1745
+    },
+    {
+      "epoch": 2.1645021645021645,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019880194997104123,
+      "loss": 0.9687,
+      "step": 1750
+    },
+    {
+      "epoch": 2.170686456400742,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019877962351890993,
+      "loss": 0.9574,
+      "step": 1755
+    },
+    {
+      "epoch": 2.17687074829932,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019875709222582594,
+      "loss": 0.9632,
+      "step": 1760
+    },
+    {
+      "epoch": 2.1830550401978974,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019873435613851275,
+      "loss": 0.9643,
+      "step": 1765
+    },
+    {
+      "epoch": 2.189239332096475,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00019871141530411853,
+      "loss": 0.9664,
+      "step": 1770
+    },
+    {
+      "epoch": 2.1954236239950524,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00019868826977021615,
+      "loss": 0.9668,
+      "step": 1775
+    },
+    {
+      "epoch": 2.2016079158936304,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019866491958480284,
+      "loss": 0.9619,
+      "step": 1780
+    },
+    {
+      "epoch": 2.207792207792208,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001986413647963003,
+      "loss": 0.9564,
+      "step": 1785
+    },
+    {
+      "epoch": 2.2139764996907854,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019861760545355442,
+      "loss": 0.9649,
+      "step": 1790
+    },
+    {
+      "epoch": 2.220160791589363,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00019859364160583544,
+      "loss": 0.9453,
+      "step": 1795
+    },
+    {
+      "epoch": 2.226345083487941,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019856947330283752,
+      "loss": 0.9669,
+      "step": 1800
+    },
+    {
+      "epoch": 2.2325293753865183,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0001985451005946789,
+      "loss": 0.9721,
+      "step": 1805
+    },
+    {
+      "epoch": 2.238713667285096,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019852052353190166,
+      "loss": 0.9657,
+      "step": 1810
+    },
+    {
+      "epoch": 2.2448979591836733,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019849574216547171,
+      "loss": 0.9648,
+      "step": 1815
+    },
+    {
+      "epoch": 2.2510822510822512,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0001984707565467785,
+      "loss": 0.9544,
+      "step": 1820
+    },
+    {
+      "epoch": 2.2572665429808287,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0001984455667276352,
+      "loss": 0.9737,
+      "step": 1825
+    },
+    {
+      "epoch": 2.2634508348794062,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019842017276027832,
+      "loss": 0.9538,
+      "step": 1830
+    },
+    {
+      "epoch": 2.2696351267779837,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019839457469736775,
+      "loss": 0.9745,
+      "step": 1835
+    },
+    {
+      "epoch": 2.2758194186765617,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019836877259198662,
+      "loss": 0.958,
+      "step": 1840
+    },
+    {
+      "epoch": 2.282003710575139,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019834276649764124,
+      "loss": 0.9637,
+      "step": 1845
+    },
+    {
+      "epoch": 2.2881880024737167,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001983165564682608,
+      "loss": 0.9706,
+      "step": 1850
+    },
+    {
+      "epoch": 2.2943722943722946,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019829014255819753,
+      "loss": 0.9671,
+      "step": 1855
+    },
+    {
+      "epoch": 2.300556586270872,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019826352482222638,
+      "loss": 0.9571,
+      "step": 1860
+    },
+    {
+      "epoch": 2.3067408781694496,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.000198236703315545,
+      "loss": 0.9581,
+      "step": 1865
+    },
+    {
+      "epoch": 2.312925170068027,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019820967809377357,
+      "loss": 0.9676,
+      "step": 1870
+    },
+    {
+      "epoch": 2.3191094619666046,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001981824492129548,
+      "loss": 0.9489,
+      "step": 1875
+    },
+    {
+      "epoch": 2.3252937538651826,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00019815501672955358,
+      "loss": 0.9609,
+      "step": 1880
+    },
+    {
+      "epoch": 2.33147804576376,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0001981273807004572,
+      "loss": 0.9569,
+      "step": 1885
+    },
+    {
+      "epoch": 2.3376623376623376,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0001980995411829749,
+      "loss": 0.959,
+      "step": 1890
+    },
+    {
+      "epoch": 2.3438466295609155,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00019807149823483798,
+      "loss": 0.9614,
+      "step": 1895
+    },
+    {
+      "epoch": 2.350030921459493,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019804325191419956,
+      "loss": 0.9687,
+      "step": 1900
+    },
+    {
+      "epoch": 2.3562152133580705,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0001980148022796345,
+      "loss": 0.9598,
+      "step": 1905
+    },
+    {
+      "epoch": 2.362399505256648,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019798614939013932,
+      "loss": 0.9572,
+      "step": 1910
+    },
+    {
+      "epoch": 2.3685837971552255,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019795729330513196,
+      "loss": 0.9586,
+      "step": 1915
+    },
+    {
+      "epoch": 2.3747680890538034,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019792823408445174,
+      "loss": 0.9532,
+      "step": 1920
+    },
+    {
+      "epoch": 2.380952380952381,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019789897178835926,
+      "loss": 0.9444,
+      "step": 1925
+    },
+    {
+      "epoch": 2.3871366728509584,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001978695064775363,
+      "loss": 0.9584,
+      "step": 1930
+    },
+    {
+      "epoch": 2.3933209647495364,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001978398382130855,
+      "loss": 0.9628,
+      "step": 1935
+    },
+    {
+      "epoch": 2.399505256648114,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00019780996705653044,
+      "loss": 0.9545,
+      "step": 1940
+    },
+    {
+      "epoch": 2.4056895485466914,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00019777989306981542,
+      "loss": 0.9496,
+      "step": 1945
+    },
+    {
+      "epoch": 2.411873840445269,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019774961631530545,
+      "loss": 0.9541,
+      "step": 1950
+    },
+    {
+      "epoch": 2.418058132343847,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00019771913685578585,
+      "loss": 0.9581,
+      "step": 1955
+    },
+    {
+      "epoch": 2.4242424242424243,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0001976884547544624,
+      "loss": 0.9535,
+      "step": 1960
+    },
+    {
+      "epoch": 2.430426716141002,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019765757007496115,
+      "loss": 0.9595,
+      "step": 1965
+    },
+    {
+      "epoch": 2.4366110080395793,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001976264828813281,
+      "loss": 0.9693,
+      "step": 1970
+    },
+    {
+      "epoch": 2.4427952999381572,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00019759519323802932,
+      "loss": 0.9582,
+      "step": 1975
+    },
+    {
+      "epoch": 2.4489795918367347,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019756370120995066,
+      "loss": 0.9525,
+      "step": 1980
+    },
+    {
+      "epoch": 2.4551638837353122,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00019753200686239763,
+      "loss": 0.95,
+      "step": 1985
+    },
+    {
+      "epoch": 2.4613481756338897,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001975001102610954,
+      "loss": 0.9582,
+      "step": 1990
+    },
+    {
+      "epoch": 2.4675324675324677,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019746801147218842,
+      "loss": 0.9541,
+      "step": 1995
+    },
+    {
+      "epoch": 2.473716759431045,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001974357105622405,
+      "loss": 0.9529,
+      "step": 2000
+    },
+    {
+      "epoch": 2.4799010513296227,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019740320759823458,
+      "loss": 0.9606,
+      "step": 2005
+    },
+    {
+      "epoch": 2.4860853432282,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001973705026475726,
+      "loss": 0.9632,
+      "step": 2010
+    },
+    {
+      "epoch": 2.492269635126778,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019733759577807538,
+      "loss": 0.9567,
+      "step": 2015
+    },
+    {
+      "epoch": 2.4984539270253556,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019730448705798239,
+      "loss": 0.9492,
+      "step": 2020
+    },
+    {
+      "epoch": 2.504638218923933,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0001972711765559518,
+      "loss": 0.96,
+      "step": 2025
+    },
+    {
+      "epoch": 2.5108225108225106,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0001972376643410601,
+      "loss": 0.9524,
+      "step": 2030
+    },
+    {
+      "epoch": 2.5170068027210886,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00019720395048280215,
+      "loss": 0.9538,
+      "step": 2035
+    },
+    {
+      "epoch": 2.523191094619666,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00019717003505109095,
+      "loss": 0.9492,
+      "step": 2040
+    },
+    {
+      "epoch": 2.5293753865182436,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001971359181162575,
+      "loss": 0.9496,
+      "step": 2045
+    },
+    {
+      "epoch": 2.5355596784168215,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00019710159974905064,
+      "loss": 0.9625,
+      "step": 2050
+    },
+    {
+      "epoch": 2.541743970315399,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00019706708002063694,
+      "loss": 0.9555,
+      "step": 2055
+    },
+    {
+      "epoch": 2.5479282622139765,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00019703235900260055,
+      "loss": 0.9655,
+      "step": 2060
+    },
+    {
+      "epoch": 2.554112554112554,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00019699743676694303,
+      "loss": 0.9554,
+      "step": 2065
+    },
+    {
+      "epoch": 2.5602968460111315,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00019696231338608316,
+      "loss": 0.9522,
+      "step": 2070
+    },
+    {
+      "epoch": 2.5664811379097094,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019692698893285693,
+      "loss": 0.952,
+      "step": 2075
+    },
+    {
+      "epoch": 2.572665429808287,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00019689146348051719,
+      "loss": 0.9524,
+      "step": 2080
+    },
+    {
+      "epoch": 2.5788497217068644,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019685573710273376,
+      "loss": 0.9523,
+      "step": 2085
+    },
+    {
+      "epoch": 2.5850340136054424,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001968198098735929,
+      "loss": 0.9491,
+      "step": 2090
+    },
+    {
+      "epoch": 2.59121830550402,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0001967836818675976,
+      "loss": 0.9496,
+      "step": 2095
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001967473531596671,
+      "loss": 0.9408,
+      "step": 2100
+    },
+    {
+      "epoch": 2.603586889301175,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001967108238251368,
+      "loss": 0.9526,
+      "step": 2105
+    },
+    {
+      "epoch": 2.6097711811997524,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019667409393975822,
+      "loss": 0.9476,
+      "step": 2110
+    },
+    {
+      "epoch": 2.6159554730983303,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019663716357969874,
+      "loss": 0.9495,
+      "step": 2115
+    },
+    {
+      "epoch": 2.622139764996908,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019660003282154147,
+      "loss": 0.945,
+      "step": 2120
+    },
+    {
+      "epoch": 2.6283240568954853,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00019656270174228503,
+      "loss": 0.9506,
+      "step": 2125
+    },
+    {
+      "epoch": 2.6345083487940633,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00019652517041934356,
+      "loss": 0.9483,
+      "step": 2130
+    },
+    {
+      "epoch": 2.6406926406926408,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001964874389305464,
+      "loss": 0.948,
+      "step": 2135
+    },
+    {
+      "epoch": 2.6468769325912183,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019644950735413788,
+      "loss": 0.9464,
+      "step": 2140
+    },
+    {
+      "epoch": 2.6530612244897958,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00019641137576877744,
+      "loss": 0.944,
+      "step": 2145
+    },
+    {
+      "epoch": 2.6592455163883733,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019637304425353916,
+      "loss": 0.9437,
+      "step": 2150
+    },
+    {
+      "epoch": 2.665429808286951,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019633451288791166,
+      "loss": 0.9494,
+      "step": 2155
+    },
+    {
+      "epoch": 2.6716141001855287,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001962957817517982,
+      "loss": 0.9497,
+      "step": 2160
+    },
+    {
+      "epoch": 2.6777983920841066,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00019625685092551612,
+      "loss": 0.9436,
+      "step": 2165
+    },
+    {
+      "epoch": 2.683982683982684,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001962177204897969,
+      "loss": 0.9509,
+      "step": 2170
+    },
+    {
+      "epoch": 2.6901669758812616,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019617839052578603,
+      "loss": 0.9474,
+      "step": 2175
+    },
+    {
+      "epoch": 2.696351267779839,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001961388611150427,
+      "loss": 0.942,
+      "step": 2180
+    },
+    {
+      "epoch": 2.7025355596784166,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00019609913233953967,
+      "loss": 0.9558,
+      "step": 2185
+    },
+    {
+      "epoch": 2.7087198515769946,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00019605920428166323,
+      "loss": 0.9616,
+      "step": 2190
+    },
+    {
+      "epoch": 2.714904143475572,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.0001960190770242128,
+      "loss": 0.9594,
+      "step": 2195
+    },
+    {
+      "epoch": 2.7210884353741496,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00019597875065040094,
+      "loss": 0.9455,
+      "step": 2200
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00019593822524385316,
+      "loss": 0.9458,
+      "step": 2205
+    },
+    {
+      "epoch": 2.733457019171305,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00019589750088860766,
+      "loss": 0.9468,
+      "step": 2210
+    },
+    {
+      "epoch": 2.7396413110698825,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019585657766911524,
+      "loss": 0.9448,
+      "step": 2215
+    },
+    {
+      "epoch": 2.74582560296846,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.000195815455670239,
+      "loss": 0.9415,
+      "step": 2220
+    },
+    {
+      "epoch": 2.7520098948670375,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00019577413497725438,
+      "loss": 0.9419,
+      "step": 2225
+    },
+    {
+      "epoch": 2.7581941867656155,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019573261567584874,
+      "loss": 0.9409,
+      "step": 2230
+    },
+    {
+      "epoch": 2.764378478664193,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001956908978521214,
+      "loss": 0.9544,
+      "step": 2235
+    },
+    {
+      "epoch": 2.7705627705627704,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00019564898159258324,
+      "loss": 0.9327,
+      "step": 2240
+    },
+    {
+      "epoch": 2.7767470624613484,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019560686698415677,
+      "loss": 0.9478,
+      "step": 2245
+    },
+    {
+      "epoch": 2.782931354359926,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019556455411417573,
+      "loss": 0.9384,
+      "step": 2250
+    },
+    {
+      "epoch": 2.7891156462585034,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00019552204307038502,
+      "loss": 0.9451,
+      "step": 2255
+    },
+    {
+      "epoch": 2.795299938157081,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001954793339409405,
+      "loss": 0.9485,
+      "step": 2260
+    },
+    {
+      "epoch": 2.8014842300556584,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001954364268144088,
+      "loss": 0.9571,
+      "step": 2265
+    },
+    {
+      "epoch": 2.8076685219542363,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00019539332177976714,
+      "loss": 0.9504,
+      "step": 2270
+    },
+    {
+      "epoch": 2.813852813852814,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00019535001892640317,
+      "loss": 0.9422,
+      "step": 2275
+    },
+    {
+      "epoch": 2.8200371057513913,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00019530651834411474,
+      "loss": 0.9473,
+      "step": 2280
+    },
+    {
+      "epoch": 2.8262213976499693,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00019526282012310975,
+      "loss": 0.9467,
+      "step": 2285
+    },
+    {
+      "epoch": 2.8324056895485468,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00019521892435400587,
+      "loss": 0.955,
+      "step": 2290
+    },
+    {
+      "epoch": 2.8385899814471243,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019517483112783054,
+      "loss": 0.9507,
+      "step": 2295
+    },
+    {
+      "epoch": 2.8447742733457018,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00019513054053602055,
+      "loss": 0.9447,
+      "step": 2300
+    },
+    {
+      "epoch": 2.8509585652442793,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00019508605267042214,
+      "loss": 0.9553,
+      "step": 2305
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00019504136762329047,
+      "loss": 0.9454,
+      "step": 2310
+    },
+    {
+      "epoch": 2.8633271490414347,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019499648548728965,
+      "loss": 0.9529,
+      "step": 2315
+    },
+    {
+      "epoch": 2.869511440940012,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019495140635549261,
+      "loss": 0.9455,
+      "step": 2320
+    },
+    {
+      "epoch": 2.87569573283859,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00019490613032138062,
+      "loss": 0.9468,
+      "step": 2325
+    },
+    {
+      "epoch": 2.8818800247371676,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00019486065747884333,
+      "loss": 0.9527,
+      "step": 2330
+    },
+    {
+      "epoch": 2.888064316635745,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001948149879221786,
+      "loss": 0.9526,
+      "step": 2335
+    },
+    {
+      "epoch": 2.8942486085343226,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0001947691217460921,
+      "loss": 0.9494,
+      "step": 2340
+    },
+    {
+      "epoch": 2.9004329004329006,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019472305904569729,
+      "loss": 0.9309,
+      "step": 2345
+    },
+    {
+      "epoch": 2.906617192331478,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001946767999165152,
+      "loss": 0.9505,
+      "step": 2350
+    },
+    {
+      "epoch": 2.9128014842300556,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0001946303444544741,
+      "loss": 0.9355,
+      "step": 2355
+    },
+    {
+      "epoch": 2.9189857761286335,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00019458369275590954,
+      "loss": 0.9423,
+      "step": 2360
+    },
+    {
+      "epoch": 2.925170068027211,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00019453684491756382,
+      "loss": 0.9392,
+      "step": 2365
+    },
+    {
+      "epoch": 2.9313543599257885,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00019448980103658613,
+      "loss": 0.9435,
+      "step": 2370
+    },
+    {
+      "epoch": 2.937538651824366,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019444256121053217,
+      "loss": 0.9505,
+      "step": 2375
+    },
+    {
+      "epoch": 2.9437229437229435,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019439512553736394,
+      "loss": 0.945,
+      "step": 2380
+    },
+    {
+      "epoch": 2.9499072356215215,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019434749411544958,
+      "loss": 0.9408,
+      "step": 2385
+    },
+    {
+      "epoch": 2.956091527520099,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0001942996670435632,
+      "loss": 0.9521,
+      "step": 2390
+    },
+    {
+      "epoch": 2.9622758194186765,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00019425164442088451,
+      "loss": 0.9453,
+      "step": 2395
+    },
+    {
+      "epoch": 2.9684601113172544,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001942034263469989,
+      "loss": 0.9481,
+      "step": 2400
+    },
+    {
+      "epoch": 2.974644403215832,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.000194155012921897,
+      "loss": 0.9466,
+      "step": 2405
+    },
+    {
+      "epoch": 2.9808286951144094,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.0001941064042459745,
+      "loss": 0.9376,
+      "step": 2410
+    },
+    {
+      "epoch": 2.987012987012987,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019405760042003203,
+      "loss": 0.9422,
+      "step": 2415
+    },
+    {
+      "epoch": 2.9931972789115644,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019400860154527493,
+      "loss": 0.9541,
+      "step": 2420
+    },
+    {
+      "epoch": 2.9993815708101423,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0001939594077233129,
+      "loss": 0.9526,
+      "step": 2425
+    },
+    {
+      "epoch": 2.9993815708101423,
+      "eval_loss": 2.465909481048584,
+      "eval_runtime": 0.6411,
+      "eval_samples_per_second": 15.599,
+      "eval_steps_per_second": 1.56,
+      "step": 2425
+    },
+    {
+      "epoch": 3.00556586270872,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001939100190561601,
+      "loss": 0.935,
+      "step": 2430
+    },
+    {
+      "epoch": 3.0117501546072973,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00019386043564623452,
+      "loss": 0.9371,
+      "step": 2435
+    },
+    {
+      "epoch": 3.0179344465058753,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019381065759635822,
+      "loss": 0.932,
+      "step": 2440
+    },
+    {
+      "epoch": 3.0241187384044528,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00019376068500975667,
+      "loss": 0.93,
+      "step": 2445
+    },
+    {
+      "epoch": 3.0303030303030303,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0001937105179900589,
+      "loss": 0.9412,
+      "step": 2450
+    },
+    {
+      "epoch": 3.0364873222016078,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019366015664129714,
+      "loss": 0.9272,
+      "step": 2455
+    },
+    {
+      "epoch": 3.0426716141001857,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019360960106790643,
+      "loss": 0.9351,
+      "step": 2460
+    },
+    {
+      "epoch": 3.048855905998763,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00019355885137472488,
+      "loss": 0.9307,
+      "step": 2465
+    },
+    {
+      "epoch": 3.0550401978973407,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019350790766699282,
+      "loss": 0.9332,
+      "step": 2470
+    },
+    {
+      "epoch": 3.061224489795918,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019345677005035315,
+      "loss": 0.9351,
+      "step": 2475
+    },
+    {
+      "epoch": 3.067408781694496,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001934054386308508,
+      "loss": 0.9241,
+      "step": 2480
+    },
+    {
+      "epoch": 3.0735930735930737,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019335391351493257,
+      "loss": 0.9182,
+      "step": 2485
+    },
+    {
+      "epoch": 3.079777365491651,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00019330219480944694,
+      "loss": 0.9133,
+      "step": 2490
+    },
+    {
+      "epoch": 3.0859616573902287,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00019325028262164384,
+      "loss": 0.9386,
+      "step": 2495
+    },
+    {
+      "epoch": 3.0921459492888066,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.0001931981770591745,
+      "loss": 0.9302,
+      "step": 2500
+    },
+    {
+      "epoch": 3.098330241187384,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00019314587823009103,
+      "loss": 0.9421,
+      "step": 2505
+    },
+    {
+      "epoch": 3.1045145330859616,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00019309338624284644,
+      "loss": 0.9286,
+      "step": 2510
+    },
+    {
+      "epoch": 3.110698824984539,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001930407012062942,
+      "loss": 0.9291,
+      "step": 2515
+    },
+    {
+      "epoch": 3.116883116883117,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00019298782322968815,
+      "loss": 0.9332,
+      "step": 2520
+    },
+    {
+      "epoch": 3.1230674087816945,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00019293475242268223,
+      "loss": 0.9427,
+      "step": 2525
+    },
+    {
+      "epoch": 3.129251700680272,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001928814888953303,
+      "loss": 0.9277,
+      "step": 2530
+    },
+    {
+      "epoch": 3.1354359925788495,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0001928280327580858,
+      "loss": 0.9286,
+      "step": 2535
+    },
+    {
+      "epoch": 3.1416202844774275,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0001927743841218016,
+      "loss": 0.9214,
+      "step": 2540
+    },
+    {
+      "epoch": 3.147804576376005,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00019272054309772978,
+      "loss": 0.9328,
+      "step": 2545
+    },
+    {
+      "epoch": 3.1539888682745825,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019266650979752136,
+      "loss": 0.9427,
+      "step": 2550
+    },
+    {
+      "epoch": 3.16017316017316,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001926122843332261,
+      "loss": 0.9285,
+      "step": 2555
+    },
+    {
+      "epoch": 3.166357452071738,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00019255786681729225,
+      "loss": 0.9344,
+      "step": 2560
+    },
+    {
+      "epoch": 3.1725417439703154,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019250325736256633,
+      "loss": 0.9332,
+      "step": 2565
+    },
+    {
+      "epoch": 3.178726035868893,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019244845608229293,
+      "loss": 0.9357,
+      "step": 2570
+    },
+    {
+      "epoch": 3.1849103277674704,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00019239346309011426,
+      "loss": 0.937,
+      "step": 2575
+    },
+    {
+      "epoch": 3.1910946196660483,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00019233827850007027,
+      "loss": 0.9332,
+      "step": 2580
+    },
+    {
+      "epoch": 3.197278911564626,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00019228290242659816,
+      "loss": 0.937,
+      "step": 2585
+    },
+    {
+      "epoch": 3.2034632034632033,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00019222733498453222,
+      "loss": 0.9214,
+      "step": 2590
+    },
+    {
+      "epoch": 3.2096474953617813,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001921715762891036,
+      "loss": 0.9329,
+      "step": 2595
+    },
+    {
+      "epoch": 3.215831787260359,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019211562645594002,
+      "loss": 0.9317,
+      "step": 2600
+    },
+    {
+      "epoch": 3.2220160791589363,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00019205948560106556,
+      "loss": 0.9232,
+      "step": 2605
+    },
+    {
+      "epoch": 3.228200371057514,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019200315384090044,
+      "loss": 0.9325,
+      "step": 2610
+    },
+    {
+      "epoch": 3.2343846629560917,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019194663129226084,
+      "loss": 0.9319,
+      "step": 2615
+    },
+    {
+      "epoch": 3.2405689548546692,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019188991807235844,
+      "loss": 0.9368,
+      "step": 2620
+    },
+    {
+      "epoch": 3.2467532467532467,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019183301429880043,
+      "loss": 0.9295,
+      "step": 2625
+    },
+    {
+      "epoch": 3.2529375386518242,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0001917759200895891,
+      "loss": 0.9421,
+      "step": 2630
+    },
+    {
+      "epoch": 3.259121830550402,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00019171863556312167,
+      "loss": 0.9335,
+      "step": 2635
+    },
+    {
+      "epoch": 3.2653061224489797,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019166116083819002,
+      "loss": 0.9317,
+      "step": 2640
+    },
+    {
+      "epoch": 3.271490414347557,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00019160349603398043,
+      "loss": 0.9297,
+      "step": 2645
+    },
+    {
+      "epoch": 3.2776747062461347,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00019154564127007336,
+      "loss": 0.9371,
+      "step": 2650
+    },
+    {
+      "epoch": 3.2838589981447126,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019148759666644325,
+      "loss": 0.9286,
+      "step": 2655
+    },
+    {
+      "epoch": 3.29004329004329,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0001914293623434581,
+      "loss": 0.9404,
+      "step": 2660
+    },
+    {
+      "epoch": 3.2962275819418676,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00019137093842187944,
+      "loss": 0.9292,
+      "step": 2665
+    },
+    {
+      "epoch": 3.302411873840445,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019131232502286188,
+      "loss": 0.9224,
+      "step": 2670
+    },
+    {
+      "epoch": 3.308596165739023,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00019125352226795307,
+      "loss": 0.9313,
+      "step": 2675
+    },
+    {
+      "epoch": 3.3147804576376005,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00019119453027909323,
+      "loss": 0.928,
+      "step": 2680
+    },
+    {
+      "epoch": 3.320964749536178,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00019113534917861502,
+      "loss": 0.9357,
+      "step": 2685
+    },
+    {
+      "epoch": 3.3271490414347555,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0001910759790892433,
+      "loss": 0.9339,
+      "step": 2690
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001910164201340948,
+      "loss": 0.9342,
+      "step": 2695
+    },
+    {
+      "epoch": 3.339517625231911,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0001909566724366779,
+      "loss": 0.9217,
+      "step": 2700
+    },
+    {
+      "epoch": 3.3457019171304885,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019089673612089243,
+      "loss": 0.9239,
+      "step": 2705
+    },
+    {
+      "epoch": 3.3518862090290664,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019083661131102933,
+      "loss": 0.9289,
+      "step": 2710
+    },
+    {
+      "epoch": 3.358070500927644,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019077629813177036,
+      "loss": 0.925,
+      "step": 2715
+    },
+    {
+      "epoch": 3.3642547928262214,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00019071579670818808,
+      "loss": 0.9251,
+      "step": 2720
+    },
+    {
+      "epoch": 3.370439084724799,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019065510716574516,
+      "loss": 0.9272,
+      "step": 2725
+    },
+    {
+      "epoch": 3.3766233766233764,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00019059422963029464,
+      "loss": 0.9264,
+      "step": 2730
+    },
+    {
+      "epoch": 3.3828076685219544,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019053316422807922,
+      "loss": 0.9169,
+      "step": 2735
+    },
+    {
+      "epoch": 3.388991960420532,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00019047191108573125,
+      "loss": 0.9299,
+      "step": 2740
+    },
+    {
+      "epoch": 3.3951762523191094,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019041047033027236,
+      "loss": 0.9293,
+      "step": 2745
+    },
+    {
+      "epoch": 3.4013605442176873,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00019034884208911335,
+      "loss": 0.9163,
+      "step": 2750
+    },
+    {
+      "epoch": 3.407544836116265,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00019028702649005364,
+      "loss": 0.927,
+      "step": 2755
+    },
+    {
+      "epoch": 3.4137291280148423,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00019022502366128135,
+      "loss": 0.9231,
+      "step": 2760
+    },
+    {
+      "epoch": 3.41991341991342,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00019016283373137274,
+      "loss": 0.9216,
+      "step": 2765
+    },
+    {
+      "epoch": 3.4260977118119973,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019010045682929213,
+      "loss": 0.9191,
+      "step": 2770
+    },
+    {
+      "epoch": 3.4322820037105752,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00019003789308439148,
+      "loss": 0.9167,
+      "step": 2775
+    },
+    {
+      "epoch": 3.4384662956091527,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00018997514262641035,
+      "loss": 0.9203,
+      "step": 2780
+    },
+    {
+      "epoch": 3.4446505875077302,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00018991220558547533,
+      "loss": 0.9316,
+      "step": 2785
+    },
+    {
+      "epoch": 3.450834879406308,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001898490820921001,
+      "loss": 0.9255,
+      "step": 2790
+    },
+    {
+      "epoch": 3.4570191713048857,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00018978577227718484,
+      "loss": 0.9166,
+      "step": 2795
+    },
+    {
+      "epoch": 3.463203463203463,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00018972227627201617,
+      "loss": 0.928,
+      "step": 2800
+    },
+    {
+      "epoch": 3.4693877551020407,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00018965859420826684,
+      "loss": 0.9247,
+      "step": 2805
+    },
+    {
+      "epoch": 3.4755720470006186,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.0001895947262179954,
+      "loss": 0.9221,
+      "step": 2810
+    },
+    {
+      "epoch": 3.481756338899196,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00018953067243364594,
+      "loss": 0.9183,
+      "step": 2815
+    },
+    {
+      "epoch": 3.4879406307977736,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00018946643298804793,
+      "loss": 0.925,
+      "step": 2820
+    },
+    {
+      "epoch": 3.494124922696351,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0001894020080144157,
+      "loss": 0.9341,
+      "step": 2825
+    },
+    {
+      "epoch": 3.500309214594929,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00018933739764634847,
+      "loss": 0.9188,
+      "step": 2830
+    },
+    {
+      "epoch": 3.5064935064935066,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00018927260201782978,
+      "loss": 0.9297,
+      "step": 2835
+    },
+    {
+      "epoch": 3.512677798392084,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001892076212632274,
+      "loss": 0.9197,
+      "step": 2840
+    },
+    {
+      "epoch": 3.5188620902906615,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00018914245551729305,
+      "loss": 0.9139,
+      "step": 2845
+    },
+    {
+      "epoch": 3.5250463821892395,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00018907710491516199,
+      "loss": 0.9252,
+      "step": 2850
+    },
+    {
+      "epoch": 3.531230674087817,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00018901156959235285,
+      "loss": 0.926,
+      "step": 2855
+    },
+    {
+      "epoch": 3.5374149659863945,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00018894584968476733,
+      "loss": 0.926,
+      "step": 2860
+    },
+    {
+      "epoch": 3.5435992578849724,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001888799453286899,
+      "loss": 0.9311,
+      "step": 2865
+    },
+    {
+      "epoch": 3.54978354978355,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00018881385666078755,
+      "loss": 0.9264,
+      "step": 2870
+    },
+    {
+      "epoch": 3.5559678416821274,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00018874758381810943,
+      "loss": 0.9353,
+      "step": 2875
+    },
+    {
+      "epoch": 3.562152133580705,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00018868112693808665,
+      "loss": 0.931,
+      "step": 2880
+    },
+    {
+      "epoch": 3.5683364254792824,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0001886144861585319,
+      "loss": 0.9156,
+      "step": 2885
+    },
+    {
+      "epoch": 3.5745207173778604,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00018854766161763932,
+      "loss": 0.933,
+      "step": 2890
+    },
+    {
+      "epoch": 3.580705009276438,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0001884806534539841,
+      "loss": 0.9268,
+      "step": 2895
+    },
+    {
+      "epoch": 3.5868893011750154,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00018841346180652213,
+      "loss": 0.92,
+      "step": 2900
+    },
+    {
+      "epoch": 3.5930735930735933,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00018834608681458988,
+      "loss": 0.9125,
+      "step": 2905
+    },
+    {
+      "epoch": 3.599257884972171,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00018827852861790398,
+      "loss": 0.9187,
+      "step": 2910
+    },
+    {
+      "epoch": 3.6054421768707483,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00018821078735656101,
+      "loss": 0.9253,
+      "step": 2915
+    },
+    {
+      "epoch": 3.611626468769326,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00018814286317103714,
+      "loss": 0.9273,
+      "step": 2920
+    },
+    {
+      "epoch": 3.6178107606679033,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00018807475620218788,
+      "loss": 0.9167,
+      "step": 2925
+    },
+    {
+      "epoch": 3.6239950525664812,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00018800646659124782,
+      "loss": 0.9192,
+      "step": 2930
+    },
+    {
+      "epoch": 3.6301793444650587,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00018793799447983025,
+      "loss": 0.9288,
+      "step": 2935
+    },
+    {
+      "epoch": 3.6363636363636362,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00018786934000992688,
+      "loss": 0.9198,
+      "step": 2940
+    },
+    {
+      "epoch": 3.642547928262214,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00018780050332390768,
+      "loss": 0.9157,
+      "step": 2945
+    },
+    {
+      "epoch": 3.6487322201607917,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00018773148456452046,
+      "loss": 0.9139,
+      "step": 2950
+    },
+    {
+      "epoch": 3.654916512059369,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00018766228387489048,
+      "loss": 0.9218,
+      "step": 2955
+    },
+    {
+      "epoch": 3.6611008039579467,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00018759290139852048,
+      "loss": 0.9315,
+      "step": 2960
+    },
+    {
+      "epoch": 3.667285095856524,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00018752333727928993,
+      "loss": 0.9291,
+      "step": 2965
+    },
+    {
+      "epoch": 3.673469387755102,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00018745359166145523,
+      "loss": 0.915,
+      "step": 2970
+    },
+    {
+      "epoch": 3.6796536796536796,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00018738366468964898,
+      "loss": 0.9188,
+      "step": 2975
+    },
+    {
+      "epoch": 3.685837971552257,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00018731355650887985,
+      "loss": 0.917,
+      "step": 2980
+    },
+    {
+      "epoch": 3.692022263450835,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00018724326726453244,
+      "loss": 0.9351,
+      "step": 2985
+    },
+    {
+      "epoch": 3.6982065553494126,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00018717279710236666,
+      "loss": 0.9178,
+      "step": 2990
+    },
+    {
+      "epoch": 3.70439084724799,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0001871021461685177,
+      "loss": 0.9245,
+      "step": 2995
+    },
+    {
+      "epoch": 3.7105751391465676,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00018703131460949554,
+      "loss": 0.9176,
+      "step": 3000
+    },
+    {
+      "epoch": 3.716759431045145,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0001869603025721848,
+      "loss": 0.9334,
+      "step": 3005
+    },
+    {
+      "epoch": 3.722943722943723,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00018688911020384432,
+      "loss": 0.9213,
+      "step": 3010
+    },
+    {
+      "epoch": 3.7291280148423005,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0001868177376521069,
+      "loss": 0.9271,
+      "step": 3015
+    },
+    {
+      "epoch": 3.7353123067408784,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.000186746185064979,
+      "loss": 0.9239,
+      "step": 3020
+    },
+    {
+      "epoch": 3.741496598639456,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00018667445259084036,
+      "loss": 0.9201,
+      "step": 3025
+    },
+    {
+      "epoch": 3.7476808905380334,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 0.9236,
+      "step": 3030
+    },
+    {
+      "epoch": 3.753865182436611,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00018653044857691508,
+      "loss": 0.9212,
+      "step": 3035
+    },
+    {
+      "epoch": 3.7600494743351884,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00018645817733575193,
+      "loss": 0.9157,
+      "step": 3040
+    },
+    {
+      "epoch": 3.7662337662337664,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00018638572680482448,
+      "loss": 0.9181,
+      "step": 3045
+    },
+    {
+      "epoch": 3.772418058132344,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00018631309713437467,
+      "loss": 0.9329,
+      "step": 3050
+    },
+    {
+      "epoch": 3.7786023500309214,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00018624028847501585,
+      "loss": 0.9188,
+      "step": 3055
+    },
+    {
+      "epoch": 3.7847866419294993,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0001861673009777325,
+      "loss": 0.9255,
+      "step": 3060
+    },
+    {
+      "epoch": 3.790970933828077,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00018609413479388003,
+      "loss": 0.921,
+      "step": 3065
+    },
+    {
+      "epoch": 3.7971552257266543,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00018602079007518438,
+      "loss": 0.9226,
+      "step": 3070
+    },
+    {
+      "epoch": 3.803339517625232,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00018594726697374175,
+      "loss": 0.9174,
+      "step": 3075
+    },
+    {
+      "epoch": 3.8095238095238093,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00018587356564201817,
+      "loss": 0.9104,
+      "step": 3080
+    },
+    {
+      "epoch": 3.8157081014223873,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00018579968623284933,
+      "loss": 0.9279,
+      "step": 3085
+    },
+    {
+      "epoch": 3.8218923933209648,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001857256288994402,
+      "loss": 0.9153,
+      "step": 3090
+    },
+    {
+      "epoch": 3.8280766852195423,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00018565139379536473,
+      "loss": 0.9225,
+      "step": 3095
+    },
+    {
+      "epoch": 3.83426097711812,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00018557698107456549,
+      "loss": 0.9158,
+      "step": 3100
+    },
+    {
+      "epoch": 3.8404452690166977,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00018550239089135334,
+      "loss": 0.924,
+      "step": 3105
+    },
+    {
+      "epoch": 3.846629560915275,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00018542762340040722,
+      "loss": 0.92,
+      "step": 3110
+    },
+    {
+      "epoch": 3.8528138528138527,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00018535267875677372,
+      "loss": 0.9123,
+      "step": 3115
+    },
+    {
+      "epoch": 3.85899814471243,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00018527755711586678,
+      "loss": 0.9152,
+      "step": 3120
+    },
+    {
+      "epoch": 3.865182436611008,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00018520225863346743,
+      "loss": 0.9285,
+      "step": 3125
+    },
+    {
+      "epoch": 3.8713667285095856,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00018512678346572337,
+      "loss": 0.9203,
+      "step": 3130
+    },
+    {
+      "epoch": 3.877551020408163,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0001850511317691487,
+      "loss": 0.9214,
+      "step": 3135
+    },
+    {
+      "epoch": 3.883735312306741,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00018497530370062363,
+      "loss": 0.9185,
+      "step": 3140
+    },
+    {
+      "epoch": 3.8899196042053186,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00018489929941739407,
+      "loss": 0.917,
+      "step": 3145
+    },
+    {
+      "epoch": 3.896103896103896,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001848231190770714,
+      "loss": 0.9221,
+      "step": 3150
+    },
+    {
+      "epoch": 3.9022881880024736,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00018474676283763205,
+      "loss": 0.9148,
+      "step": 3155
+    },
+    {
+      "epoch": 3.908472479901051,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00018467023085741717,
+      "loss": 0.9258,
+      "step": 3160
+    },
+    {
+      "epoch": 3.914656771799629,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0001845935232951325,
+      "loss": 0.9044,
+      "step": 3165
+    },
+    {
+      "epoch": 3.9208410636982065,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00018451664030984773,
+      "loss": 0.9217,
+      "step": 3170
+    },
+    {
+      "epoch": 3.927025355596784,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001844395820609964,
+      "loss": 0.9186,
+      "step": 3175
+    },
+    {
+      "epoch": 3.933209647495362,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00018436234870837547,
+      "loss": 0.9087,
+      "step": 3180
+    },
+    {
+      "epoch": 3.9393939393939394,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00018428494041214507,
+      "loss": 0.9143,
+      "step": 3185
+    },
+    {
+      "epoch": 3.945578231292517,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00018420735733282807,
+      "loss": 0.923,
+      "step": 3190
+    },
+    {
+      "epoch": 3.9517625231910944,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00018412959963130975,
+      "loss": 0.9181,
+      "step": 3195
+    },
+    {
+      "epoch": 3.9579468150896724,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00018405166746883762,
+      "loss": 0.9207,
+      "step": 3200
+    },
+    {
+      "epoch": 3.96413110698825,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00018397356100702085,
+      "loss": 0.9107,
+      "step": 3205
+    },
+    {
+      "epoch": 3.9703153988868274,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00018389528040783012,
+      "loss": 0.9207,
+      "step": 3210
+    },
+    {
+      "epoch": 3.9764996907854053,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00018381682583359723,
+      "loss": 0.9226,
+      "step": 3215
+    },
+    {
+      "epoch": 3.982683982683983,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00018373819744701476,
+      "loss": 0.9065,
+      "step": 3220
+    },
+    {
+      "epoch": 3.9888682745825603,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00018365939541113566,
+      "loss": 0.9145,
+      "step": 3225
+    },
+    {
+      "epoch": 3.995052566481138,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00018358041988937305,
+      "loss": 0.925,
+      "step": 3230
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 2.475017786026001,
+      "eval_runtime": 0.5414,
+      "eval_samples_per_second": 18.47,
+      "eval_steps_per_second": 1.847,
+      "step": 3234
+    },
+    {
+      "epoch": 4.001236858379715,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00018350127104549977,
+      "loss": 0.9148,
+      "step": 3235
+    },
+    {
+      "epoch": 4.007421150278293,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00018342194904364813,
+      "loss": 0.9175,
+      "step": 3240
+    },
+    {
+      "epoch": 4.01360544217687,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00018334245404830944,
+      "loss": 0.8931,
+      "step": 3245
+    },
+    {
+      "epoch": 4.019789734075449,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00018326278622433386,
+      "loss": 0.9231,
+      "step": 3250
+    },
+    {
+      "epoch": 4.025974025974026,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00018318294573692985,
+      "loss": 0.9013,
+      "step": 3255
+    },
+    {
+      "epoch": 4.032158317872604,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00018310293275166392,
+      "loss": 0.9067,
+      "step": 3260
+    },
+    {
+      "epoch": 4.038342609771181,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00018302274743446043,
+      "loss": 0.9111,
+      "step": 3265
+    },
+    {
+      "epoch": 4.044526901669759,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00018294238995160094,
+      "loss": 0.9146,
+      "step": 3270
+    },
+    {
+      "epoch": 4.050711193568336,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0001828618604697241,
+      "loss": 0.8986,
+      "step": 3275
+    },
+    {
+      "epoch": 4.056895485466914,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00018278115915582526,
+      "loss": 0.8951,
+      "step": 3280
+    },
+    {
+      "epoch": 4.063079777365492,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00018270028617725607,
+      "loss": 0.9153,
+      "step": 3285
+    },
+    {
+      "epoch": 4.06926406926407,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001826192417017242,
+      "loss": 0.9018,
+      "step": 3290
+    },
+    {
+      "epoch": 4.075448361162647,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0001825380258972929,
+      "loss": 0.9111,
+      "step": 3295
+    },
+    {
+      "epoch": 4.081632653061225,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00018245663893238075,
+      "loss": 0.9133,
+      "step": 3300
+    },
+    {
+      "epoch": 4.087816944959802,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00018237508097576123,
+      "loss": 0.9063,
+      "step": 3305
+    },
+    {
+      "epoch": 4.09400123685838,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001822933521965625,
+      "loss": 0.8988,
+      "step": 3310
+    },
+    {
+      "epoch": 4.100185528756957,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00018221145276426683,
+      "loss": 0.9129,
+      "step": 3315
+    },
+    {
+      "epoch": 4.106369820655535,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00018212938284871047,
+      "loss": 0.9105,
+      "step": 3320
+    },
+    {
+      "epoch": 4.112554112554113,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00018204714262008316,
+      "loss": 0.8963,
+      "step": 3325
+    },
+    {
+      "epoch": 4.1187384044526905,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00018196473224892784,
+      "loss": 0.9094,
+      "step": 3330
+    },
+    {
+      "epoch": 4.124922696351268,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00018188215190614027,
+      "loss": 0.8976,
+      "step": 3335
+    },
+    {
+      "epoch": 4.1311069882498455,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001817994017629687,
+      "loss": 0.9048,
+      "step": 3340
+    },
+    {
+      "epoch": 4.137291280148423,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00018171648199101346,
+      "loss": 0.9181,
+      "step": 3345
+    },
+    {
+      "epoch": 4.1434755720470005,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00018163339276222666,
+      "loss": 0.9077,
+      "step": 3350
+    },
+    {
+      "epoch": 4.149659863945578,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00018155013424891184,
+      "loss": 0.9036,
+      "step": 3355
+    },
+    {
+      "epoch": 4.1558441558441555,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00018146670662372354,
+      "loss": 0.9122,
+      "step": 3360
+    },
+    {
+      "epoch": 4.162028447742734,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00018138311005966705,
+      "loss": 0.9029,
+      "step": 3365
+    },
+    {
+      "epoch": 4.168212739641311,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001812993447300979,
+      "loss": 0.9063,
+      "step": 3370
+    },
+    {
+      "epoch": 4.174397031539889,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00018121541080872176,
+      "loss": 0.9096,
+      "step": 3375
+    },
+    {
+      "epoch": 4.180581323438466,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00018113130846959368,
+      "loss": 0.9121,
+      "step": 3380
+    },
+    {
+      "epoch": 4.186765615337044,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00018104703788711816,
+      "loss": 0.9073,
+      "step": 3385
+    },
+    {
+      "epoch": 4.192949907235621,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0001809625992360485,
+      "loss": 0.9113,
+      "step": 3390
+    },
+    {
+      "epoch": 4.199134199134199,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00018087799269148654,
+      "loss": 0.8939,
+      "step": 3395
+    },
+    {
+      "epoch": 4.205318491032776,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00018079321842888227,
+      "loss": 0.8998,
+      "step": 3400
+    },
+    {
+      "epoch": 4.211502782931355,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00018070827662403349,
+      "loss": 0.897,
+      "step": 3405
+    },
+    {
+      "epoch": 4.217687074829932,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00018062316745308542,
+      "loss": 0.9122,
+      "step": 3410
+    },
+    {
+      "epoch": 4.22387136672851,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00018053789109253042,
+      "loss": 0.9,
+      "step": 3415
+    },
+    {
+      "epoch": 4.230055658627087,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0001804524477192075,
+      "loss": 0.9108,
+      "step": 3420
+    },
+    {
+      "epoch": 4.236239950525665,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00018036683751030194,
+      "loss": 0.9214,
+      "step": 3425
+    },
+    {
+      "epoch": 4.242424242424242,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001802810606433451,
+      "loss": 0.8972,
+      "step": 3430
+    },
+    {
+      "epoch": 4.24860853432282,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001801951172962139,
+      "loss": 0.9142,
+      "step": 3435
+    },
+    {
+      "epoch": 4.254792826221397,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00018010900764713048,
+      "loss": 0.8953,
+      "step": 3440
+    },
+    {
+      "epoch": 4.260977118119976,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001800227318746619,
+      "loss": 0.9033,
+      "step": 3445
+    },
+    {
+      "epoch": 4.267161410018553,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001799362901577196,
+      "loss": 0.896,
+      "step": 3450
+    },
+    {
+      "epoch": 4.273345701917131,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00017984968267555925,
+      "loss": 0.8991,
+      "step": 3455
+    },
+    {
+      "epoch": 4.279529993815708,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00017976290960778024,
+      "loss": 0.9085,
+      "step": 3460
+    },
+    {
+      "epoch": 4.285714285714286,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001796759711343253,
+      "loss": 0.9209,
+      "step": 3465
+    },
+    {
+      "epoch": 4.291898577612863,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001795888674354802,
+      "loss": 0.9091,
+      "step": 3470
+    },
+    {
+      "epoch": 4.298082869511441,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00017950159869187333,
+      "loss": 0.9103,
+      "step": 3475
+    },
+    {
+      "epoch": 4.304267161410019,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00017941416508447536,
+      "loss": 0.9115,
+      "step": 3480
+    },
+    {
+      "epoch": 4.3104514533085965,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001793265667945988,
+      "loss": 0.9025,
+      "step": 3485
+    },
+    {
+      "epoch": 4.316635745207174,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0001792388040038977,
+      "loss": 0.9113,
+      "step": 3490
+    },
+    {
+      "epoch": 4.3228200371057515,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001791508768943672,
+      "loss": 0.9182,
+      "step": 3495
+    },
+    {
+      "epoch": 4.329004329004329,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00017906278564834324,
+      "loss": 0.9105,
+      "step": 3500
+    },
+    {
+      "epoch": 4.3351886209029065,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00017897453044850208,
+      "loss": 0.9133,
+      "step": 3505
+    },
+    {
+      "epoch": 4.341372912801484,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00017888611147786002,
+      "loss": 0.9112,
+      "step": 3510
+    },
+    {
+      "epoch": 4.3475572047000615,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00017879752891977296,
+      "loss": 0.9023,
+      "step": 3515
+    },
+    {
+      "epoch": 4.35374149659864,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00017870878295793598,
+      "loss": 0.8993,
+      "step": 3520
+    },
+    {
+      "epoch": 4.359925788497217,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00017861987377638312,
+      "loss": 0.9052,
+      "step": 3525
+    },
+    {
+      "epoch": 4.366110080395795,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0001785308015594868,
+      "loss": 0.8988,
+      "step": 3530
+    },
+    {
+      "epoch": 4.372294372294372,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00017844156649195759,
+      "loss": 0.9054,
+      "step": 3535
+    },
+    {
+      "epoch": 4.37847866419295,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00017835216875884368,
+      "loss": 0.9039,
+      "step": 3540
+    },
+    {
+      "epoch": 4.384662956091527,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00017826260854553072,
+      "loss": 0.9082,
+      "step": 3545
+    },
+    {
+      "epoch": 4.390847247990105,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00017817288603774116,
+      "loss": 0.9025,
+      "step": 3550
+    },
+    {
+      "epoch": 4.397031539888682,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00017808300142153406,
+      "loss": 0.9022,
+      "step": 3555
+    },
+    {
+      "epoch": 4.403215831787261,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00017799295488330467,
+      "loss": 0.8981,
+      "step": 3560
+    },
+    {
+      "epoch": 4.409400123685838,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.000177902746609784,
+      "loss": 0.9073,
+      "step": 3565
+    },
+    {
+      "epoch": 4.415584415584416,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00017781237678803847,
+      "loss": 0.908,
+      "step": 3570
+    },
+    {
+      "epoch": 4.421768707482993,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00017772184560546942,
+      "loss": 0.8984,
+      "step": 3575
+    },
+    {
+      "epoch": 4.427952999381571,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00017763115324981294,
+      "loss": 0.9122,
+      "step": 3580
+    },
+    {
+      "epoch": 4.434137291280148,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00017754029990913926,
+      "loss": 0.9038,
+      "step": 3585
+    },
+    {
+      "epoch": 4.440321583178726,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00017744928577185243,
+      "loss": 0.9047,
+      "step": 3590
+    },
+    {
+      "epoch": 4.446505875077303,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00017735811102669003,
+      "loss": 0.8955,
+      "step": 3595
+    },
+    {
+      "epoch": 4.452690166975882,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00017726677586272263,
+      "loss": 0.8993,
+      "step": 3600
+    },
+    {
+      "epoch": 4.458874458874459,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0001771752804693535,
+      "loss": 0.9106,
+      "step": 3605
+    },
+    {
+      "epoch": 4.465058750773037,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00017708362503631814,
+      "loss": 0.895,
+      "step": 3610
+    },
+    {
+      "epoch": 4.471243042671614,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00017699180975368396,
+      "loss": 0.9059,
+      "step": 3615
+    },
+    {
+      "epoch": 4.477427334570192,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00017689983481184989,
+      "loss": 0.9031,
+      "step": 3620
+    },
+    {
+      "epoch": 4.483611626468769,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001768077004015458,
+      "loss": 0.9148,
+      "step": 3625
+    },
+    {
+      "epoch": 4.489795918367347,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00017671540671383243,
+      "loss": 0.8983,
+      "step": 3630
+    },
+    {
+      "epoch": 4.495980210265925,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00017662295394010072,
+      "loss": 0.9008,
+      "step": 3635
+    },
+    {
+      "epoch": 4.5021645021645025,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00017653034227207152,
+      "loss": 0.9089,
+      "step": 3640
+    },
+    {
+      "epoch": 4.50834879406308,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00017643757190179523,
+      "loss": 0.9176,
+      "step": 3645
+    },
+    {
+      "epoch": 4.5145330859616575,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00017634464302165124,
+      "loss": 0.9026,
+      "step": 3650
+    },
+    {
+      "epoch": 4.520717377860235,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00017625155582434777,
+      "loss": 0.9066,
+      "step": 3655
+    },
+    {
+      "epoch": 4.5269016697588125,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.0001761583105029213,
+      "loss": 0.9125,
+      "step": 3660
+    },
+    {
+      "epoch": 4.53308596165739,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00017606490725073615,
+      "loss": 0.9058,
+      "step": 3665
+    },
+    {
+      "epoch": 4.5392702535559675,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00017597134626148427,
+      "loss": 0.8999,
+      "step": 3670
+    },
+    {
+      "epoch": 4.545454545454545,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00017587762772918467,
+      "loss": 0.9047,
+      "step": 3675
+    },
+    {
+      "epoch": 4.551638837353123,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001757837518481829,
+      "loss": 0.902,
+      "step": 3680
+    },
+    {
+      "epoch": 4.557823129251701,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00017568971881315104,
+      "loss": 0.9083,
+      "step": 3685
+    },
+    {
+      "epoch": 4.564007421150278,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00017559552881908695,
+      "loss": 0.9084,
+      "step": 3690
+    },
+    {
+      "epoch": 4.570191713048856,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00017550118206131402,
+      "loss": 0.905,
+      "step": 3695
+    },
+    {
+      "epoch": 4.576376004947433,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00017540667873548063,
+      "loss": 0.8922,
+      "step": 3700
+    },
+    {
+      "epoch": 4.582560296846011,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00017531201903755994,
+      "loss": 0.9123,
+      "step": 3705
+    },
+    {
+      "epoch": 4.588744588744589,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00017521720316384935,
+      "loss": 0.895,
+      "step": 3710
+    },
+    {
+      "epoch": 4.594928880643167,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00017512223131097007,
+      "loss": 0.9068,
+      "step": 3715
+    },
+    {
+      "epoch": 4.601113172541744,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00017502710367586687,
+      "loss": 0.8987,
+      "step": 3720
+    },
+    {
+      "epoch": 4.607297464440322,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0001749318204558075,
+      "loss": 0.9054,
+      "step": 3725
+    },
+    {
+      "epoch": 4.613481756338899,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00017483638184838239,
+      "loss": 0.9075,
+      "step": 3730
+    },
+    {
+      "epoch": 4.619666048237477,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001747407880515041,
+      "loss": 0.9086,
+      "step": 3735
+    },
+    {
+      "epoch": 4.625850340136054,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001746450392634071,
+      "loss": 0.9125,
+      "step": 3740
+    },
+    {
+      "epoch": 4.632034632034632,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001745491356826473,
+      "loss": 0.9123,
+      "step": 3745
+    },
+    {
+      "epoch": 4.638218923933209,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001744530775081015,
+      "loss": 0.9056,
+      "step": 3750
+    },
+    {
+      "epoch": 4.644403215831788,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001743568649389672,
+      "loss": 0.9066,
+      "step": 3755
+    },
+    {
+      "epoch": 4.650587507730365,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00017426049817476197,
+      "loss": 0.9052,
+      "step": 3760
+    },
+    {
+      "epoch": 4.656771799628943,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00017416397741532315,
+      "loss": 0.9075,
+      "step": 3765
+    },
+    {
+      "epoch": 4.66295609152752,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00017406730286080753,
+      "loss": 0.8945,
+      "step": 3770
+    },
+    {
+      "epoch": 4.669140383426098,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00017397047471169063,
+      "loss": 0.9086,
+      "step": 3775
+    },
+    {
+      "epoch": 4.675324675324675,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00017387349316876666,
+      "loss": 0.9103,
+      "step": 3780
+    },
+    {
+      "epoch": 4.681508967223253,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0001737763584331479,
+      "loss": 0.9139,
+      "step": 3785
+    },
+    {
+      "epoch": 4.687693259121831,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00017367907070626424,
+      "loss": 0.8962,
+      "step": 3790
+    },
+    {
+      "epoch": 4.6938775510204085,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00017358163018986282,
+      "loss": 0.9195,
+      "step": 3795
+    },
+    {
+      "epoch": 4.700061842918986,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00017348403708600772,
+      "loss": 0.9006,
+      "step": 3800
+    },
+    {
+      "epoch": 4.7062461348175635,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00017338629159707936,
+      "loss": 0.9051,
+      "step": 3805
+    },
+    {
+      "epoch": 4.712430426716141,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001732883939257742,
+      "loss": 0.9039,
+      "step": 3810
+    },
+    {
+      "epoch": 4.7186147186147185,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.0001731903442751043,
+      "loss": 0.8953,
+      "step": 3815
+    },
+    {
+      "epoch": 4.724799010513296,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00017309214284839678,
+      "loss": 0.9031,
+      "step": 3820
+    },
+    {
+      "epoch": 4.7309833024118735,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00017299378984929366,
+      "loss": 0.8986,
+      "step": 3825
+    },
+    {
+      "epoch": 4.737167594310451,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00017289528548175114,
+      "loss": 0.8992,
+      "step": 3830
+    },
+    {
+      "epoch": 4.743351886209029,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.0001727966299500394,
+      "loss": 0.9061,
+      "step": 3835
+    },
+    {
+      "epoch": 4.749536178107607,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00017269782345874203,
+      "loss": 0.8924,
+      "step": 3840
+    },
+    {
+      "epoch": 4.755720470006184,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00017259886621275573,
+      "loss": 0.8993,
+      "step": 3845
+    },
+    {
+      "epoch": 4.761904761904762,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001724997584172898,
+      "loss": 0.8885,
+      "step": 3850
+    },
+    {
+      "epoch": 4.768089053803339,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0001724005002778657,
+      "loss": 0.906,
+      "step": 3855
+    },
+    {
+      "epoch": 4.774273345701917,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00017230109200031668,
+      "loss": 0.9108,
+      "step": 3860
+    },
+    {
+      "epoch": 4.780457637600494,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00017220153379078737,
+      "loss": 0.9037,
+      "step": 3865
+    },
+    {
+      "epoch": 4.786641929499073,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00017210182585573327,
+      "loss": 0.904,
+      "step": 3870
+    },
+    {
+      "epoch": 4.79282622139765,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00017200196840192042,
+      "loss": 0.9066,
+      "step": 3875
+    },
+    {
+      "epoch": 4.799010513296228,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00017190196163642483,
+      "loss": 0.8982,
+      "step": 3880
+    },
+    {
+      "epoch": 4.805194805194805,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00017180180576663228,
+      "loss": 0.9065,
+      "step": 3885
+    },
+    {
+      "epoch": 4.811379097093383,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001717015010002376,
+      "loss": 0.8865,
+      "step": 3890
+    },
+    {
+      "epoch": 4.81756338899196,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00017160104754524445,
+      "loss": 0.9034,
+      "step": 3895
+    },
+    {
+      "epoch": 4.823747680890538,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00017150044560996488,
+      "loss": 0.8984,
+      "step": 3900
+    },
+    {
+      "epoch": 4.829931972789115,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00017139969540301878,
+      "loss": 0.9073,
+      "step": 3905
+    },
+    {
+      "epoch": 4.836116264687694,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00017129879713333356,
+      "loss": 0.9078,
+      "step": 3910
+    },
+    {
+      "epoch": 4.842300556586271,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00017119775101014358,
+      "loss": 0.9127,
+      "step": 3915
+    },
+    {
+      "epoch": 4.848484848484849,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00017109655724298995,
+      "loss": 0.9068,
+      "step": 3920
+    },
+    {
+      "epoch": 4.854669140383426,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00017099521604171982,
+      "loss": 0.8936,
+      "step": 3925
+    },
+    {
+      "epoch": 4.860853432282004,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00017089372761648616,
+      "loss": 0.9042,
+      "step": 3930
+    },
+    {
+      "epoch": 4.867037724180581,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001707920921777472,
+      "loss": 0.9065,
+      "step": 3935
+    },
+    {
+      "epoch": 4.873222016079159,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00017069030993626603,
+      "loss": 0.8962,
+      "step": 3940
+    },
+    {
+      "epoch": 4.879406307977737,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00017058838110311017,
+      "loss": 0.9011,
+      "step": 3945
+    },
+    {
+      "epoch": 4.8855905998763145,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00017048630588965117,
+      "loss": 0.8989,
+      "step": 3950
+    },
+    {
+      "epoch": 4.891774891774892,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001703840845075641,
+      "loss": 0.9101,
+      "step": 3955
+    },
+    {
+      "epoch": 4.8979591836734695,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00017028171716882714,
+      "loss": 0.8964,
+      "step": 3960
+    },
+    {
+      "epoch": 4.904143475572047,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00017017920408572115,
+      "loss": 0.9077,
+      "step": 3965
+    },
+    {
+      "epoch": 4.9103277674706245,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00017007654547082922,
+      "loss": 0.9015,
+      "step": 3970
+    },
+    {
+      "epoch": 4.916512059369202,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00016997374153703625,
+      "loss": 0.8991,
+      "step": 3975
+    },
+    {
+      "epoch": 4.9226963512677795,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00016987079249752843,
+      "loss": 0.9045,
+      "step": 3980
+    },
+    {
+      "epoch": 4.928880643166357,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0001697676985657929,
+      "loss": 0.9004,
+      "step": 3985
+    },
+    {
+      "epoch": 4.935064935064935,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00016966445995561727,
+      "loss": 0.8999,
+      "step": 3990
+    },
+    {
+      "epoch": 4.941249226963513,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00016956107688108923,
+      "loss": 0.9044,
+      "step": 3995
+    },
+    {
+      "epoch": 4.94743351886209,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00016945754955659595,
+      "loss": 0.9037,
+      "step": 4000
+    },
+    {
+      "epoch": 4.953617810760668,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00016935387819682376,
+      "loss": 0.9038,
+      "step": 4005
+    },
+    {
+      "epoch": 4.959802102659245,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00016925006301675763,
+      "loss": 0.8995,
+      "step": 4010
+    },
+    {
+      "epoch": 4.965986394557823,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00016914610423168094,
+      "loss": 0.9127,
+      "step": 4015
+    },
+    {
+      "epoch": 4.9721706864564,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001690420020571747,
+      "loss": 0.8946,
+      "step": 4020
+    },
+    {
+      "epoch": 4.978354978354979,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00016893775670911732,
+      "loss": 0.9108,
+      "step": 4025
+    },
+    {
+      "epoch": 4.984539270253556,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00016883336840368412,
+      "loss": 0.9083,
+      "step": 4030
+    },
+    {
+      "epoch": 4.990723562152134,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001687288373573469,
+      "loss": 0.9043,
+      "step": 4035
+    },
+    {
+      "epoch": 4.996907854050711,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0001686241637868734,
+      "loss": 0.9099,
+      "step": 4040
+    },
+    {
+      "epoch": 4.999381570810142,
+      "eval_loss": 2.483851194381714,
+      "eval_runtime": 0.6403,
+      "eval_samples_per_second": 15.619,
+      "eval_steps_per_second": 1.562,
+      "step": 4042
+    },
+    {
+      "epoch": 5.003092145949289,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00016851934790932692,
+      "loss": 0.903,
+      "step": 4045
+    },
+    {
+      "epoch": 5.009276437847866,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00016841438994206595,
+      "loss": 0.8894,
+      "step": 4050
+    },
+    {
+      "epoch": 5.015460729746444,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001683092901027436,
+      "loss": 0.9058,
+      "step": 4055
+    },
+    {
+      "epoch": 5.021645021645021,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001682040486093071,
+      "loss": 0.8982,
+      "step": 4060
+    },
+    {
+      "epoch": 5.0278293135436,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.0001680986656799975,
+      "loss": 0.8948,
+      "step": 4065
+    },
+    {
+      "epoch": 5.034013605442177,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00016799314153334916,
+      "loss": 0.8902,
+      "step": 4070
+    },
+    {
+      "epoch": 5.040197897340755,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00016788747638818926,
+      "loss": 0.8845,
+      "step": 4075
+    },
+    {
+      "epoch": 5.046382189239332,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00016778167046363734,
+      "loss": 0.8875,
+      "step": 4080
+    },
+    {
+      "epoch": 5.05256648113791,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001676757239791049,
+      "loss": 0.8976,
+      "step": 4085
+    },
+    {
+      "epoch": 5.058750773036487,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00016756963715429502,
+      "loss": 0.887,
+      "step": 4090
+    },
+    {
+      "epoch": 5.064935064935065,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00016746341020920167,
+      "loss": 0.8852,
+      "step": 4095
+    },
+    {
+      "epoch": 5.071119356833642,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00016735704336410943,
+      "loss": 0.8966,
+      "step": 4100
+    },
+    {
+      "epoch": 5.0773036487322205,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.000167250536839593,
+      "loss": 0.9055,
+      "step": 4105
+    },
+    {
+      "epoch": 5.083487940630798,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0001671438908565167,
+      "loss": 0.8904,
+      "step": 4110
+    },
+    {
+      "epoch": 5.0896722325293755,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00016703710563603416,
+      "loss": 0.8829,
+      "step": 4115
+    },
+    {
+      "epoch": 5.095856524427953,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00016693018139958763,
+      "loss": 0.8835,
+      "step": 4120
+    },
+    {
+      "epoch": 5.1020408163265305,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00016682311836890766,
+      "loss": 0.8927,
+      "step": 4125
+    },
+    {
+      "epoch": 5.108225108225108,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00016671591676601272,
+      "loss": 0.8803,
+      "step": 4130
+    },
+    {
+      "epoch": 5.1144094001236855,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001666085768132085,
+      "loss": 0.8923,
+      "step": 4135
+    },
+    {
+      "epoch": 5.120593692022264,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00016650109873308765,
+      "loss": 0.8963,
+      "step": 4140
+    },
+    {
+      "epoch": 5.126777983920841,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00016639348274852925,
+      "loss": 0.8986,
+      "step": 4145
+    },
+    {
+      "epoch": 5.132962275819419,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00016628572908269841,
+      "loss": 0.8841,
+      "step": 4150
+    },
+    {
+      "epoch": 5.139146567717996,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00016617783795904565,
+      "loss": 0.8892,
+      "step": 4155
+    },
+    {
+      "epoch": 5.145330859616574,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00016606980960130665,
+      "loss": 0.892,
+      "step": 4160
+    },
+    {
+      "epoch": 5.151515151515151,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00016596164423350157,
+      "loss": 0.8884,
+      "step": 4165
+    },
+    {
+      "epoch": 5.157699443413729,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00016585334207993476,
+      "loss": 0.9061,
+      "step": 4170
+    },
+    {
+      "epoch": 5.163883735312306,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00016574490336519418,
+      "loss": 0.8863,
+      "step": 4175
+    },
+    {
+      "epoch": 5.170068027210885,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00016563632831415102,
+      "loss": 0.8963,
+      "step": 4180
+    },
+    {
+      "epoch": 5.176252319109462,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00016552761715195918,
+      "loss": 0.8934,
+      "step": 4185
+    },
+    {
+      "epoch": 5.18243661100804,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00016541877010405477,
+      "loss": 0.9016,
+      "step": 4190
+    },
+    {
+      "epoch": 5.188620902906617,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00016530978739615578,
+      "loss": 0.8868,
+      "step": 4195
+    },
+    {
+      "epoch": 5.194805194805195,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00016520066925426144,
+      "loss": 0.8907,
+      "step": 4200
+    },
+    {
+      "epoch": 5.200989486703772,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00016509141590465189,
+      "loss": 0.8899,
+      "step": 4205
+    },
+    {
+      "epoch": 5.20717377860235,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00016498202757388758,
+      "loss": 0.89,
+      "step": 4210
+    },
+    {
+      "epoch": 5.213358070500927,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00016487250448880893,
+      "loss": 0.8961,
+      "step": 4215
+    },
+    {
+      "epoch": 5.219542362399506,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001647628468765358,
+      "loss": 0.8962,
+      "step": 4220
+    },
+    {
+      "epoch": 5.225726654298083,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00016465305496446702,
+      "loss": 0.8974,
+      "step": 4225
+    },
+    {
+      "epoch": 5.231910946196661,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0001645431289802799,
+      "loss": 0.8987,
+      "step": 4230
+    },
+    {
+      "epoch": 5.238095238095238,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00016443306915192975,
+      "loss": 0.8891,
+      "step": 4235
+    },
+    {
+      "epoch": 5.244279529993816,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00016432287570764952,
+      "loss": 0.8979,
+      "step": 4240
+    },
+    {
+      "epoch": 5.250463821892393,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00016421254887594917,
+      "loss": 0.8971,
+      "step": 4245
+    },
+    {
+      "epoch": 5.256648113790971,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0001641020888856153,
+      "loss": 0.8943,
+      "step": 4250
+    },
+    {
+      "epoch": 5.262832405689548,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00016399149596571064,
+      "loss": 0.901,
+      "step": 4255
+    },
+    {
+      "epoch": 5.2690166975881265,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00016388077034557355,
+      "loss": 0.874,
+      "step": 4260
+    },
+    {
+      "epoch": 5.275200989486704,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001637699122548176,
+      "loss": 0.894,
+      "step": 4265
+    },
+    {
+      "epoch": 5.2813852813852815,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0001636589219233311,
+      "loss": 0.8834,
+      "step": 4270
+    },
+    {
+      "epoch": 5.287569573283859,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0001635477995812765,
+      "loss": 0.9,
+      "step": 4275
+    },
+    {
+      "epoch": 5.2937538651824365,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00016343654545909007,
+      "loss": 0.897,
+      "step": 4280
+    },
+    {
+      "epoch": 5.299938157081014,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00016332515978748134,
+      "loss": 0.8738,
+      "step": 4285
+    },
+    {
+      "epoch": 5.3061224489795915,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00016321364279743266,
+      "loss": 0.8839,
+      "step": 4290
+    },
+    {
+      "epoch": 5.312306740878169,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00016310199472019865,
+      "loss": 0.8876,
+      "step": 4295
+    },
+    {
+      "epoch": 5.318491032776747,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00016299021578730579,
+      "loss": 0.8807,
+      "step": 4300
+    },
+    {
+      "epoch": 5.324675324675325,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00016287830623055188,
+      "loss": 0.8866,
+      "step": 4305
+    },
+    {
+      "epoch": 5.330859616573902,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00016276626628200568,
+      "loss": 0.8949,
+      "step": 4310
+    },
+    {
+      "epoch": 5.33704390847248,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00016265409617400632,
+      "loss": 0.8963,
+      "step": 4315
+    },
+    {
+      "epoch": 5.343228200371057,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00016254179613916278,
+      "loss": 0.8974,
+      "step": 4320
+    },
+    {
+      "epoch": 5.349412492269635,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00016242936641035357,
+      "loss": 0.8935,
+      "step": 4325
+    },
+    {
+      "epoch": 5.355596784168212,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.000162316807220726,
+      "loss": 0.8888,
+      "step": 4330
+    },
+    {
+      "epoch": 5.361781076066791,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00016220411880369601,
+      "loss": 0.8833,
+      "step": 4335
+    },
+    {
+      "epoch": 5.367965367965368,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00016209130139294744,
+      "loss": 0.8855,
+      "step": 4340
+    },
+    {
+      "epoch": 5.374149659863946,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00016197835522243162,
+      "loss": 0.8867,
+      "step": 4345
+    },
+    {
+      "epoch": 5.380333951762523,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00016186528052636692,
+      "loss": 0.8878,
+      "step": 4350
+    },
+    {
+      "epoch": 5.386518243661101,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00016175207753923822,
+      "loss": 0.8938,
+      "step": 4355
+    },
+    {
+      "epoch": 5.392702535559678,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00016163874649579647,
+      "loss": 0.8857,
+      "step": 4360
+    },
+    {
+      "epoch": 5.398886827458256,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001615252876310581,
+      "loss": 0.889,
+      "step": 4365
+    },
+    {
+      "epoch": 5.405071119356833,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00016141170118030463,
+      "loss": 0.8858,
+      "step": 4370
+    },
+    {
+      "epoch": 5.411255411255412,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00016129798737908225,
+      "loss": 0.8901,
+      "step": 4375
+    },
+    {
+      "epoch": 5.417439703153989,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0001611841464632011,
+      "loss": 0.8935,
+      "step": 4380
+    },
+    {
+      "epoch": 5.423623995052567,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00016107017866873505,
+      "loss": 0.8871,
+      "step": 4385
+    },
+    {
+      "epoch": 5.429808286951144,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00016095608423202098,
+      "loss": 0.8979,
+      "step": 4390
+    },
+    {
+      "epoch": 5.435992578849722,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00016084186338965843,
+      "loss": 0.8893,
+      "step": 4395
+    },
+    {
+      "epoch": 5.442176870748299,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00016072751637850904,
+      "loss": 0.8928,
+      "step": 4400
+    },
+    {
+      "epoch": 5.448361162646877,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00016061304343569614,
+      "loss": 0.8916,
+      "step": 4405
+    },
+    {
+      "epoch": 5.454545454545454,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00016049844479860422,
+      "loss": 0.8866,
+      "step": 4410
+    },
+    {
+      "epoch": 5.4607297464440325,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00016038372070487832,
+      "loss": 0.9007,
+      "step": 4415
+    },
+    {
+      "epoch": 5.46691403834261,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00016026887139242372,
+      "loss": 0.8922,
+      "step": 4420
+    },
+    {
+      "epoch": 5.4730983302411875,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00016015389709940538,
+      "loss": 0.8912,
+      "step": 4425
+    },
+    {
+      "epoch": 5.479282622139765,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001600387980642474,
+      "loss": 0.8891,
+      "step": 4430
+    },
+    {
+      "epoch": 5.4854669140383425,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00015992357452563255,
+      "loss": 0.8973,
+      "step": 4435
+    },
+    {
+      "epoch": 5.49165120593692,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001598082267225018,
+      "loss": 0.9042,
+      "step": 4440
+    },
+    {
+      "epoch": 5.4978354978354975,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00015969275489405383,
+      "loss": 0.8975,
+      "step": 4445
+    },
+    {
+      "epoch": 5.504019789734075,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0001595771592797445,
+      "loss": 0.8788,
+      "step": 4450
+    },
+    {
+      "epoch": 5.510204081632653,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00015946144011928638,
+      "loss": 0.8986,
+      "step": 4455
+    },
+    {
+      "epoch": 5.516388373531231,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0001593455976526482,
+      "loss": 0.8966,
+      "step": 4460
+    },
+    {
+      "epoch": 5.522572665429808,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00015922963212005442,
+      "loss": 0.8858,
+      "step": 4465
+    },
+    {
+      "epoch": 5.528756957328386,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001591135437619847,
+      "loss": 0.8931,
+      "step": 4470
+    },
+    {
+      "epoch": 5.534941249226963,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001589973328191734,
+      "loss": 0.8913,
+      "step": 4475
+    },
+    {
+      "epoch": 5.541125541125541,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00015888099953260905,
+      "loss": 0.8913,
+      "step": 4480
+    },
+    {
+      "epoch": 5.547309833024118,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.000158764544143534,
+      "loss": 0.8886,
+      "step": 4485
+    },
+    {
+      "epoch": 5.553494124922697,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0001586479668934437,
+      "loss": 0.9026,
+      "step": 4490
+    },
+    {
+      "epoch": 5.559678416821274,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00015853126802408633,
+      "loss": 0.8857,
+      "step": 4495
+    },
+    {
+      "epoch": 5.565862708719852,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001584144477774623,
+      "loss": 0.8948,
+      "step": 4500
+    },
+    {
+      "epoch": 5.572047000618429,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001582975063958237,
+      "loss": 0.8877,
+      "step": 4505
+    },
+    {
+      "epoch": 5.578231292517007,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0001581804441216738,
+      "loss": 0.8958,
+      "step": 4510
+    },
+    {
+      "epoch": 5.584415584415584,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00015806326119776663,
+      "loss": 0.8914,
+      "step": 4515
+    },
+    {
+      "epoch": 5.590599876314162,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00015794595786710632,
+      "loss": 0.8937,
+      "step": 4520
+    },
+    {
+      "epoch": 5.596784168212739,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0001578285343729468,
+      "loss": 0.8957,
+      "step": 4525
+    },
+    {
+      "epoch": 5.602968460111317,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00015771099095879108,
+      "loss": 0.8916,
+      "step": 4530
+    },
+    {
+      "epoch": 5.609152752009895,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00015759332786839092,
+      "loss": 0.891,
+      "step": 4535
+    },
+    {
+      "epoch": 5.615337043908473,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00015747554534574626,
+      "loss": 0.8814,
+      "step": 4540
+    },
+    {
+      "epoch": 5.62152133580705,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001573576436351046,
+      "loss": 0.8883,
+      "step": 4545
+    },
+    {
+      "epoch": 5.627705627705628,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001572396229809608,
+      "loss": 0.8992,
+      "step": 4550
+    },
+    {
+      "epoch": 5.633889919604205,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00015712148362805617,
+      "loss": 0.883,
+      "step": 4555
+    },
+    {
+      "epoch": 5.640074211502783,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00015700322582137827,
+      "loss": 0.8908,
+      "step": 4560
+    },
+    {
+      "epoch": 5.646258503401361,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00015688484980616032,
+      "loss": 0.8953,
+      "step": 4565
+    },
+    {
+      "epoch": 5.6524427952999385,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001567663558278806,
+      "loss": 0.8968,
+      "step": 4570
+    },
+    {
+      "epoch": 5.658627087198516,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001566477441322621,
+      "loss": 0.8788,
+      "step": 4575
+    },
+    {
+      "epoch": 5.6648113790970935,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001565290149652718,
+      "loss": 0.8847,
+      "step": 4580
+    },
+    {
+      "epoch": 5.670995670995671,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00015641016857312044,
+      "loss": 0.8916,
+      "step": 4585
+    },
+    {
+      "epoch": 5.6771799628942485,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00015629120520226165,
+      "loss": 0.8942,
+      "step": 4590
+    },
+    {
+      "epoch": 5.683364254792826,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00015617212509939186,
+      "loss": 0.893,
+      "step": 4595
+    },
+    {
+      "epoch": 5.6895485466914035,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00015605292851144942,
+      "loss": 0.8956,
+      "step": 4600
+    },
+    {
+      "epoch": 5.695732838589981,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00015593361568561428,
+      "loss": 0.8978,
+      "step": 4605
+    },
+    {
+      "epoch": 5.701917130488559,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00015581418686930743,
+      "loss": 0.8867,
+      "step": 4610
+    },
+    {
+      "epoch": 5.708101422387137,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001556946423101905,
+      "loss": 0.8914,
+      "step": 4615
+    },
+    {
+      "epoch": 5.714285714285714,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00015557498225616487,
+      "loss": 0.892,
+      "step": 4620
+    },
+    {
+      "epoch": 5.720470006184292,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00015545520695537169,
+      "loss": 0.8906,
+      "step": 4625
+    },
+    {
+      "epoch": 5.726654298082869,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00015533531665619098,
+      "loss": 0.8893,
+      "step": 4630
+    },
+    {
+      "epoch": 5.732838589981447,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00015521531160724126,
+      "loss": 0.8866,
+      "step": 4635
+    },
+    {
+      "epoch": 5.739022881880024,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00015509519205737896,
+      "loss": 0.907,
+      "step": 4640
+    },
+    {
+      "epoch": 5.745207173778603,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00015497495825569807,
+      "loss": 0.8901,
+      "step": 4645
+    },
+    {
+      "epoch": 5.75139146567718,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.0001548546104515294,
+      "loss": 0.8843,
+      "step": 4650
+    },
+    {
+      "epoch": 5.757575757575758,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00015473414889444014,
+      "loss": 0.8911,
+      "step": 4655
+    },
+    {
+      "epoch": 5.763760049474335,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0001546135738342335,
+      "loss": 0.8918,
+      "step": 4660
+    },
+    {
+      "epoch": 5.769944341372913,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00015449288552094796,
+      "loss": 0.8893,
+      "step": 4665
+    },
+    {
+      "epoch": 5.77612863327149,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001543720842048569,
+      "loss": 0.8921,
+      "step": 4670
+    },
+    {
+      "epoch": 5.782312925170068,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.000154251170136468,
+      "loss": 0.8927,
+      "step": 4675
+    },
+    {
+      "epoch": 5.788497217068645,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00015413014356652286,
+      "loss": 0.8886,
+      "step": 4680
+    },
+    {
+      "epoch": 5.794681508967223,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001540090047459962,
+      "loss": 0.8811,
+      "step": 4685
+    },
+    {
+      "epoch": 5.800865800865801,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00015388775392609564,
+      "loss": 0.8938,
+      "step": 4690
+    },
+    {
+      "epoch": 5.807050092764379,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00015376639135826107,
+      "loss": 0.8838,
+      "step": 4695
+    },
+    {
+      "epoch": 5.813234384662956,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.000153644917294164,
+      "loss": 0.8913,
+      "step": 4700
+    },
+    {
+      "epoch": 5.819418676561534,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001535233319857073,
+      "loss": 0.8946,
+      "step": 4705
+    },
+    {
+      "epoch": 5.825602968460111,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001534016356850244,
+      "loss": 0.8948,
+      "step": 4710
+    },
+    {
+      "epoch": 5.831787260358689,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00015327982864447894,
+      "loss": 0.8854,
+      "step": 4715
+    },
+    {
+      "epoch": 5.837971552257266,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00015315791111666425,
+      "loss": 0.8961,
+      "step": 4720
+    },
+    {
+      "epoch": 5.8441558441558445,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00015303588335440274,
+      "loss": 0.8981,
+      "step": 4725
+    },
+    {
+      "epoch": 5.850340136054422,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00015291374561074536,
+      "loss": 0.8957,
+      "step": 4730
+    },
+    {
+      "epoch": 5.8565244279529995,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00015279149813897126,
+      "loss": 0.8876,
+      "step": 4735
+    },
+    {
+      "epoch": 5.862708719851577,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.000152669141192587,
+      "loss": 0.8829,
+      "step": 4740
+    },
+    {
+      "epoch": 5.8688930117501545,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00015254667502532626,
+      "loss": 0.8997,
+      "step": 4745
+    },
+    {
+      "epoch": 5.875077303648732,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00015242409989114916,
+      "loss": 0.8926,
+      "step": 4750
+    },
+    {
+      "epoch": 5.8812615955473095,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00015230141604424181,
+      "loss": 0.8902,
+      "step": 4755
+    },
+    {
+      "epoch": 5.887445887445887,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00015217862373901575,
+      "loss": 0.8962,
+      "step": 4760
+    },
+    {
+      "epoch": 5.893630179344465,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001520557232301074,
+      "loss": 0.8931,
+      "step": 4765
+    },
+    {
+      "epoch": 5.899814471243043,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001519327147723776,
+      "loss": 0.8883,
+      "step": 4770
+    },
+    {
+      "epoch": 5.90599876314162,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001518095986209111,
+      "loss": 0.8953,
+      "step": 4775
+    },
+    {
+      "epoch": 5.912183055040198,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00015168637503101584,
+      "loss": 0.9026,
+      "step": 4780
+    },
+    {
+      "epoch": 5.918367346938775,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00015156304425822267,
+      "loss": 0.8927,
+      "step": 4785
+    },
+    {
+      "epoch": 5.924551638837353,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00015143960655828468,
+      "loss": 0.8968,
+      "step": 4790
+    },
+    {
+      "epoch": 5.93073593073593,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00015131606218717666,
+      "loss": 0.9022,
+      "step": 4795
+    },
+    {
+      "epoch": 5.936920222634509,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00015119241140109467,
+      "loss": 0.8945,
+      "step": 4800
+    },
+    {
+      "epoch": 5.943104514533086,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00015106865445645536,
+      "loss": 0.8783,
+      "step": 4805
+    },
+    {
+      "epoch": 5.949288806431664,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001509447916098956,
+      "loss": 0.8833,
+      "step": 4810
+    },
+    {
+      "epoch": 5.955473098330241,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00015082082311827183,
+      "loss": 0.8861,
+      "step": 4815
+    },
+    {
+      "epoch": 5.961657390228819,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001506967492386596,
+      "loss": 0.8853,
+      "step": 4820
+    },
+    {
+      "epoch": 5.967841682127396,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00015057257022835295,
+      "loss": 0.8897,
+      "step": 4825
+    },
+    {
+      "epoch": 5.974025974025974,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.000150448286344864,
+      "loss": 0.8951,
+      "step": 4830
+    },
+    {
+      "epoch": 5.980210265924551,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00015032389784592226,
+      "loss": 0.8823,
+      "step": 4835
+    },
+    {
+      "epoch": 5.986394557823129,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00015019940498947428,
+      "loss": 0.8795,
+      "step": 4840
+    },
+    {
+      "epoch": 5.992578849721707,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.000150074808033683,
+      "loss": 0.9034,
+      "step": 4845
+    },
+    {
+      "epoch": 5.998763141620285,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00014995010723692714,
+      "loss": 0.8939,
+      "step": 4850
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 2.4935665130615234,
+      "eval_runtime": 0.5407,
+      "eval_samples_per_second": 18.496,
+      "eval_steps_per_second": 1.85,
+      "step": 4851
+    },
+    {
+      "epoch": 6.004947433518862,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00014982530285780082,
+      "loss": 0.8858,
+      "step": 4855
+    },
+    {
+      "epoch": 6.01113172541744,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00014970039515511304,
+      "loss": 0.8882,
+      "step": 4860
+    },
+    {
+      "epoch": 6.017316017316017,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0001495753843878869,
+      "loss": 0.8914,
+      "step": 4865
+    },
+    {
+      "epoch": 6.023500309214595,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00014945027081535937,
+      "loss": 0.8648,
+      "step": 4870
+    },
+    {
+      "epoch": 6.029684601113172,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00014932505469698052,
+      "loss": 0.87,
+      "step": 4875
+    },
+    {
+      "epoch": 6.035868893011751,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00014919973629241314,
+      "loss": 0.8876,
+      "step": 4880
+    },
+    {
+      "epoch": 6.042053184910328,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00014907431586153201,
+      "loss": 0.8852,
+      "step": 4885
+    },
+    {
+      "epoch": 6.0482374768089056,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001489487936644237,
+      "loss": 0.8752,
+      "step": 4890
+    },
+    {
+      "epoch": 6.054421768707483,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00014882316996138556,
+      "loss": 0.8855,
+      "step": 4895
+    },
+    {
+      "epoch": 6.0606060606060606,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00014869744501292561,
+      "loss": 0.8878,
+      "step": 4900
+    },
+    {
+      "epoch": 6.066790352504638,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00014857161907976183,
+      "loss": 0.8818,
+      "step": 4905
+    },
+    {
+      "epoch": 6.0729746444032155,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00014844569242282148,
+      "loss": 0.8859,
+      "step": 4910
+    },
+    {
+      "epoch": 6.079158936301793,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00014831966530324078,
+      "loss": 0.876,
+      "step": 4915
+    },
+    {
+      "epoch": 6.085343228200371,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00014819353798236427,
+      "loss": 0.8756,
+      "step": 4920
+    },
+    {
+      "epoch": 6.091527520098949,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00014806731072174428,
+      "loss": 0.8867,
+      "step": 4925
+    },
+    {
+      "epoch": 6.097711811997526,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001479409837831404,
+      "loss": 0.8736,
+      "step": 4930
+    },
+    {
+      "epoch": 6.103896103896104,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00014781455742851892,
+      "loss": 0.878,
+      "step": 4935
+    },
+    {
+      "epoch": 6.110080395794681,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00014768803192005223,
+      "loss": 0.8853,
+      "step": 4940
+    },
+    {
+      "epoch": 6.116264687693259,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00014756140752011842,
+      "loss": 0.8778,
+      "step": 4945
+    },
+    {
+      "epoch": 6.122448979591836,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00014743468449130063,
+      "loss": 0.891,
+      "step": 4950
+    },
+    {
+      "epoch": 6.128633271490414,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00014730786309638652,
+      "loss": 0.8794,
+      "step": 4955
+    },
+    {
+      "epoch": 6.134817563388992,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00014718094359836772,
+      "loss": 0.8913,
+      "step": 4960
+    },
+    {
+      "epoch": 6.14100185528757,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001470539262604393,
+      "loss": 0.8825,
+      "step": 4965
+    },
+    {
+      "epoch": 6.147186147186147,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00014692681134599925,
+      "loss": 0.8865,
+      "step": 4970
+    },
+    {
+      "epoch": 6.153370439084725,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00014679959911864784,
+      "loss": 0.8829,
+      "step": 4975
+    },
+    {
+      "epoch": 6.159554730983302,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001466722898421873,
+      "loss": 0.8781,
+      "step": 4980
+    },
+    {
+      "epoch": 6.16573902288188,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00014654488378062087,
+      "loss": 0.8809,
+      "step": 4985
+    },
+    {
+      "epoch": 6.171923314780457,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00014641738119815266,
+      "loss": 0.885,
+      "step": 4990
+    },
+    {
+      "epoch": 6.178107606679035,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00014628978235918695,
+      "loss": 0.8823,
+      "step": 4995
+    },
+    {
+      "epoch": 6.184291898577613,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00014616208752832758,
+      "loss": 0.8882,
+      "step": 5000
+    },
+    {
+      "epoch": 6.190476190476191,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0001460342969703774,
+      "loss": 0.885,
+      "step": 5005
+    },
+    {
+      "epoch": 6.196660482374768,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00014590641095033787,
+      "loss": 0.8901,
+      "step": 5010
+    },
+    {
+      "epoch": 6.202844774273346,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001457784297334083,
+      "loss": 0.8722,
+      "step": 5015
+    },
+    {
+      "epoch": 6.209029066171923,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0001456503535849855,
+      "loss": 0.8808,
+      "step": 5020
+    },
+    {
+      "epoch": 6.215213358070501,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00014552218277066314,
+      "loss": 0.8734,
+      "step": 5025
+    },
+    {
+      "epoch": 6.221397649969078,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.0001453939175562312,
+      "loss": 0.8814,
+      "step": 5030
+    },
+    {
+      "epoch": 6.227581941867657,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00014526555820767534,
+      "loss": 0.8775,
+      "step": 5035
+    },
+    {
+      "epoch": 6.233766233766234,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00014513710499117647,
+      "loss": 0.8756,
+      "step": 5040
+    },
+    {
+      "epoch": 6.239950525664812,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0001450085581731102,
+      "loss": 0.8814,
+      "step": 5045
+    },
+    {
+      "epoch": 6.246134817563389,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00014487991802004623,
+      "loss": 0.8778,
+      "step": 5050
+    },
+    {
+      "epoch": 6.252319109461967,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00014475118479874774,
+      "loss": 0.8744,
+      "step": 5055
+    },
+    {
+      "epoch": 6.258503401360544,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00014462235877617098,
+      "loss": 0.876,
+      "step": 5060
+    },
+    {
+      "epoch": 6.264687693259122,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00014449344021946468,
+      "loss": 0.888,
+      "step": 5065
+    },
+    {
+      "epoch": 6.270871985157699,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001443644293959693,
+      "loss": 0.8792,
+      "step": 5070
+    },
+    {
+      "epoch": 6.2770562770562774,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001442353265732168,
+      "loss": 0.8672,
+      "step": 5075
+    },
+    {
+      "epoch": 6.283240568954855,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00014410613201892985,
+      "loss": 0.8773,
+      "step": 5080
+    },
+    {
+      "epoch": 6.289424860853432,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001439768460010213,
+      "loss": 0.8835,
+      "step": 5085
+    },
+    {
+      "epoch": 6.29560915275201,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001438474687875938,
+      "loss": 0.8845,
+      "step": 5090
+    },
+    {
+      "epoch": 6.301793444650587,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.000143718000646939,
+      "loss": 0.8753,
+      "step": 5095
+    },
+    {
+      "epoch": 6.307977736549165,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00014358844184753712,
+      "loss": 0.8719,
+      "step": 5100
+    },
+    {
+      "epoch": 6.314162028447742,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00014345879265805644,
+      "loss": 0.8807,
+      "step": 5105
+    },
+    {
+      "epoch": 6.32034632034632,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00014332905334735261,
+      "loss": 0.8832,
+      "step": 5110
+    },
+    {
+      "epoch": 6.326530612244898,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00014319922418446824,
+      "loss": 0.8805,
+      "step": 5115
+    },
+    {
+      "epoch": 6.332714904143476,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00014306930543863219,
+      "loss": 0.8839,
+      "step": 5120
+    },
+    {
+      "epoch": 6.338899196042053,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0001429392973792592,
+      "loss": 0.8863,
+      "step": 5125
+    },
+    {
+      "epoch": 6.345083487940631,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00014280920027594907,
+      "loss": 0.8783,
+      "step": 5130
+    },
+    {
+      "epoch": 6.351267779839208,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001426790143984864,
+      "loss": 0.8823,
+      "step": 5135
+    },
+    {
+      "epoch": 6.357452071737786,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00014254874001683976,
+      "loss": 0.8857,
+      "step": 5140
+    },
+    {
+      "epoch": 6.363636363636363,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00014241837740116132,
+      "loss": 0.8865,
+      "step": 5145
+    },
+    {
+      "epoch": 6.369820655534941,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00014228792682178623,
+      "loss": 0.8779,
+      "step": 5150
+    },
+    {
+      "epoch": 6.376004947433519,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00014215738854923203,
+      "loss": 0.8813,
+      "step": 5155
+    },
+    {
+      "epoch": 6.382189239332097,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00014202676285419812,
+      "loss": 0.8824,
+      "step": 5160
+    },
+    {
+      "epoch": 6.388373531230674,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00014189605000756514,
+      "loss": 0.8897,
+      "step": 5165
+    },
+    {
+      "epoch": 6.394557823129252,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00014176525028039452,
+      "loss": 0.8711,
+      "step": 5170
+    },
+    {
+      "epoch": 6.400742115027829,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00014163436394392786,
+      "loss": 0.8929,
+      "step": 5175
+    },
+    {
+      "epoch": 6.406926406926407,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00014150339126958633,
+      "loss": 0.8739,
+      "step": 5180
+    },
+    {
+      "epoch": 6.413110698824984,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001413723325289701,
+      "loss": 0.876,
+      "step": 5185
+    },
+    {
+      "epoch": 6.419294990723563,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00014124118799385796,
+      "loss": 0.8756,
+      "step": 5190
+    },
+    {
+      "epoch": 6.42547928262214,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00014110995793620648,
+      "loss": 0.8882,
+      "step": 5195
+    },
+    {
+      "epoch": 6.431663574520718,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00014097864262814955,
+      "loss": 0.8771,
+      "step": 5200
+    },
+    {
+      "epoch": 6.437847866419295,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.000140847242341998,
+      "loss": 0.8794,
+      "step": 5205
+    },
+    {
+      "epoch": 6.444032158317873,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00014071575735023875,
+      "loss": 0.8804,
+      "step": 5210
+    },
+    {
+      "epoch": 6.45021645021645,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00014058418792553445,
+      "loss": 0.874,
+      "step": 5215
+    },
+    {
+      "epoch": 6.456400742115028,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0001404525343407228,
+      "loss": 0.8731,
+      "step": 5220
+    },
+    {
+      "epoch": 6.462585034013605,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00014032079686881603,
+      "loss": 0.8762,
+      "step": 5225
+    },
+    {
+      "epoch": 6.4687693259121835,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00014018897578300035,
+      "loss": 0.8906,
+      "step": 5230
+    },
+    {
+      "epoch": 6.474953617810761,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00014005707135663527,
+      "loss": 0.8737,
+      "step": 5235
+    },
+    {
+      "epoch": 6.4811379097093385,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0001399250838632533,
+      "loss": 0.8806,
+      "step": 5240
+    },
+    {
+      "epoch": 6.487322201607916,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00013979301357655905,
+      "loss": 0.8737,
+      "step": 5245
+    },
+    {
+      "epoch": 6.4935064935064934,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001396608607704289,
+      "loss": 0.878,
+      "step": 5250
+    },
+    {
+      "epoch": 6.499690785405071,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00013952862571891034,
+      "loss": 0.8754,
+      "step": 5255
+    },
+    {
+      "epoch": 6.5058750773036484,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00013939630869622133,
+      "loss": 0.8818,
+      "step": 5260
+    },
+    {
+      "epoch": 6.512059369202227,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00013926390997674997,
+      "loss": 0.8834,
+      "step": 5265
+    },
+    {
+      "epoch": 6.518243661100804,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00013913142983505364,
+      "loss": 0.8787,
+      "step": 5270
+    },
+    {
+      "epoch": 6.524427952999382,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00013899886854585862,
+      "loss": 0.8813,
+      "step": 5275
+    },
+    {
+      "epoch": 6.530612244897959,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00013886622638405952,
+      "loss": 0.8835,
+      "step": 5280
+    },
+    {
+      "epoch": 6.536796536796537,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00013873350362471855,
+      "loss": 0.8811,
+      "step": 5285
+    },
+    {
+      "epoch": 6.542980828695114,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00013860070054306516,
+      "loss": 0.891,
+      "step": 5290
+    },
+    {
+      "epoch": 6.549165120593692,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00013846781741449525,
+      "loss": 0.8807,
+      "step": 5295
+    },
+    {
+      "epoch": 6.555349412492269,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001383348545145708,
+      "loss": 0.883,
+      "step": 5300
+    },
+    {
+      "epoch": 6.561533704390847,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0001382018121190192,
+      "loss": 0.8807,
+      "step": 5305
+    },
+    {
+      "epoch": 6.567717996289425,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0001380686905037327,
+      "loss": 0.8844,
+      "step": 5310
+    },
+    {
+      "epoch": 6.573902288188003,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0001379354899447678,
+      "loss": 0.893,
+      "step": 5315
+    },
+    {
+      "epoch": 6.58008658008658,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00013780221071834476,
+      "loss": 0.8825,
+      "step": 5320
+    },
+    {
+      "epoch": 6.586270871985158,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00013766885310084688,
+      "loss": 0.8899,
+      "step": 5325
+    },
+    {
+      "epoch": 6.592455163883735,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0001375354173688201,
+      "loss": 0.8774,
+      "step": 5330
+    },
+    {
+      "epoch": 6.598639455782313,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00013740190379897226,
+      "loss": 0.8802,
+      "step": 5335
+    },
+    {
+      "epoch": 6.60482374768089,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00013726831266817278,
+      "loss": 0.877,
+      "step": 5340
+    },
+    {
+      "epoch": 6.611008039579469,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00013713464425345174,
+      "loss": 0.8791,
+      "step": 5345
+    },
+    {
+      "epoch": 6.617192331478046,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00013700089883199966,
+      "loss": 0.8822,
+      "step": 5350
+    },
+    {
+      "epoch": 6.623376623376624,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0001368670766811665,
+      "loss": 0.8786,
+      "step": 5355
+    },
+    {
+      "epoch": 6.629560915275201,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001367331780784616,
+      "loss": 0.876,
+      "step": 5360
+    },
+    {
+      "epoch": 6.635745207173779,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001365992033015527,
+      "loss": 0.875,
+      "step": 5365
+    },
+    {
+      "epoch": 6.641929499072356,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00013646515262826552,
+      "loss": 0.8762,
+      "step": 5370
+    },
+    {
+      "epoch": 6.648113790970934,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00013633102633658318,
+      "loss": 0.8862,
+      "step": 5375
+    },
+    {
+      "epoch": 6.654298082869511,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00013619682470464558,
+      "loss": 0.889,
+      "step": 5380
+    },
+    {
+      "epoch": 6.660482374768089,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00013606254801074895,
+      "loss": 0.883,
+      "step": 5385
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00013592819653334505,
+      "loss": 0.8794,
+      "step": 5390
+    },
+    {
+      "epoch": 6.6728509585652445,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001357937705510408,
+      "loss": 0.8812,
+      "step": 5395
+    },
+    {
+      "epoch": 6.679035250463822,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001356592703425976,
+      "loss": 0.8742,
+      "step": 5400
+    },
+    {
+      "epoch": 6.6852195423623995,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00013552469618693076,
+      "loss": 0.8759,
+      "step": 5405
+    },
+    {
+      "epoch": 6.691403834260977,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00013539004836310894,
+      "loss": 0.8747,
+      "step": 5410
+    },
+    {
+      "epoch": 6.6975881261595545,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00013525532715035366,
+      "loss": 0.8803,
+      "step": 5415
+    },
+    {
+      "epoch": 6.703772418058133,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0001351205328280385,
+      "loss": 0.8769,
+      "step": 5420
+    },
+    {
+      "epoch": 6.70995670995671,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00013498566567568865,
+      "loss": 0.8749,
+      "step": 5425
+    },
+    {
+      "epoch": 6.716141001855288,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00013485072597298038,
+      "loss": 0.8708,
+      "step": 5430
+    },
+    {
+      "epoch": 6.722325293753865,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00013471571399974045,
+      "loss": 0.8736,
+      "step": 5435
+    },
+    {
+      "epoch": 6.728509585652443,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00013458063003594543,
+      "loss": 0.8774,
+      "step": 5440
+    },
+    {
+      "epoch": 6.73469387755102,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00013444547436172117,
+      "loss": 0.877,
+      "step": 5445
+    },
+    {
+      "epoch": 6.740878169449598,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001343102472573423,
+      "loss": 0.8998,
+      "step": 5450
+    },
+    {
+      "epoch": 6.747062461348175,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00013417494900323142,
+      "loss": 0.8903,
+      "step": 5455
+    },
+    {
+      "epoch": 6.753246753246753,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00013403957987995882,
+      "loss": 0.8768,
+      "step": 5460
+    },
+    {
+      "epoch": 6.759431045145331,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00013390414016824174,
+      "loss": 0.8859,
+      "step": 5465
+    },
+    {
+      "epoch": 6.765615337043909,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00013376863014894375,
+      "loss": 0.8784,
+      "step": 5470
+    },
+    {
+      "epoch": 6.771799628942486,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00013363305010307425,
+      "loss": 0.8793,
+      "step": 5475
+    },
+    {
+      "epoch": 6.777983920841064,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00013349740031178784,
+      "loss": 0.8688,
+      "step": 5480
+    },
+    {
+      "epoch": 6.784168212739641,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0001333616810563837,
+      "loss": 0.8727,
+      "step": 5485
+    },
+    {
+      "epoch": 6.790352504638219,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00013322589261830517,
+      "loss": 0.8718,
+      "step": 5490
+    },
+    {
+      "epoch": 6.796536796536796,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.000133090035279139,
+      "loss": 0.8866,
+      "step": 5495
+    },
+    {
+      "epoch": 6.802721088435375,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00013295410932061478,
+      "loss": 0.8834,
+      "step": 5500
+    },
+    {
+      "epoch": 6.808905380333952,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001328181150246045,
+      "loss": 0.8827,
+      "step": 5505
+    },
+    {
+      "epoch": 6.81508967223253,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00013268205267312174,
+      "loss": 0.8842,
+      "step": 5510
+    },
+    {
+      "epoch": 6.821273964131107,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001325459225483213,
+      "loss": 0.8813,
+      "step": 5515
+    },
+    {
+      "epoch": 6.827458256029685,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00013240972493249847,
+      "loss": 0.8872,
+      "step": 5520
+    },
+    {
+      "epoch": 6.833642547928262,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001322734601080885,
+      "loss": 0.8884,
+      "step": 5525
+    },
+    {
+      "epoch": 6.83982683982684,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00013213712835766607,
+      "loss": 0.8877,
+      "step": 5530
+    },
+    {
+      "epoch": 6.846011131725417,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0001320007299639446,
+      "loss": 0.8777,
+      "step": 5535
+    },
+    {
+      "epoch": 6.852195423623995,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0001318642652097757,
+      "loss": 0.8732,
+      "step": 5540
+    },
+    {
+      "epoch": 6.858379715522573,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00013172773437814865,
+      "loss": 0.8859,
+      "step": 5545
+    },
+    {
+      "epoch": 6.8645640074211505,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00013159113775218964,
+      "loss": 0.8738,
+      "step": 5550
+    },
+    {
+      "epoch": 6.870748299319728,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00013145447561516138,
+      "loss": 0.8849,
+      "step": 5555
+    },
+    {
+      "epoch": 6.8769325912183055,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00013131774825046245,
+      "loss": 0.8766,
+      "step": 5560
+    },
+    {
+      "epoch": 6.883116883116883,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001311809559416267,
+      "loss": 0.881,
+      "step": 5565
+    },
+    {
+      "epoch": 6.8893011750154605,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00013104409897232258,
+      "loss": 0.8833,
+      "step": 5570
+    },
+    {
+      "epoch": 6.895485466914038,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00013090717762635266,
+      "loss": 0.8787,
+      "step": 5575
+    },
+    {
+      "epoch": 6.901669758812616,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00013077019218765305,
+      "loss": 0.8843,
+      "step": 5580
+    },
+    {
+      "epoch": 6.907854050711194,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0001306331429402927,
+      "loss": 0.8901,
+      "step": 5585
+    },
+    {
+      "epoch": 6.914038342609771,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00013049603016847296,
+      "loss": 0.8918,
+      "step": 5590
+    },
+    {
+      "epoch": 6.920222634508349,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00013035885415652685,
+      "loss": 0.8775,
+      "step": 5595
+    },
+    {
+      "epoch": 6.926406926406926,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00013022161518891855,
+      "loss": 0.8947,
+      "step": 5600
+    },
+    {
+      "epoch": 6.932591218305504,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00013008431355024283,
+      "loss": 0.8815,
+      "step": 5605
+    },
+    {
+      "epoch": 6.938775510204081,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00012994694952522435,
+      "loss": 0.8678,
+      "step": 5610
+    },
+    {
+      "epoch": 6.944959802102659,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00012980952339871718,
+      "loss": 0.8793,
+      "step": 5615
+    },
+    {
+      "epoch": 6.951144094001237,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00012967203545570418,
+      "loss": 0.8782,
+      "step": 5620
+    },
+    {
+      "epoch": 6.957328385899815,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00012953448598129643,
+      "loss": 0.8754,
+      "step": 5625
+    },
+    {
+      "epoch": 6.963512677798392,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001293968752607325,
+      "loss": 0.8751,
+      "step": 5630
+    },
+    {
+      "epoch": 6.96969696969697,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00012925920357937808,
+      "loss": 0.8816,
+      "step": 5635
+    },
+    {
+      "epoch": 6.975881261595547,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00012912147122272523,
+      "loss": 0.8865,
+      "step": 5640
+    },
+    {
+      "epoch": 6.982065553494125,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001289836784763918,
+      "loss": 0.8764,
+      "step": 5645
+    },
+    {
+      "epoch": 6.988249845392702,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00012884582562612095,
+      "loss": 0.8862,
+      "step": 5650
+    },
+    {
+      "epoch": 6.994434137291281,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001287079129577804,
+      "loss": 0.8795,
+      "step": 5655
+    },
+    {
+      "epoch": 6.999381570810142,
+      "eval_loss": 2.5032334327697754,
+      "eval_runtime": 0.6588,
+      "eval_samples_per_second": 15.178,
+      "eval_steps_per_second": 1.518,
+      "step": 5659
+    },
+    {
+      "epoch": 7.000618429189858,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00012856994075736197,
+      "loss": 0.8769,
+      "step": 5660
+    },
+    {
+      "epoch": 7.006802721088436,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00012843190931098093,
+      "loss": 0.8531,
+      "step": 5665
+    },
+    {
+      "epoch": 7.012987012987013,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00012829381890487536,
+      "loss": 0.8766,
+      "step": 5670
+    },
+    {
+      "epoch": 7.019171304885591,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00012815566982540567,
+      "loss": 0.8794,
+      "step": 5675
+    },
+    {
+      "epoch": 7.025355596784168,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00012801746235905384,
+      "loss": 0.8736,
+      "step": 5680
+    },
+    {
+      "epoch": 7.031539888682746,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00012787919679242306,
+      "loss": 0.8789,
+      "step": 5685
+    },
+    {
+      "epoch": 7.037724180581323,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00012774087341223695,
+      "loss": 0.8716,
+      "step": 5690
+    },
+    {
+      "epoch": 7.0439084724799015,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.000127602492505339,
+      "loss": 0.8688,
+      "step": 5695
+    },
+    {
+      "epoch": 7.050092764378479,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00012746405435869198,
+      "loss": 0.8736,
+      "step": 5700
+    },
+    {
+      "epoch": 7.0562770562770565,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0001273255592593774,
+      "loss": 0.8623,
+      "step": 5705
+    },
+    {
+      "epoch": 7.062461348175634,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00012718700749459486,
+      "loss": 0.8647,
+      "step": 5710
+    },
+    {
+      "epoch": 7.0686456400742115,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00012704839935166143,
+      "loss": 0.8648,
+      "step": 5715
+    },
+    {
+      "epoch": 7.074829931972789,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001269097351180112,
+      "loss": 0.8635,
+      "step": 5720
+    },
+    {
+      "epoch": 7.0810142238713665,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00012677101508119445,
+      "loss": 0.8724,
+      "step": 5725
+    },
+    {
+      "epoch": 7.087198515769944,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00012663223952887723,
+      "loss": 0.8703,
+      "step": 5730
+    },
+    {
+      "epoch": 7.093382807668522,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00012649340874884075,
+      "loss": 0.8688,
+      "step": 5735
+    },
+    {
+      "epoch": 7.0995670995671,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001263545230289807,
+      "loss": 0.8691,
+      "step": 5740
+    },
+    {
+      "epoch": 7.105751391465677,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001262155826573067,
+      "loss": 0.8708,
+      "step": 5745
+    },
+    {
+      "epoch": 7.111935683364255,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00012607658792194174,
+      "loss": 0.872,
+      "step": 5750
+    },
+    {
+      "epoch": 7.118119975262832,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0001259375391111215,
+      "loss": 0.8774,
+      "step": 5755
+    },
+    {
+      "epoch": 7.12430426716141,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001257984365131938,
+      "loss": 0.8759,
+      "step": 5760
+    },
+    {
+      "epoch": 7.130488559059987,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001256592804166181,
+      "loss": 0.8642,
+      "step": 5765
+    },
+    {
+      "epoch": 7.136672850958565,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00012552007110996463,
+      "loss": 0.8667,
+      "step": 5770
+    },
+    {
+      "epoch": 7.142857142857143,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00012538080888191408,
+      "loss": 0.8672,
+      "step": 5775
+    },
+    {
+      "epoch": 7.149041434755721,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00012524149402125685,
+      "loss": 0.8582,
+      "step": 5780
+    },
+    {
+      "epoch": 7.155225726654298,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001251021268168925,
+      "loss": 0.8789,
+      "step": 5785
+    },
+    {
+      "epoch": 7.161410018552876,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00012496270755782914,
+      "loss": 0.8787,
+      "step": 5790
+    },
+    {
+      "epoch": 7.167594310451453,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00012482323653318278,
+      "loss": 0.8723,
+      "step": 5795
+    },
+    {
+      "epoch": 7.173778602350031,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00012468371403217684,
+      "loss": 0.8649,
+      "step": 5800
+    },
+    {
+      "epoch": 7.179962894248608,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00012454414034414142,
+      "loss": 0.877,
+      "step": 5805
+    },
+    {
+      "epoch": 7.186147186147186,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00012440451575851285,
+      "loss": 0.8851,
+      "step": 5810
+    },
+    {
+      "epoch": 7.192331478045764,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00012426484056483292,
+      "loss": 0.8703,
+      "step": 5815
+    },
+    {
+      "epoch": 7.198515769944342,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00012412511505274844,
+      "loss": 0.8664,
+      "step": 5820
+    },
+    {
+      "epoch": 7.204700061842919,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00012398533951201053,
+      "loss": 0.8617,
+      "step": 5825
+    },
+    {
+      "epoch": 7.210884353741497,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00012384551423247407,
+      "loss": 0.8691,
+      "step": 5830
+    },
+    {
+      "epoch": 7.217068645640074,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00012370563950409703,
+      "loss": 0.8808,
+      "step": 5835
+    },
+    {
+      "epoch": 7.223252937538652,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00012356571561693996,
+      "loss": 0.8674,
+      "step": 5840
+    },
+    {
+      "epoch": 7.229437229437229,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00012342574286116544,
+      "loss": 0.8644,
+      "step": 5845
+    },
+    {
+      "epoch": 7.235621521335807,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00012328572152703725,
+      "loss": 0.8745,
+      "step": 5850
+    },
+    {
+      "epoch": 7.241805813234385,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00012314565190491998,
+      "loss": 0.8642,
+      "step": 5855
+    },
+    {
+      "epoch": 7.2479901051329625,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00012300553428527832,
+      "loss": 0.8785,
+      "step": 5860
+    },
+    {
+      "epoch": 7.25417439703154,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00012286536895867654,
+      "loss": 0.8719,
+      "step": 5865
+    },
+    {
+      "epoch": 7.2603586889301175,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00012272515621577782,
+      "loss": 0.8651,
+      "step": 5870
+    },
+    {
+      "epoch": 7.266542980828695,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00012258489634734367,
+      "loss": 0.87,
+      "step": 5875
+    },
+    {
+      "epoch": 7.2727272727272725,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00012244458964423327,
+      "loss": 0.8693,
+      "step": 5880
+    },
+    {
+      "epoch": 7.27891156462585,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.000122304236397403,
+      "loss": 0.8658,
+      "step": 5885
+    },
+    {
+      "epoch": 7.285095856524428,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00012216383689790574,
+      "loss": 0.8749,
+      "step": 5890
+    },
+    {
+      "epoch": 7.291280148423006,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00012202339143689023,
+      "loss": 0.8737,
+      "step": 5895
+    },
+    {
+      "epoch": 7.297464440321583,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00012188290030560063,
+      "loss": 0.871,
+      "step": 5900
+    },
+    {
+      "epoch": 7.303648732220161,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00012174236379537572,
+      "loss": 0.8725,
+      "step": 5905
+    },
+    {
+      "epoch": 7.309833024118738,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00012160178219764837,
+      "loss": 0.8718,
+      "step": 5910
+    },
+    {
+      "epoch": 7.316017316017316,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00012146115580394499,
+      "loss": 0.8763,
+      "step": 5915
+    },
+    {
+      "epoch": 7.322201607915893,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00012132048490588492,
+      "loss": 0.8658,
+      "step": 5920
+    },
+    {
+      "epoch": 7.328385899814471,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00012117976979517973,
+      "loss": 0.8633,
+      "step": 5925
+    },
+    {
+      "epoch": 7.334570191713049,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00012103901076363269,
+      "loss": 0.8827,
+      "step": 5930
+    },
+    {
+      "epoch": 7.340754483611627,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00012089820810313812,
+      "loss": 0.8769,
+      "step": 5935
+    },
+    {
+      "epoch": 7.346938775510204,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001207573621056809,
+      "loss": 0.8767,
+      "step": 5940
+    },
+    {
+      "epoch": 7.353123067408782,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00012061647306333568,
+      "loss": 0.8909,
+      "step": 5945
+    },
+    {
+      "epoch": 7.359307359307359,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00012047554126826643,
+      "loss": 0.8697,
+      "step": 5950
+    },
+    {
+      "epoch": 7.365491651205937,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00012033456701272576,
+      "loss": 0.8715,
+      "step": 5955
+    },
+    {
+      "epoch": 7.371675943104514,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00012019355058905435,
+      "loss": 0.8731,
+      "step": 5960
+    },
+    {
+      "epoch": 7.377860235003092,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00012005249228968032,
+      "loss": 0.8818,
+      "step": 5965
+    },
+    {
+      "epoch": 7.38404452690167,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00011991139240711857,
+      "loss": 0.8656,
+      "step": 5970
+    },
+    {
+      "epoch": 7.390228818800248,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00011977025123397033,
+      "loss": 0.8862,
+      "step": 5975
+    },
+    {
+      "epoch": 7.396413110698825,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00011962906906292238,
+      "loss": 0.8726,
+      "step": 5980
+    },
+    {
+      "epoch": 7.402597402597403,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00011948784618674653,
+      "loss": 0.8633,
+      "step": 5985
+    },
+    {
+      "epoch": 7.40878169449598,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00011934658289829902,
+      "loss": 0.8705,
+      "step": 5990
+    },
+    {
+      "epoch": 7.414965986394558,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00011920527949051991,
+      "loss": 0.8736,
+      "step": 5995
+    },
+    {
+      "epoch": 7.421150278293135,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00011906393625643244,
+      "loss": 0.8721,
+      "step": 6000
+    },
+    {
+      "epoch": 7.427334570191713,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00011892255348914239,
+      "loss": 0.8733,
+      "step": 6005
+    },
+    {
+      "epoch": 7.433518862090291,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00011878113148183758,
+      "loss": 0.8753,
+      "step": 6010
+    },
+    {
+      "epoch": 7.4397031539888685,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00011863967052778721,
+      "loss": 0.8805,
+      "step": 6015
+    },
+    {
+      "epoch": 7.445887445887446,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00011849817092034118,
+      "loss": 0.881,
+      "step": 6020
+    },
+    {
+      "epoch": 7.4520717377860235,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00011835663295292963,
+      "loss": 0.8815,
+      "step": 6025
+    },
+    {
+      "epoch": 7.458256029684601,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00011821505691906216,
+      "loss": 0.8684,
+      "step": 6030
+    },
+    {
+      "epoch": 7.4644403215831785,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00011807344311232738,
+      "loss": 0.88,
+      "step": 6035
+    },
+    {
+      "epoch": 7.470624613481756,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00011793179182639218,
+      "loss": 0.8808,
+      "step": 6040
+    },
+    {
+      "epoch": 7.476808905380334,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0001177901033550012,
+      "loss": 0.8732,
+      "step": 6045
+    },
+    {
+      "epoch": 7.482993197278912,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00011764837799197622,
+      "loss": 0.8767,
+      "step": 6050
+    },
+    {
+      "epoch": 7.489177489177489,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00011750661603121544,
+      "loss": 0.8694,
+      "step": 6055
+    },
+    {
+      "epoch": 7.495361781076067,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 0.8702,
+      "step": 6060
+    },
+    {
+      "epoch": 7.501546072974644,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00011722298349245844,
+      "loss": 0.8751,
+      "step": 6065
+    },
+    {
+      "epoch": 7.507730364873222,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0001170811135026357,
+      "loss": 0.8758,
+      "step": 6070
+    },
+    {
+      "epoch": 7.513914656771799,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00011693920809142305,
+      "loss": 0.8732,
+      "step": 6075
+    },
+    {
+      "epoch": 7.520098948670377,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00011679726755309205,
+      "loss": 0.8649,
+      "step": 6080
+    },
+    {
+      "epoch": 7.526283240568954,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00011665529218198721,
+      "loss": 0.8688,
+      "step": 6085
+    },
+    {
+      "epoch": 7.532467532467533,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00011651328227252517,
+      "loss": 0.8727,
+      "step": 6090
+    },
+    {
+      "epoch": 7.53865182436611,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0001163712381191943,
+      "loss": 0.8796,
+      "step": 6095
+    },
+    {
+      "epoch": 7.544836116264688,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00011622916001655388,
+      "loss": 0.8723,
+      "step": 6100
+    },
+    {
+      "epoch": 7.551020408163265,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00011608704825923369,
+      "loss": 0.874,
+      "step": 6105
+    },
+    {
+      "epoch": 7.557204700061843,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00011594490314193323,
+      "loss": 0.8792,
+      "step": 6110
+    },
+    {
+      "epoch": 7.56338899196042,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00011580272495942119,
+      "loss": 0.8685,
+      "step": 6115
+    },
+    {
+      "epoch": 7.569573283858999,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00011566051400653486,
+      "loss": 0.8656,
+      "step": 6120
+    },
+    {
+      "epoch": 7.575757575757576,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00011551827057817945,
+      "loss": 0.8799,
+      "step": 6125
+    },
+    {
+      "epoch": 7.581941867656154,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00011537599496932752,
+      "loss": 0.8796,
+      "step": 6130
+    },
+    {
+      "epoch": 7.588126159554731,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00011523368747501839,
+      "loss": 0.8738,
+      "step": 6135
+    },
+    {
+      "epoch": 7.594310451453309,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00011509134839035748,
+      "loss": 0.8787,
+      "step": 6140
+    },
+    {
+      "epoch": 7.600494743351886,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00011494897801051574,
+      "loss": 0.8755,
+      "step": 6145
+    },
+    {
+      "epoch": 7.606679035250464,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00011480657663072896,
+      "loss": 0.878,
+      "step": 6150
+    },
+    {
+      "epoch": 7.612863327149041,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00011466414454629731,
+      "loss": 0.8772,
+      "step": 6155
+    },
+    {
+      "epoch": 7.619047619047619,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001145216820525845,
+      "loss": 0.8697,
+      "step": 6160
+    },
+    {
+      "epoch": 7.625231910946197,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00011437918944501749,
+      "loss": 0.8816,
+      "step": 6165
+    },
+    {
+      "epoch": 7.6314162028447745,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00011423666701908547,
+      "loss": 0.8607,
+      "step": 6170
+    },
+    {
+      "epoch": 7.637600494743352,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00011409411507033962,
+      "loss": 0.8783,
+      "step": 6175
+    },
+    {
+      "epoch": 7.6437847866419295,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00011395153389439233,
+      "loss": 0.8765,
+      "step": 6180
+    },
+    {
+      "epoch": 7.649969078540507,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00011380892378691646,
+      "loss": 0.8723,
+      "step": 6185
+    },
+    {
+      "epoch": 7.6561533704390845,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00011366628504364509,
+      "loss": 0.8783,
+      "step": 6190
+    },
+    {
+      "epoch": 7.662337662337662,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00011352361796037047,
+      "loss": 0.8764,
+      "step": 6195
+    },
+    {
+      "epoch": 7.66852195423624,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00011338092283294377,
+      "loss": 0.8831,
+      "step": 6200
+    },
+    {
+      "epoch": 7.674706246134818,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00011323819995727421,
+      "loss": 0.8728,
+      "step": 6205
+    },
+    {
+      "epoch": 7.680890538033395,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00011309544962932862,
+      "loss": 0.8765,
+      "step": 6210
+    },
+    {
+      "epoch": 7.687074829931973,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001129526721451307,
+      "loss": 0.8725,
+      "step": 6215
+    },
+    {
+      "epoch": 7.69325912183055,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00011280986780076057,
+      "loss": 0.8677,
+      "step": 6220
+    },
+    {
+      "epoch": 7.699443413729128,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00011266703689235394,
+      "loss": 0.8769,
+      "step": 6225
+    },
+    {
+      "epoch": 7.705627705627705,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00011252417971610163,
+      "loss": 0.8719,
+      "step": 6230
+    },
+    {
+      "epoch": 7.711811997526283,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00011238129656824898,
+      "loss": 0.8735,
+      "step": 6235
+    },
+    {
+      "epoch": 7.71799628942486,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00011223838774509514,
+      "loss": 0.8828,
+      "step": 6240
+    },
+    {
+      "epoch": 7.724180581323439,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00011209545354299251,
+      "loss": 0.8807,
+      "step": 6245
+    },
+    {
+      "epoch": 7.730364873222016,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00011195249425834615,
+      "loss": 0.8844,
+      "step": 6250
+    },
+    {
+      "epoch": 7.736549165120594,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00011180951018761314,
+      "loss": 0.8783,
+      "step": 6255
+    },
+    {
+      "epoch": 7.742733457019171,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00011166650162730188,
+      "loss": 0.8608,
+      "step": 6260
+    },
+    {
+      "epoch": 7.748917748917749,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001115234688739716,
+      "loss": 0.8747,
+      "step": 6265
+    },
+    {
+      "epoch": 7.755102040816326,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00011138041222423177,
+      "loss": 0.8773,
+      "step": 6270
+    },
+    {
+      "epoch": 7.761286332714904,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00011123733197474128,
+      "loss": 0.8672,
+      "step": 6275
+    },
+    {
+      "epoch": 7.767470624613482,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00011109422842220805,
+      "loss": 0.8731,
+      "step": 6280
+    },
+    {
+      "epoch": 7.77365491651206,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00011095110186338835,
+      "loss": 0.8684,
+      "step": 6285
+    },
+    {
+      "epoch": 7.779839208410637,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00011080795259508608,
+      "loss": 0.8699,
+      "step": 6290
+    },
+    {
+      "epoch": 7.786023500309215,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00011066478091415223,
+      "loss": 0.8673,
+      "step": 6295
+    },
+    {
+      "epoch": 7.792207792207792,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00011052158711748434,
+      "loss": 0.8707,
+      "step": 6300
+    },
+    {
+      "epoch": 7.79839208410637,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00011037837150202576,
+      "loss": 0.8729,
+      "step": 6305
+    },
+    {
+      "epoch": 7.804576376004947,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00011023513436476511,
+      "loss": 0.8627,
+      "step": 6310
+    },
+    {
+      "epoch": 7.810760667903525,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00011009187600273566,
+      "loss": 0.8787,
+      "step": 6315
+    },
+    {
+      "epoch": 7.816944959802103,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00010994859671301462,
+      "loss": 0.8708,
+      "step": 6320
+    },
+    {
+      "epoch": 7.8231292517006805,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00010980529679272266,
+      "loss": 0.8691,
+      "step": 6325
+    },
+    {
+      "epoch": 7.829313543599258,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001096619765390232,
+      "loss": 0.8678,
+      "step": 6330
+    },
+    {
+      "epoch": 7.8354978354978355,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00010951863624912185,
+      "loss": 0.8747,
+      "step": 6335
+    },
+    {
+      "epoch": 7.841682127396413,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00010937527622026575,
+      "loss": 0.8727,
+      "step": 6340
+    },
+    {
+      "epoch": 7.8478664192949905,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00010923189674974301,
+      "loss": 0.8742,
+      "step": 6345
+    },
+    {
+      "epoch": 7.854050711193568,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00010908849813488203,
+      "loss": 0.8675,
+      "step": 6350
+    },
+    {
+      "epoch": 7.860235003092146,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00010894508067305088,
+      "loss": 0.8758,
+      "step": 6355
+    },
+    {
+      "epoch": 7.866419294990724,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00010880164466165674,
+      "loss": 0.8716,
+      "step": 6360
+    },
+    {
+      "epoch": 7.872603586889301,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00010865819039814526,
+      "loss": 0.8738,
+      "step": 6365
+    },
+    {
+      "epoch": 7.878787878787879,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00010851471817999997,
+      "loss": 0.8819,
+      "step": 6370
+    },
+    {
+      "epoch": 7.884972170686456,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00010837122830474158,
+      "loss": 0.8703,
+      "step": 6375
+    },
+    {
+      "epoch": 7.891156462585034,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00010822772106992747,
+      "loss": 0.8778,
+      "step": 6380
+    },
+    {
+      "epoch": 7.897340754483611,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00010808419677315093,
+      "loss": 0.8737,
+      "step": 6385
+    },
+    {
+      "epoch": 7.903525046382189,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00010794065571204072,
+      "loss": 0.8775,
+      "step": 6390
+    },
+    {
+      "epoch": 7.909709338280766,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00010779709818426033,
+      "loss": 0.8834,
+      "step": 6395
+    },
+    {
+      "epoch": 7.915893630179345,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0001076535244875074,
+      "loss": 0.8678,
+      "step": 6400
+    },
+    {
+      "epoch": 7.922077922077922,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001075099349195131,
+      "loss": 0.8648,
+      "step": 6405
+    },
+    {
+      "epoch": 7.9282622139765,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00010736632977804149,
+      "loss": 0.8707,
+      "step": 6410
+    },
+    {
+      "epoch": 7.934446505875077,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00010722270936088898,
+      "loss": 0.8705,
+      "step": 6415
+    },
+    {
+      "epoch": 7.940630797773655,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00010707907396588361,
+      "loss": 0.8754,
+      "step": 6420
+    },
+    {
+      "epoch": 7.946815089672232,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00010693542389088452,
+      "loss": 0.8723,
+      "step": 6425
+    },
+    {
+      "epoch": 7.95299938157081,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00010679175943378119,
+      "loss": 0.8746,
+      "step": 6430
+    },
+    {
+      "epoch": 7.959183673469388,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00010664808089249305,
+      "loss": 0.8709,
+      "step": 6435
+    },
+    {
+      "epoch": 7.965367965367966,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00010650438856496872,
+      "loss": 0.871,
+      "step": 6440
+    },
+    {
+      "epoch": 7.971552257266543,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00010636068274918536,
+      "loss": 0.8661,
+      "step": 6445
+    },
+    {
+      "epoch": 7.977736549165121,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00010621696374314807,
+      "loss": 0.8711,
+      "step": 6450
+    },
+    {
+      "epoch": 7.983920841063698,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001060732318448894,
+      "loss": 0.8759,
+      "step": 6455
+    },
+    {
+      "epoch": 7.990105132962276,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00010592948735246854,
+      "loss": 0.8632,
+      "step": 6460
+    },
+    {
+      "epoch": 7.996289424860853,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00010578573056397085,
+      "loss": 0.8694,
+      "step": 6465
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 2.506842851638794,
+      "eval_runtime": 0.5393,
+      "eval_samples_per_second": 18.543,
+      "eval_steps_per_second": 1.854,
+      "step": 6468
+    },
+    {
+      "epoch": 8.00247371675943,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00010564196177750725,
+      "loss": 0.8705,
+      "step": 6470
+    },
+    {
+      "epoch": 8.008658008658008,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00010549818129121338,
+      "loss": 0.8568,
+      "step": 6475
+    },
+    {
+      "epoch": 8.014842300556586,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0001053543894032493,
+      "loss": 0.8752,
+      "step": 6480
+    },
+    {
+      "epoch": 8.021026592455163,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00010521058641179861,
+      "loss": 0.8689,
+      "step": 6485
+    },
+    {
+      "epoch": 8.02721088435374,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00010506677261506797,
+      "loss": 0.8673,
+      "step": 6490
+    },
+    {
+      "epoch": 8.03339517625232,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00010492294831128641,
+      "loss": 0.8757,
+      "step": 6495
+    },
+    {
+      "epoch": 8.039579468150897,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00010477911379870488,
+      "loss": 0.8522,
+      "step": 6500
+    },
+    {
+      "epoch": 8.045763760049475,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00010463526937559536,
+      "loss": 0.8703,
+      "step": 6505
+    },
+    {
+      "epoch": 8.051948051948052,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00010449141534025045,
+      "loss": 0.8631,
+      "step": 6510
+    },
+    {
+      "epoch": 8.05813234384663,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00010434755199098261,
+      "loss": 0.8672,
+      "step": 6515
+    },
+    {
+      "epoch": 8.064316635745207,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00010420367962612372,
+      "loss": 0.8732,
+      "step": 6520
+    },
+    {
+      "epoch": 8.070500927643785,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00010405979854402425,
+      "loss": 0.8687,
+      "step": 6525
+    },
+    {
+      "epoch": 8.076685219542362,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00010391590904305284,
+      "loss": 0.8756,
+      "step": 6530
+    },
+    {
+      "epoch": 8.08286951144094,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00010377201142159554,
+      "loss": 0.8664,
+      "step": 6535
+    },
+    {
+      "epoch": 8.089053803339517,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00010362810597805526,
+      "loss": 0.867,
+      "step": 6540
+    },
+    {
+      "epoch": 8.095238095238095,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00010348419301085113,
+      "loss": 0.8612,
+      "step": 6545
+    },
+    {
+      "epoch": 8.101422387136672,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00010334027281841781,
+      "loss": 0.8646,
+      "step": 6550
+    },
+    {
+      "epoch": 8.10760667903525,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00010319634569920504,
+      "loss": 0.8753,
+      "step": 6555
+    },
+    {
+      "epoch": 8.113790970933827,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00010305241195167687,
+      "loss": 0.8595,
+      "step": 6560
+    },
+    {
+      "epoch": 8.119975262832405,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00010290847187431113,
+      "loss": 0.864,
+      "step": 6565
+    },
+    {
+      "epoch": 8.126159554730984,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00010276452576559879,
+      "loss": 0.8822,
+      "step": 6570
+    },
+    {
+      "epoch": 8.132343846629562,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00010262057392404328,
+      "loss": 0.8721,
+      "step": 6575
+    },
+    {
+      "epoch": 8.13852813852814,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00010247661664815986,
+      "loss": 0.8611,
+      "step": 6580
+    },
+    {
+      "epoch": 8.144712430426717,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00010233265423647523,
+      "loss": 0.8668,
+      "step": 6585
+    },
+    {
+      "epoch": 8.150896722325294,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00010218868698752658,
+      "loss": 0.8696,
+      "step": 6590
+    },
+    {
+      "epoch": 8.157081014223872,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00010204471519986124,
+      "loss": 0.8711,
+      "step": 6595
+    },
+    {
+      "epoch": 8.16326530612245,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00010190073917203589,
+      "loss": 0.87,
+      "step": 6600
+    },
+    {
+      "epoch": 8.169449598021027,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00010175675920261602,
+      "loss": 0.8702,
+      "step": 6605
+    },
+    {
+      "epoch": 8.175633889919604,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00010161277559017528,
+      "loss": 0.864,
+      "step": 6610
+    },
+    {
+      "epoch": 8.181818181818182,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00010146878863329492,
+      "loss": 0.8583,
+      "step": 6615
+    },
+    {
+      "epoch": 8.18800247371676,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00010132479863056303,
+      "loss": 0.8689,
+      "step": 6620
+    },
+    {
+      "epoch": 8.194186765615337,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00010118080588057416,
+      "loss": 0.8629,
+      "step": 6625
+    },
+    {
+      "epoch": 8.200371057513914,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00010103681068192845,
+      "loss": 0.8656,
+      "step": 6630
+    },
+    {
+      "epoch": 8.206555349412492,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00010089281333323112,
+      "loss": 0.8723,
+      "step": 6635
+    },
+    {
+      "epoch": 8.21273964131107,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00010074881413309193,
+      "loss": 0.8689,
+      "step": 6640
+    },
+    {
+      "epoch": 8.218923933209647,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00010060481338012435,
+      "loss": 0.8706,
+      "step": 6645
+    },
+    {
+      "epoch": 8.225108225108226,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00010046081137294516,
+      "loss": 0.8672,
+      "step": 6650
+    },
+    {
+      "epoch": 8.231292517006803,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00010031680841017377,
+      "loss": 0.8601,
+      "step": 6655
+    },
+    {
+      "epoch": 8.237476808905381,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00010017280479043147,
+      "loss": 0.866,
+      "step": 6660
+    },
+    {
+      "epoch": 8.243661100803958,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00010002880081234103,
+      "loss": 0.8567,
+      "step": 6665
+    },
+    {
+      "epoch": 8.249845392702536,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.988479677452584e-05,
+      "loss": 0.8731,
+      "step": 6670
+    },
+    {
+      "epoch": 8.256029684601113,
+      "grad_norm": 0.451171875,
+      "learning_rate": 9.97407929756095e-05,
+      "loss": 0.8686,
+      "step": 6675
+    },
+    {
+      "epoch": 8.262213976499691,
+      "grad_norm": 0.4765625,
+      "learning_rate": 9.959678971421508e-05,
+      "loss": 0.8743,
+      "step": 6680
+    },
+    {
+      "epoch": 8.268398268398268,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.94527872889646e-05,
+      "loss": 0.8558,
+      "step": 6685
+    },
+    {
+      "epoch": 8.274582560296846,
+      "grad_norm": 0.466796875,
+      "learning_rate": 9.930878599847821e-05,
+      "loss": 0.8717,
+      "step": 6690
+    },
+    {
+      "epoch": 8.280766852195423,
+      "grad_norm": 0.46484375,
+      "learning_rate": 9.916478614137383e-05,
+      "loss": 0.8627,
+      "step": 6695
+    },
+    {
+      "epoch": 8.286951144094001,
+      "grad_norm": 0.470703125,
+      "learning_rate": 9.902078801626636e-05,
+      "loss": 0.8667,
+      "step": 6700
+    },
+    {
+      "epoch": 8.293135435992578,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.887679192176712e-05,
+      "loss": 0.8584,
+      "step": 6705
+    },
+    {
+      "epoch": 8.299319727891156,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.873279815648318e-05,
+      "loss": 0.8681,
+      "step": 6710
+    },
+    {
+      "epoch": 8.305504019789733,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.858880701901682e-05,
+      "loss": 0.8734,
+      "step": 6715
+    },
+    {
+      "epoch": 8.311688311688311,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.844481880796491e-05,
+      "loss": 0.8738,
+      "step": 6720
+    },
+    {
+      "epoch": 8.317872603586888,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.830083382191819e-05,
+      "loss": 0.8735,
+      "step": 6725
+    },
+    {
+      "epoch": 8.324056895485468,
+      "grad_norm": 0.44921875,
+      "learning_rate": 9.815685235946068e-05,
+      "loss": 0.8685,
+      "step": 6730
+    },
+    {
+      "epoch": 8.330241187384045,
+      "grad_norm": 0.447265625,
+      "learning_rate": 9.801287471916919e-05,
+      "loss": 0.8737,
+      "step": 6735
+    },
+    {
+      "epoch": 8.336425479282623,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.786890119961253e-05,
+      "loss": 0.8647,
+      "step": 6740
+    },
+    {
+      "epoch": 8.3426097711812,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.772493209935099e-05,
+      "loss": 0.8701,
+      "step": 6745
+    },
+    {
+      "epoch": 8.348794063079778,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.758096771693573e-05,
+      "loss": 0.8589,
+      "step": 6750
+    },
+    {
+      "epoch": 8.354978354978355,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.743700835090804e-05,
+      "loss": 0.8674,
+      "step": 6755
+    },
+    {
+      "epoch": 8.361162646876933,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.729305429979887e-05,
+      "loss": 0.8682,
+      "step": 6760
+    },
+    {
+      "epoch": 8.36734693877551,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.714910586212816e-05,
+      "loss": 0.8732,
+      "step": 6765
+    },
+    {
+      "epoch": 8.373531230674088,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.700516333640415e-05,
+      "loss": 0.8845,
+      "step": 6770
+    },
+    {
+      "epoch": 8.379715522572665,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.686122702112285e-05,
+      "loss": 0.8646,
+      "step": 6775
+    },
+    {
+      "epoch": 8.385899814471243,
+      "grad_norm": 0.451171875,
+      "learning_rate": 9.671729721476746e-05,
+      "loss": 0.8635,
+      "step": 6780
+    },
+    {
+      "epoch": 8.39208410636982,
+      "grad_norm": 0.470703125,
+      "learning_rate": 9.657337421580759e-05,
+      "loss": 0.8722,
+      "step": 6785
+    },
+    {
+      "epoch": 8.398268398268398,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.642945832269874e-05,
+      "loss": 0.8505,
+      "step": 6790
+    },
+    {
+      "epoch": 8.404452690166975,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.628554983388173e-05,
+      "loss": 0.865,
+      "step": 6795
+    },
+    {
+      "epoch": 8.410636982065553,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.614164904778196e-05,
+      "loss": 0.8605,
+      "step": 6800
+    },
+    {
+      "epoch": 8.416821273964132,
+      "grad_norm": 0.431640625,
+      "learning_rate": 9.599775626280892e-05,
+      "loss": 0.867,
+      "step": 6805
+    },
+    {
+      "epoch": 8.42300556586271,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.585387177735547e-05,
+      "loss": 0.8689,
+      "step": 6810
+    },
+    {
+      "epoch": 8.429189857761287,
+      "grad_norm": 0.4765625,
+      "learning_rate": 9.570999588979728e-05,
+      "loss": 0.8721,
+      "step": 6815
+    },
+    {
+      "epoch": 8.435374149659864,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.556612889849214e-05,
+      "loss": 0.8601,
+      "step": 6820
+    },
+    {
+      "epoch": 8.441558441558442,
+      "grad_norm": 0.466796875,
+      "learning_rate": 9.542227110177945e-05,
+      "loss": 0.8559,
+      "step": 6825
+    },
+    {
+      "epoch": 8.44774273345702,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.527842279797953e-05,
+      "loss": 0.866,
+      "step": 6830
+    },
+    {
+      "epoch": 8.453927025355597,
+      "grad_norm": 0.447265625,
+      "learning_rate": 9.513458428539298e-05,
+      "loss": 0.8635,
+      "step": 6835
+    },
+    {
+      "epoch": 8.460111317254174,
+      "grad_norm": 0.466796875,
+      "learning_rate": 9.499075586230013e-05,
+      "loss": 0.8694,
+      "step": 6840
+    },
+    {
+      "epoch": 8.466295609152752,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.484693782696041e-05,
+      "loss": 0.8716,
+      "step": 6845
+    },
+    {
+      "epoch": 8.47247990105133,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.470313047761167e-05,
+      "loss": 0.8623,
+      "step": 6850
+    },
+    {
+      "epoch": 8.478664192949907,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.455933411246958e-05,
+      "loss": 0.8603,
+      "step": 6855
+    },
+    {
+      "epoch": 8.484848484848484,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.44155490297271e-05,
+      "loss": 0.863,
+      "step": 6860
+    },
+    {
+      "epoch": 8.491032776747062,
+      "grad_norm": 0.447265625,
+      "learning_rate": 9.427177552755374e-05,
+      "loss": 0.8601,
+      "step": 6865
+    },
+    {
+      "epoch": 8.49721706864564,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.412801390409497e-05,
+      "loss": 0.8738,
+      "step": 6870
+    },
+    {
+      "epoch": 8.503401360544217,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.398426445747171e-05,
+      "loss": 0.8707,
+      "step": 6875
+    },
+    {
+      "epoch": 8.509585652442794,
+      "grad_norm": 0.60546875,
+      "learning_rate": 9.38405274857796e-05,
+      "loss": 0.8667,
+      "step": 6880
+    },
+    {
+      "epoch": 8.515769944341374,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.369680328708836e-05,
+      "loss": 0.876,
+      "step": 6885
+    },
+    {
+      "epoch": 8.521954236239951,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.355309215944124e-05,
+      "loss": 0.8676,
+      "step": 6890
+    },
+    {
+      "epoch": 8.528138528138529,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.340939440085445e-05,
+      "loss": 0.8488,
+      "step": 6895
+    },
+    {
+      "epoch": 8.534322820037106,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.326571030931637e-05,
+      "loss": 0.8724,
+      "step": 6900
+    },
+    {
+      "epoch": 8.540507111935684,
+      "grad_norm": 0.447265625,
+      "learning_rate": 9.312204018278716e-05,
+      "loss": 0.8657,
+      "step": 6905
+    },
+    {
+      "epoch": 8.546691403834261,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.297838431919794e-05,
+      "loss": 0.8611,
+      "step": 6910
+    },
+    {
+      "epoch": 8.552875695732839,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.283474301645026e-05,
+      "loss": 0.8634,
+      "step": 6915
+    },
+    {
+      "epoch": 8.559059987631416,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.269111657241548e-05,
+      "loss": 0.8627,
+      "step": 6920
+    },
+    {
+      "epoch": 8.565244279529994,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.254750528493417e-05,
+      "loss": 0.8572,
+      "step": 6925
+    },
+    {
+      "epoch": 8.571428571428571,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.240390945181543e-05,
+      "loss": 0.8648,
+      "step": 6930
+    },
+    {
+      "epoch": 8.577612863327149,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.226032937083635e-05,
+      "loss": 0.8621,
+      "step": 6935
+    },
+    {
+      "epoch": 8.583797155225726,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.211676533974131e-05,
+      "loss": 0.8634,
+      "step": 6940
+    },
+    {
+      "epoch": 8.589981447124304,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.197321765624152e-05,
+      "loss": 0.8667,
+      "step": 6945
+    },
+    {
+      "epoch": 8.596165739022881,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.182968661801412e-05,
+      "loss": 0.8707,
+      "step": 6950
+    },
+    {
+      "epoch": 8.602350030921459,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.168617252270183e-05,
+      "loss": 0.8772,
+      "step": 6955
+    },
+    {
+      "epoch": 8.608534322820038,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.154267566791223e-05,
+      "loss": 0.8584,
+      "step": 6960
+    },
+    {
+      "epoch": 8.614718614718615,
+      "grad_norm": 0.470703125,
+      "learning_rate": 9.139919635121714e-05,
+      "loss": 0.8657,
+      "step": 6965
+    },
+    {
+      "epoch": 8.620902906617193,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.125573487015203e-05,
+      "loss": 0.8632,
+      "step": 6970
+    },
+    {
+      "epoch": 8.62708719851577,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.111229152221535e-05,
+      "loss": 0.8705,
+      "step": 6975
+    },
+    {
+      "epoch": 8.633271490414348,
+      "grad_norm": 0.458984375,
+      "learning_rate": 9.096886660486797e-05,
+      "loss": 0.869,
+      "step": 6980
+    },
+    {
+      "epoch": 8.639455782312925,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.082546041553253e-05,
+      "loss": 0.875,
+      "step": 6985
+    },
+    {
+      "epoch": 8.645640074211503,
+      "grad_norm": 0.470703125,
+      "learning_rate": 9.068207325159284e-05,
+      "loss": 0.8667,
+      "step": 6990
+    },
+    {
+      "epoch": 8.65182436611008,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.053870541039327e-05,
+      "loss": 0.8749,
+      "step": 6995
+    },
+    {
+      "epoch": 8.658008658008658,
+      "grad_norm": 0.451171875,
+      "learning_rate": 9.039535718923804e-05,
+      "loss": 0.8608,
+      "step": 7000
+    },
+    {
+      "epoch": 8.664192949907235,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.02520288853908e-05,
+      "loss": 0.8621,
+      "step": 7005
+    },
+    {
+      "epoch": 8.670377241805813,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.01087207960739e-05,
+      "loss": 0.8716,
+      "step": 7010
+    },
+    {
+      "epoch": 8.67656153370439,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.996543321846759e-05,
+      "loss": 0.8588,
+      "step": 7015
+    },
+    {
+      "epoch": 8.682745825602968,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.982216644970979e-05,
+      "loss": 0.8595,
+      "step": 7020
+    },
+    {
+      "epoch": 8.688930117501545,
+      "grad_norm": 0.46484375,
+      "learning_rate": 8.967892078689513e-05,
+      "loss": 0.8813,
+      "step": 7025
+    },
+    {
+      "epoch": 8.695114409400123,
+      "grad_norm": 0.435546875,
+      "learning_rate": 8.953569652707459e-05,
+      "loss": 0.8618,
+      "step": 7030
+    },
+    {
+      "epoch": 8.7012987012987,
+      "grad_norm": 0.44921875,
+      "learning_rate": 8.939249396725467e-05,
+      "loss": 0.8558,
+      "step": 7035
+    },
+    {
+      "epoch": 8.70748299319728,
+      "grad_norm": 0.4765625,
+      "learning_rate": 8.924931340439694e-05,
+      "loss": 0.8712,
+      "step": 7040
+    },
+    {
+      "epoch": 8.713667285095857,
+      "grad_norm": 0.44921875,
+      "learning_rate": 8.910615513541729e-05,
+      "loss": 0.8496,
+      "step": 7045
+    },
+    {
+      "epoch": 8.719851576994435,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.896301945718541e-05,
+      "loss": 0.8614,
+      "step": 7050
+    },
+    {
+      "epoch": 8.726035868893012,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.881990666652417e-05,
+      "loss": 0.8733,
+      "step": 7055
+    },
+    {
+      "epoch": 8.73222016079159,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.867681706020894e-05,
+      "loss": 0.8747,
+      "step": 7060
+    },
+    {
+      "epoch": 8.738404452690167,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.853375093496699e-05,
+      "loss": 0.8717,
+      "step": 7065
+    },
+    {
+      "epoch": 8.744588744588745,
+      "grad_norm": 0.458984375,
+      "learning_rate": 8.839070858747697e-05,
+      "loss": 0.8632,
+      "step": 7070
+    },
+    {
+      "epoch": 8.750773036487322,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.824769031436822e-05,
+      "loss": 0.867,
+      "step": 7075
+    },
+    {
+      "epoch": 8.7569573283859,
+      "grad_norm": 0.4921875,
+      "learning_rate": 8.810469641222001e-05,
+      "loss": 0.8627,
+      "step": 7080
+    },
+    {
+      "epoch": 8.763141620284477,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.796172717756124e-05,
+      "loss": 0.8723,
+      "step": 7085
+    },
+    {
+      "epoch": 8.769325912183055,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.781878290686959e-05,
+      "loss": 0.878,
+      "step": 7090
+    },
+    {
+      "epoch": 8.775510204081632,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.767586389657098e-05,
+      "loss": 0.8723,
+      "step": 7095
+    },
+    {
+      "epoch": 8.78169449598021,
+      "grad_norm": 0.49609375,
+      "learning_rate": 8.753297044303896e-05,
+      "loss": 0.8715,
+      "step": 7100
+    },
+    {
+      "epoch": 8.787878787878787,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.739010284259406e-05,
+      "loss": 0.8626,
+      "step": 7105
+    },
+    {
+      "epoch": 8.794063079777365,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.724726139150318e-05,
+      "loss": 0.8616,
+      "step": 7110
+    },
+    {
+      "epoch": 8.800247371675944,
+      "grad_norm": 0.4375,
+      "learning_rate": 8.710444638597905e-05,
+      "loss": 0.8726,
+      "step": 7115
+    },
+    {
+      "epoch": 8.806431663574521,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.696165812217953e-05,
+      "loss": 0.8644,
+      "step": 7120
+    },
+    {
+      "epoch": 8.812615955473099,
+      "grad_norm": 0.447265625,
+      "learning_rate": 8.6818896896207e-05,
+      "loss": 0.86,
+      "step": 7125
+    },
+    {
+      "epoch": 8.818800247371676,
+      "grad_norm": 0.47265625,
+      "learning_rate": 8.667616300410778e-05,
+      "loss": 0.871,
+      "step": 7130
+    },
+    {
+      "epoch": 8.824984539270254,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.653345674187157e-05,
+      "loss": 0.8669,
+      "step": 7135
+    },
+    {
+      "epoch": 8.831168831168831,
+      "grad_norm": 0.458984375,
+      "learning_rate": 8.639077840543077e-05,
+      "loss": 0.8533,
+      "step": 7140
+    },
+    {
+      "epoch": 8.837353123067409,
+      "grad_norm": 0.44921875,
+      "learning_rate": 8.62481282906597e-05,
+      "loss": 0.8853,
+      "step": 7145
+    },
+    {
+      "epoch": 8.843537414965986,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.610550669337433e-05,
+      "loss": 0.864,
+      "step": 7150
+    },
+    {
+      "epoch": 8.849721706864564,
+      "grad_norm": 0.4375,
+      "learning_rate": 8.596291390933147e-05,
+      "loss": 0.8616,
+      "step": 7155
+    },
+    {
+      "epoch": 8.855905998763141,
+      "grad_norm": 0.4375,
+      "learning_rate": 8.582035023422815e-05,
+      "loss": 0.8634,
+      "step": 7160
+    },
+    {
+      "epoch": 8.862090290661719,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.567781596370104e-05,
+      "loss": 0.872,
+      "step": 7165
+    },
+    {
+      "epoch": 8.868274582560296,
+      "grad_norm": 0.42578125,
+      "learning_rate": 8.553531139332582e-05,
+      "loss": 0.8525,
+      "step": 7170
+    },
+    {
+      "epoch": 8.874458874458874,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.539283681861661e-05,
+      "loss": 0.8559,
+      "step": 7175
+    },
+    {
+      "epoch": 8.880643166357451,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.525039253502529e-05,
+      "loss": 0.8702,
+      "step": 7180
+    },
+    {
+      "epoch": 8.886827458256029,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.510797883794097e-05,
+      "loss": 0.867,
+      "step": 7185
+    },
+    {
+      "epoch": 8.893011750154606,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.496559602268928e-05,
+      "loss": 0.8573,
+      "step": 7190
+    },
+    {
+      "epoch": 8.899196042053186,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.482324438453187e-05,
+      "loss": 0.8687,
+      "step": 7195
+    },
+    {
+      "epoch": 8.905380333951763,
+      "grad_norm": 0.4765625,
+      "learning_rate": 8.468092421866573e-05,
+      "loss": 0.8658,
+      "step": 7200
+    },
+    {
+      "epoch": 8.91156462585034,
+      "grad_norm": 0.44921875,
+      "learning_rate": 8.45386358202225e-05,
+      "loss": 0.8594,
+      "step": 7205
+    },
+    {
+      "epoch": 8.917748917748918,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.439637948426801e-05,
+      "loss": 0.8699,
+      "step": 7210
+    },
+    {
+      "epoch": 8.923933209647496,
+      "grad_norm": 0.447265625,
+      "learning_rate": 8.425415550580162e-05,
+      "loss": 0.865,
+      "step": 7215
+    },
+    {
+      "epoch": 8.930117501546073,
+      "grad_norm": 0.4453125,
+      "learning_rate": 8.411196417975558e-05,
+      "loss": 0.8646,
+      "step": 7220
+    },
+    {
+      "epoch": 8.93630179344465,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.396980580099438e-05,
+      "loss": 0.8553,
+      "step": 7225
+    },
+    {
+      "epoch": 8.942486085343228,
+      "grad_norm": 0.478515625,
+      "learning_rate": 8.382768066431425e-05,
+      "loss": 0.8629,
+      "step": 7230
+    },
+    {
+      "epoch": 8.948670377241806,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.368558906444244e-05,
+      "loss": 0.8691,
+      "step": 7235
+    },
+    {
+      "epoch": 8.954854669140383,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.354353129603668e-05,
+      "loss": 0.8654,
+      "step": 7240
+    },
+    {
+      "epoch": 8.96103896103896,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.340150765368452e-05,
+      "loss": 0.8607,
+      "step": 7245
+    },
+    {
+      "epoch": 8.967223252937538,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.325951843190274e-05,
+      "loss": 0.8741,
+      "step": 7250
+    },
+    {
+      "epoch": 8.973407544836116,
+      "grad_norm": 0.447265625,
+      "learning_rate": 8.311756392513681e-05,
+      "loss": 0.8696,
+      "step": 7255
+    },
+    {
+      "epoch": 8.979591836734693,
+      "grad_norm": 0.59375,
+      "learning_rate": 8.297564442776014e-05,
+      "loss": 0.876,
+      "step": 7260
+    },
+    {
+      "epoch": 8.98577612863327,
+      "grad_norm": 0.4921875,
+      "learning_rate": 8.283376023407357e-05,
+      "loss": 0.8813,
+      "step": 7265
+    },
+    {
+      "epoch": 8.99196042053185,
+      "grad_norm": 0.447265625,
+      "learning_rate": 8.269191163830467e-05,
+      "loss": 0.8682,
+      "step": 7270
+    },
+    {
+      "epoch": 8.998144712430427,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.255009893460724e-05,
+      "loss": 0.8608,
+      "step": 7275
+    },
+    {
+      "epoch": 8.999381570810142,
+      "eval_loss": 2.514324903488159,
+      "eval_runtime": 0.6521,
+      "eval_samples_per_second": 15.335,
+      "eval_steps_per_second": 1.534,
+      "step": 7276
+    },
+    {
+      "epoch": 9.004329004329005,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.240832241706068e-05,
+      "loss": 0.8585,
+      "step": 7280
+    },
+    {
+      "epoch": 9.010513296227582,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.226658237966933e-05,
+      "loss": 0.8596,
+      "step": 7285
+    },
+    {
+      "epoch": 9.01669758812616,
+      "grad_norm": 0.466796875,
+      "learning_rate": 8.212487911636184e-05,
+      "loss": 0.8587,
+      "step": 7290
+    },
+    {
+      "epoch": 9.022881880024737,
+      "grad_norm": 0.48828125,
+      "learning_rate": 8.198321292099064e-05,
+      "loss": 0.8649,
+      "step": 7295
+    },
+    {
+      "epoch": 9.029066171923315,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.184158408733131e-05,
+      "loss": 0.8721,
+      "step": 7300
+    },
+    {
+      "epoch": 9.035250463821892,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.169999290908188e-05,
+      "loss": 0.861,
+      "step": 7305
+    },
+    {
+      "epoch": 9.04143475572047,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.155843967986236e-05,
+      "loss": 0.8489,
+      "step": 7310
+    },
+    {
+      "epoch": 9.047619047619047,
+      "grad_norm": 0.455078125,
+      "learning_rate": 8.141692469321403e-05,
+      "loss": 0.8582,
+      "step": 7315
+    },
+    {
+      "epoch": 9.053803339517625,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.127544824259889e-05,
+      "loss": 0.8654,
+      "step": 7320
+    },
+    {
+      "epoch": 9.059987631416202,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.113401062139901e-05,
+      "loss": 0.8545,
+      "step": 7325
+    },
+    {
+      "epoch": 9.06617192331478,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.099261212291601e-05,
+      "loss": 0.8645,
+      "step": 7330
+    },
+    {
+      "epoch": 9.072356215213357,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.085125304037018e-05,
+      "loss": 0.869,
+      "step": 7335
+    },
+    {
+      "epoch": 9.078540507111935,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.070993366690029e-05,
+      "loss": 0.8699,
+      "step": 7340
+    },
+    {
+      "epoch": 9.084724799010512,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.056865429556267e-05,
+      "loss": 0.8558,
+      "step": 7345
+    },
+    {
+      "epoch": 9.090909090909092,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.042741521933071e-05,
+      "loss": 0.8524,
+      "step": 7350
+    },
+    {
+      "epoch": 9.09709338280767,
+      "grad_norm": 0.474609375,
+      "learning_rate": 8.028621673109425e-05,
+      "loss": 0.8664,
+      "step": 7355
+    },
+    {
+      "epoch": 9.103277674706247,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.014505912365893e-05,
+      "loss": 0.8638,
+      "step": 7360
+    },
+    {
+      "epoch": 9.109461966604824,
+      "grad_norm": 0.435546875,
+      "learning_rate": 8.000394268974563e-05,
+      "loss": 0.8599,
+      "step": 7365
+    },
+    {
+      "epoch": 9.115646258503402,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.986286772198986e-05,
+      "loss": 0.8649,
+      "step": 7370
+    },
+    {
+      "epoch": 9.12183055040198,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.972183451294112e-05,
+      "loss": 0.8691,
+      "step": 7375
+    },
+    {
+      "epoch": 9.128014842300557,
+      "grad_norm": 0.458984375,
+      "learning_rate": 7.958084335506239e-05,
+      "loss": 0.8663,
+      "step": 7380
+    },
+    {
+      "epoch": 9.134199134199134,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.943989454072931e-05,
+      "loss": 0.8588,
+      "step": 7385
+    },
+    {
+      "epoch": 9.140383426097712,
+      "grad_norm": 0.462890625,
+      "learning_rate": 7.929898836222983e-05,
+      "loss": 0.8668,
+      "step": 7390
+    },
+    {
+      "epoch": 9.14656771799629,
+      "grad_norm": 0.4609375,
+      "learning_rate": 7.915812511176347e-05,
+      "loss": 0.8612,
+      "step": 7395
+    },
+    {
+      "epoch": 9.152752009894867,
+      "grad_norm": 0.462890625,
+      "learning_rate": 7.90173050814406e-05,
+      "loss": 0.8627,
+      "step": 7400
+    },
+    {
+      "epoch": 9.158936301793444,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.887652856328214e-05,
+      "loss": 0.8531,
+      "step": 7405
+    },
+    {
+      "epoch": 9.165120593692022,
+      "grad_norm": 0.4609375,
+      "learning_rate": 7.873579584921869e-05,
+      "loss": 0.8705,
+      "step": 7410
+    },
+    {
+      "epoch": 9.1713048855906,
+      "grad_norm": 0.45703125,
+      "learning_rate": 7.859510723109003e-05,
+      "loss": 0.8737,
+      "step": 7415
+    },
+    {
+      "epoch": 9.177489177489177,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.84544630006445e-05,
+      "loss": 0.8651,
+      "step": 7420
+    },
+    {
+      "epoch": 9.183673469387756,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.831386344953836e-05,
+      "loss": 0.8513,
+      "step": 7425
+    },
+    {
+      "epoch": 9.189857761286333,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.817330886933527e-05,
+      "loss": 0.8665,
+      "step": 7430
+    },
+    {
+      "epoch": 9.196042053184911,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.803279955150558e-05,
+      "loss": 0.8645,
+      "step": 7435
+    },
+    {
+      "epoch": 9.202226345083488,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.789233578742582e-05,
+      "loss": 0.8693,
+      "step": 7440
+    },
+    {
+      "epoch": 9.208410636982066,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.775191786837807e-05,
+      "loss": 0.8573,
+      "step": 7445
+    },
+    {
+      "epoch": 9.214594928880643,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.761154608554927e-05,
+      "loss": 0.8614,
+      "step": 7450
+    },
+    {
+      "epoch": 9.220779220779221,
+      "grad_norm": 0.5078125,
+      "learning_rate": 7.747122073003075e-05,
+      "loss": 0.8628,
+      "step": 7455
+    },
+    {
+      "epoch": 9.226963512677798,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.733094209281756e-05,
+      "loss": 0.8563,
+      "step": 7460
+    },
+    {
+      "epoch": 9.233147804576376,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.719071046480776e-05,
+      "loss": 0.8668,
+      "step": 7465
+    },
+    {
+      "epoch": 9.239332096474953,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.705052613680211e-05,
+      "loss": 0.8607,
+      "step": 7470
+    },
+    {
+      "epoch": 9.245516388373531,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.691038939950316e-05,
+      "loss": 0.8543,
+      "step": 7475
+    },
+    {
+      "epoch": 9.251700680272108,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.677030054351477e-05,
+      "loss": 0.858,
+      "step": 7480
+    },
+    {
+      "epoch": 9.257884972170686,
+      "grad_norm": 0.419921875,
+      "learning_rate": 7.663025985934158e-05,
+      "loss": 0.8605,
+      "step": 7485
+    },
+    {
+      "epoch": 9.264069264069263,
+      "grad_norm": 0.4609375,
+      "learning_rate": 7.649026763738827e-05,
+      "loss": 0.8628,
+      "step": 7490
+    },
+    {
+      "epoch": 9.270253555967841,
+      "grad_norm": 0.48046875,
+      "learning_rate": 7.635032416795905e-05,
+      "loss": 0.8618,
+      "step": 7495
+    },
+    {
+      "epoch": 9.276437847866418,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.6210429741257e-05,
+      "loss": 0.8675,
+      "step": 7500
+    },
+    {
+      "epoch": 9.282622139764998,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.607058464738357e-05,
+      "loss": 0.8614,
+      "step": 7505
+    },
+    {
+      "epoch": 9.288806431663575,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.593078917633787e-05,
+      "loss": 0.8658,
+      "step": 7510
+    },
+    {
+      "epoch": 9.294990723562153,
+      "grad_norm": 0.443359375,
+      "learning_rate": 7.579104361801605e-05,
+      "loss": 0.8552,
+      "step": 7515
+    },
+    {
+      "epoch": 9.30117501546073,
+      "grad_norm": 0.4765625,
+      "learning_rate": 7.565134826221083e-05,
+      "loss": 0.8667,
+      "step": 7520
+    },
+    {
+      "epoch": 9.307359307359308,
+      "grad_norm": 0.4453125,
+      "learning_rate": 7.551170339861083e-05,
+      "loss": 0.8498,
+      "step": 7525
+    },
+    {
+      "epoch": 9.313543599257885,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.537210931679987e-05,
+      "loss": 0.8498,
+      "step": 7530
+    },
+    {
+      "epoch": 9.319727891156463,
+      "grad_norm": 0.466796875,
+      "learning_rate": 7.523256630625657e-05,
+      "loss": 0.8632,
+      "step": 7535
+    },
+    {
+      "epoch": 9.32591218305504,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.509307465635358e-05,
+      "loss": 0.8497,
+      "step": 7540
+    },
+    {
+      "epoch": 9.332096474953618,
+      "grad_norm": 0.466796875,
+      "learning_rate": 7.495363465635708e-05,
+      "loss": 0.8685,
+      "step": 7545
+    },
+    {
+      "epoch": 9.338280766852195,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.481424659542609e-05,
+      "loss": 0.8521,
+      "step": 7550
+    },
+    {
+      "epoch": 9.344465058750773,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.467491076261197e-05,
+      "loss": 0.8644,
+      "step": 7555
+    },
+    {
+      "epoch": 9.35064935064935,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.453562744685778e-05,
+      "loss": 0.8676,
+      "step": 7560
+    },
+    {
+      "epoch": 9.356833642547928,
+      "grad_norm": 0.44140625,
+      "learning_rate": 7.439639693699763e-05,
+      "loss": 0.8689,
+      "step": 7565
+    },
+    {
+      "epoch": 9.363017934446505,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.425721952175618e-05,
+      "loss": 0.8636,
+      "step": 7570
+    },
+    {
+      "epoch": 9.369202226345083,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.411809548974792e-05,
+      "loss": 0.8494,
+      "step": 7575
+    },
+    {
+      "epoch": 9.375386518243662,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.39790251294767e-05,
+      "loss": 0.8693,
+      "step": 7580
+    },
+    {
+      "epoch": 9.38157081014224,
+      "grad_norm": 0.45703125,
+      "learning_rate": 7.384000872933506e-05,
+      "loss": 0.8667,
+      "step": 7585
+    },
+    {
+      "epoch": 9.387755102040817,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.370104657760361e-05,
+      "loss": 0.8572,
+      "step": 7590
+    },
+    {
+      "epoch": 9.393939393939394,
+      "grad_norm": 0.47265625,
+      "learning_rate": 7.356213896245046e-05,
+      "loss": 0.8636,
+      "step": 7595
+    },
+    {
+      "epoch": 9.400123685837972,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.342328617193067e-05,
+      "loss": 0.8729,
+      "step": 7600
+    },
+    {
+      "epoch": 9.40630797773655,
+      "grad_norm": 0.435546875,
+      "learning_rate": 7.328448849398558e-05,
+      "loss": 0.8612,
+      "step": 7605
+    },
+    {
+      "epoch": 9.412492269635127,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.314574621644225e-05,
+      "loss": 0.8672,
+      "step": 7610
+    },
+    {
+      "epoch": 9.418676561533704,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.300705962701287e-05,
+      "loss": 0.8604,
+      "step": 7615
+    },
+    {
+      "epoch": 9.424860853432282,
+      "grad_norm": 0.4921875,
+      "learning_rate": 7.286842901329412e-05,
+      "loss": 0.854,
+      "step": 7620
+    },
+    {
+      "epoch": 9.43104514533086,
+      "grad_norm": 0.458984375,
+      "learning_rate": 7.272985466276661e-05,
+      "loss": 0.8513,
+      "step": 7625
+    },
+    {
+      "epoch": 9.437229437229437,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.259133686279429e-05,
+      "loss": 0.8654,
+      "step": 7630
+    },
+    {
+      "epoch": 9.443413729128014,
+      "grad_norm": 0.443359375,
+      "learning_rate": 7.245287590062384e-05,
+      "loss": 0.8533,
+      "step": 7635
+    },
+    {
+      "epoch": 9.449598021026592,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.231447206338407e-05,
+      "loss": 0.8777,
+      "step": 7640
+    },
+    {
+      "epoch": 9.45578231292517,
+      "grad_norm": 0.4453125,
+      "learning_rate": 7.217612563808528e-05,
+      "loss": 0.8644,
+      "step": 7645
+    },
+    {
+      "epoch": 9.461966604823747,
+      "grad_norm": 0.4453125,
+      "learning_rate": 7.203783691161883e-05,
+      "loss": 0.8646,
+      "step": 7650
+    },
+    {
+      "epoch": 9.468150896722324,
+      "grad_norm": 0.4453125,
+      "learning_rate": 7.189960617075633e-05,
+      "loss": 0.8791,
+      "step": 7655
+    },
+    {
+      "epoch": 9.474335188620904,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.176143370214914e-05,
+      "loss": 0.8591,
+      "step": 7660
+    },
+    {
+      "epoch": 9.480519480519481,
+      "grad_norm": 0.45703125,
+      "learning_rate": 7.162331979232783e-05,
+      "loss": 0.8641,
+      "step": 7665
+    },
+    {
+      "epoch": 9.486703772418059,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.148526472770154e-05,
+      "loss": 0.8763,
+      "step": 7670
+    },
+    {
+      "epoch": 9.492888064316636,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.134726879455734e-05,
+      "loss": 0.8668,
+      "step": 7675
+    },
+    {
+      "epoch": 9.499072356215214,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.12093322790597e-05,
+      "loss": 0.8619,
+      "step": 7680
+    },
+    {
+      "epoch": 9.505256648113791,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.107145546724989e-05,
+      "loss": 0.8695,
+      "step": 7685
+    },
+    {
+      "epoch": 9.511440940012369,
+      "grad_norm": 0.4921875,
+      "learning_rate": 7.09336386450453e-05,
+      "loss": 0.8523,
+      "step": 7690
+    },
+    {
+      "epoch": 9.517625231910946,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.079588209823906e-05,
+      "loss": 0.8676,
+      "step": 7695
+    },
+    {
+      "epoch": 9.523809523809524,
+      "grad_norm": 0.44140625,
+      "learning_rate": 7.065818611249915e-05,
+      "loss": 0.8555,
+      "step": 7700
+    },
+    {
+      "epoch": 9.529993815708101,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.052055097336805e-05,
+      "loss": 0.8612,
+      "step": 7705
+    },
+    {
+      "epoch": 9.536178107606679,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.038297696626206e-05,
+      "loss": 0.8602,
+      "step": 7710
+    },
+    {
+      "epoch": 9.542362399505256,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.02454643764707e-05,
+      "loss": 0.8636,
+      "step": 7715
+    },
+    {
+      "epoch": 9.548546691403834,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.010801348915608e-05,
+      "loss": 0.8566,
+      "step": 7720
+    },
+    {
+      "epoch": 9.554730983302411,
+      "grad_norm": 0.45703125,
+      "learning_rate": 6.99706245893524e-05,
+      "loss": 0.8603,
+      "step": 7725
+    },
+    {
+      "epoch": 9.560915275200989,
+      "grad_norm": 0.515625,
+      "learning_rate": 6.983329796196534e-05,
+      "loss": 0.858,
+      "step": 7730
+    },
+    {
+      "epoch": 9.567099567099568,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.969603389177142e-05,
+      "loss": 0.858,
+      "step": 7735
+    },
+    {
+      "epoch": 9.573283858998145,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.955883266341741e-05,
+      "loss": 0.8738,
+      "step": 7740
+    },
+    {
+      "epoch": 9.579468150896723,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.94216945614198e-05,
+      "loss": 0.8643,
+      "step": 7745
+    },
+    {
+      "epoch": 9.5856524427953,
+      "grad_norm": 0.494140625,
+      "learning_rate": 6.928461987016413e-05,
+      "loss": 0.8708,
+      "step": 7750
+    },
+    {
+      "epoch": 9.591836734693878,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.914760887390452e-05,
+      "loss": 0.8582,
+      "step": 7755
+    },
+    {
+      "epoch": 9.598021026592455,
+      "grad_norm": 0.42578125,
+      "learning_rate": 6.901066185676295e-05,
+      "loss": 0.8593,
+      "step": 7760
+    },
+    {
+      "epoch": 9.604205318491033,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.887377910272869e-05,
+      "loss": 0.8639,
+      "step": 7765
+    },
+    {
+      "epoch": 9.61038961038961,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.873696089565786e-05,
+      "loss": 0.8507,
+      "step": 7770
+    },
+    {
+      "epoch": 9.616573902288188,
+      "grad_norm": 0.45703125,
+      "learning_rate": 6.860020751927259e-05,
+      "loss": 0.8571,
+      "step": 7775
+    },
+    {
+      "epoch": 9.622758194186765,
+      "grad_norm": 0.474609375,
+      "learning_rate": 6.846351925716068e-05,
+      "loss": 0.8637,
+      "step": 7780
+    },
+    {
+      "epoch": 9.628942486085343,
+      "grad_norm": 0.443359375,
+      "learning_rate": 6.832689639277484e-05,
+      "loss": 0.8621,
+      "step": 7785
+    },
+    {
+      "epoch": 9.63512677798392,
+      "grad_norm": 0.419921875,
+      "learning_rate": 6.819033920943219e-05,
+      "loss": 0.8631,
+      "step": 7790
+    },
+    {
+      "epoch": 9.641311069882498,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.805384799031361e-05,
+      "loss": 0.8645,
+      "step": 7795
+    },
+    {
+      "epoch": 9.647495361781075,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.791742301846326e-05,
+      "loss": 0.8617,
+      "step": 7800
+    },
+    {
+      "epoch": 9.653679653679653,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.778106457678785e-05,
+      "loss": 0.8613,
+      "step": 7805
+    },
+    {
+      "epoch": 9.65986394557823,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.764477294805615e-05,
+      "loss": 0.8543,
+      "step": 7810
+    },
+    {
+      "epoch": 9.666048237476808,
+      "grad_norm": 0.42578125,
+      "learning_rate": 6.750854841489842e-05,
+      "loss": 0.8599,
+      "step": 7815
+    },
+    {
+      "epoch": 9.672232529375387,
+      "grad_norm": 0.427734375,
+      "learning_rate": 6.737239125980573e-05,
+      "loss": 0.8622,
+      "step": 7820
+    },
+    {
+      "epoch": 9.678416821273965,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.723630176512944e-05,
+      "loss": 0.8587,
+      "step": 7825
+    },
+    {
+      "epoch": 9.684601113172542,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.710028021308061e-05,
+      "loss": 0.8655,
+      "step": 7830
+    },
+    {
+      "epoch": 9.69078540507112,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.696432688572937e-05,
+      "loss": 0.8626,
+      "step": 7835
+    },
+    {
+      "epoch": 9.696969696969697,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.682844206500445e-05,
+      "loss": 0.8685,
+      "step": 7840
+    },
+    {
+      "epoch": 9.703153988868275,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.669262603269246e-05,
+      "loss": 0.8599,
+      "step": 7845
+    },
+    {
+      "epoch": 9.709338280766852,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.655687907043734e-05,
+      "loss": 0.8549,
+      "step": 7850
+    },
+    {
+      "epoch": 9.71552257266543,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.642120145973985e-05,
+      "loss": 0.8692,
+      "step": 7855
+    },
+    {
+      "epoch": 9.721706864564007,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.62855934819569e-05,
+      "loss": 0.8663,
+      "step": 7860
+    },
+    {
+      "epoch": 9.727891156462585,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.615005541830103e-05,
+      "loss": 0.8571,
+      "step": 7865
+    },
+    {
+      "epoch": 9.734075448361162,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.601458754983978e-05,
+      "loss": 0.862,
+      "step": 7870
+    },
+    {
+      "epoch": 9.74025974025974,
+      "grad_norm": 0.419921875,
+      "learning_rate": 6.587919015749511e-05,
+      "loss": 0.8637,
+      "step": 7875
+    },
+    {
+      "epoch": 9.746444032158317,
+      "grad_norm": 0.4609375,
+      "learning_rate": 6.574386352204289e-05,
+      "loss": 0.8586,
+      "step": 7880
+    },
+    {
+      "epoch": 9.752628324056895,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.560860792411219e-05,
+      "loss": 0.8576,
+      "step": 7885
+    },
+    {
+      "epoch": 9.758812615955474,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.547342364418481e-05,
+      "loss": 0.8503,
+      "step": 7890
+    },
+    {
+      "epoch": 9.764996907854051,
+      "grad_norm": 0.419921875,
+      "learning_rate": 6.533831096259467e-05,
+      "loss": 0.8615,
+      "step": 7895
+    },
+    {
+      "epoch": 9.771181199752629,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.520327015952713e-05,
+      "loss": 0.8564,
+      "step": 7900
+    },
+    {
+      "epoch": 9.777365491651206,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.506830151501861e-05,
+      "loss": 0.8571,
+      "step": 7905
+    },
+    {
+      "epoch": 9.783549783549784,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.493340530895583e-05,
+      "loss": 0.8671,
+      "step": 7910
+    },
+    {
+      "epoch": 9.789734075448361,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.479858182107527e-05,
+      "loss": 0.8657,
+      "step": 7915
+    },
+    {
+      "epoch": 9.795918367346939,
+      "grad_norm": 0.47265625,
+      "learning_rate": 6.466383133096267e-05,
+      "loss": 0.8556,
+      "step": 7920
+    },
+    {
+      "epoch": 9.802102659245516,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.452915411805238e-05,
+      "loss": 0.8502,
+      "step": 7925
+    },
+    {
+      "epoch": 9.808286951144094,
+      "grad_norm": 0.42578125,
+      "learning_rate": 6.439455046162677e-05,
+      "loss": 0.8599,
+      "step": 7930
+    },
+    {
+      "epoch": 9.814471243042671,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.426002064081565e-05,
+      "loss": 0.8641,
+      "step": 7935
+    },
+    {
+      "epoch": 9.820655534941249,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.412556493459581e-05,
+      "loss": 0.8644,
+      "step": 7940
+    },
+    {
+      "epoch": 9.826839826839826,
+      "grad_norm": 0.43359375,
+      "learning_rate": 6.399118362179028e-05,
+      "loss": 0.8574,
+      "step": 7945
+    },
+    {
+      "epoch": 9.833024118738404,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.385687698106781e-05,
+      "loss": 0.8651,
+      "step": 7950
+    },
+    {
+      "epoch": 9.839208410636981,
+      "grad_norm": 0.494140625,
+      "learning_rate": 6.372264529094233e-05,
+      "loss": 0.8756,
+      "step": 7955
+    },
+    {
+      "epoch": 9.845392702535559,
+      "grad_norm": 0.46484375,
+      "learning_rate": 6.358848882977233e-05,
+      "loss": 0.8606,
+      "step": 7960
+    },
+    {
+      "epoch": 9.851576994434136,
+      "grad_norm": 0.470703125,
+      "learning_rate": 6.345440787576031e-05,
+      "loss": 0.8615,
+      "step": 7965
+    },
+    {
+      "epoch": 9.857761286332714,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.332040270695219e-05,
+      "loss": 0.8724,
+      "step": 7970
+    },
+    {
+      "epoch": 9.863945578231293,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.31864736012367e-05,
+      "loss": 0.8641,
+      "step": 7975
+    },
+    {
+      "epoch": 9.87012987012987,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.305262083634488e-05,
+      "loss": 0.8693,
+      "step": 7980
+    },
+    {
+      "epoch": 9.876314162028448,
+      "grad_norm": 0.43359375,
+      "learning_rate": 6.291884468984941e-05,
+      "loss": 0.8602,
+      "step": 7985
+    },
+    {
+      "epoch": 9.882498453927026,
+      "grad_norm": 0.443359375,
+      "learning_rate": 6.278514543916415e-05,
+      "loss": 0.8574,
+      "step": 7990
+    },
+    {
+      "epoch": 9.888682745825603,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.265152336154345e-05,
+      "loss": 0.8529,
+      "step": 7995
+    },
+    {
+      "epoch": 9.89486703772418,
+      "grad_norm": 0.482421875,
+      "learning_rate": 6.251797873408161e-05,
+      "loss": 0.8701,
+      "step": 8000
+    },
+    {
+      "epoch": 9.901051329622758,
+      "grad_norm": 0.45703125,
+      "learning_rate": 6.238451183371241e-05,
+      "loss": 0.8587,
+      "step": 8005
+    },
+    {
+      "epoch": 9.907235621521336,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.225112293720836e-05,
+      "loss": 0.858,
+      "step": 8010
+    },
+    {
+      "epoch": 9.913419913419913,
+      "grad_norm": 0.466796875,
+      "learning_rate": 6.211781232118025e-05,
+      "loss": 0.8736,
+      "step": 8015
+    },
+    {
+      "epoch": 9.91960420531849,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.198458026207652e-05,
+      "loss": 0.8559,
+      "step": 8020
+    },
+    {
+      "epoch": 9.925788497217068,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.18514270361827e-05,
+      "loss": 0.8657,
+      "step": 8025
+    },
+    {
+      "epoch": 9.931972789115646,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.171835291962088e-05,
+      "loss": 0.8569,
+      "step": 8030
+    },
+    {
+      "epoch": 9.938157081014223,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.158535818834906e-05,
+      "loss": 0.8585,
+      "step": 8035
+    },
+    {
+      "epoch": 9.9443413729128,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.145244311816063e-05,
+      "loss": 0.863,
+      "step": 8040
+    },
+    {
+      "epoch": 9.95052566481138,
+      "grad_norm": 0.46484375,
+      "learning_rate": 6.13196079846838e-05,
+      "loss": 0.8619,
+      "step": 8045
+    },
+    {
+      "epoch": 9.956709956709958,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.1186853063381e-05,
+      "loss": 0.8664,
+      "step": 8050
+    },
+    {
+      "epoch": 9.962894248608535,
+      "grad_norm": 0.43359375,
+      "learning_rate": 6.105417862954828e-05,
+      "loss": 0.8636,
+      "step": 8055
+    },
+    {
+      "epoch": 9.969078540507113,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.092158495831486e-05,
+      "loss": 0.8536,
+      "step": 8060
+    },
+    {
+      "epoch": 9.97526283240569,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.078907232464248e-05,
+      "loss": 0.8701,
+      "step": 8065
+    },
+    {
+      "epoch": 9.981447124304268,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.065664100332478e-05,
+      "loss": 0.8686,
+      "step": 8070
+    },
+    {
+      "epoch": 9.987631416202845,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.0524291268986766e-05,
+      "loss": 0.8626,
+      "step": 8075
+    },
+    {
+      "epoch": 9.993815708101423,
+      "grad_norm": 0.419921875,
+      "learning_rate": 6.039202339608432e-05,
+      "loss": 0.8695,
+      "step": 8080
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.462890625,
+      "learning_rate": 6.025983765890353e-05,
+      "loss": 0.8421,
+      "step": 8085
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 2.518721103668213,
+      "eval_runtime": 0.5383,
+      "eval_samples_per_second": 18.576,
+      "eval_steps_per_second": 1.858,
+      "step": 8085
+    },
+    {
+      "epoch": 10.006184291898577,
+      "grad_norm": 0.443359375,
+      "learning_rate": 6.012773433156017e-05,
+      "loss": 0.8602,
+      "step": 8090
+    },
+    {
+      "epoch": 10.012368583797155,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.99957136879991e-05,
+      "loss": 0.8548,
+      "step": 8095
+    },
+    {
+      "epoch": 10.018552875695732,
+      "grad_norm": 0.451171875,
+      "learning_rate": 5.986377600199371e-05,
+      "loss": 0.8597,
+      "step": 8100
+    },
+    {
+      "epoch": 10.02473716759431,
+      "grad_norm": 0.423828125,
+      "learning_rate": 5.973192154714547e-05,
+      "loss": 0.8643,
+      "step": 8105
+    },
+    {
+      "epoch": 10.030921459492887,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.9600150596883066e-05,
+      "loss": 0.8591,
+      "step": 8110
+    },
+    {
+      "epoch": 10.037105751391465,
+      "grad_norm": 0.462890625,
+      "learning_rate": 5.946846342446214e-05,
+      "loss": 0.8699,
+      "step": 8115
+    },
+    {
+      "epoch": 10.043290043290042,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.933686030296459e-05,
+      "loss": 0.8558,
+      "step": 8120
+    },
+    {
+      "epoch": 10.049474335188622,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.920534150529797e-05,
+      "loss": 0.8567,
+      "step": 8125
+    },
+    {
+      "epoch": 10.0556586270872,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.907390730419507e-05,
+      "loss": 0.8566,
+      "step": 8130
+    },
+    {
+      "epoch": 10.061842918985777,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.894255797221313e-05,
+      "loss": 0.8462,
+      "step": 8135
+    },
+    {
+      "epoch": 10.068027210884354,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.881129378173347e-05,
+      "loss": 0.8521,
+      "step": 8140
+    },
+    {
+      "epoch": 10.074211502782932,
+      "grad_norm": 0.490234375,
+      "learning_rate": 5.868011500496084e-05,
+      "loss": 0.858,
+      "step": 8145
+    },
+    {
+      "epoch": 10.08039579468151,
+      "grad_norm": 0.453125,
+      "learning_rate": 5.854902191392284e-05,
+      "loss": 0.8568,
+      "step": 8150
+    },
+    {
+      "epoch": 10.086580086580087,
+      "grad_norm": 0.474609375,
+      "learning_rate": 5.84180147804694e-05,
+      "loss": 0.864,
+      "step": 8155
+    },
+    {
+      "epoch": 10.092764378478664,
+      "grad_norm": 0.474609375,
+      "learning_rate": 5.828709387627218e-05,
+      "loss": 0.8633,
+      "step": 8160
+    },
+    {
+      "epoch": 10.098948670377242,
+      "grad_norm": 0.44921875,
+      "learning_rate": 5.8156259472824124e-05,
+      "loss": 0.8608,
+      "step": 8165
+    },
+    {
+      "epoch": 10.10513296227582,
+      "grad_norm": 0.447265625,
+      "learning_rate": 5.802551184143865e-05,
+      "loss": 0.8512,
+      "step": 8170
+    },
+    {
+      "epoch": 10.111317254174397,
+      "grad_norm": 0.44921875,
+      "learning_rate": 5.789485125324926e-05,
+      "loss": 0.8572,
+      "step": 8175
+    },
+    {
+      "epoch": 10.117501546072974,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.7764277979209094e-05,
+      "loss": 0.8498,
+      "step": 8180
+    },
+    {
+      "epoch": 10.123685837971552,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.763379229009003e-05,
+      "loss": 0.8596,
+      "step": 8185
+    },
+    {
+      "epoch": 10.12987012987013,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.750339445648252e-05,
+      "loss": 0.8667,
+      "step": 8190
+    },
+    {
+      "epoch": 10.136054421768707,
+      "grad_norm": 0.453125,
+      "learning_rate": 5.7373084748794626e-05,
+      "loss": 0.8661,
+      "step": 8195
+    },
+    {
+      "epoch": 10.142238713667284,
+      "grad_norm": 0.451171875,
+      "learning_rate": 5.724286343725185e-05,
+      "loss": 0.8522,
+      "step": 8200
+    },
+    {
+      "epoch": 10.148423005565864,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.7112730791896207e-05,
+      "loss": 0.863,
+      "step": 8205
+    },
+    {
+      "epoch": 10.154607297464441,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.6982687082585994e-05,
+      "loss": 0.8652,
+      "step": 8210
+    },
+    {
+      "epoch": 10.160791589363019,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.685273257899505e-05,
+      "loss": 0.8557,
+      "step": 8215
+    },
+    {
+      "epoch": 10.166975881261596,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.6722867550612116e-05,
+      "loss": 0.8567,
+      "step": 8220
+    },
+    {
+      "epoch": 10.173160173160174,
+      "grad_norm": 0.45703125,
+      "learning_rate": 5.6593092266740545e-05,
+      "loss": 0.8569,
+      "step": 8225
+    },
+    {
+      "epoch": 10.179344465058751,
+      "grad_norm": 0.44921875,
+      "learning_rate": 5.6463406996497456e-05,
+      "loss": 0.8591,
+      "step": 8230
+    },
+    {
+      "epoch": 10.185528756957329,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.633381200881335e-05,
+      "loss": 0.8581,
+      "step": 8235
+    },
+    {
+      "epoch": 10.191713048855906,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.620430757243156e-05,
+      "loss": 0.8579,
+      "step": 8240
+    },
+    {
+      "epoch": 10.197897340754484,
+      "grad_norm": 0.458984375,
+      "learning_rate": 5.6074893955907535e-05,
+      "loss": 0.8621,
+      "step": 8245
+    },
+    {
+      "epoch": 10.204081632653061,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.5945571427608526e-05,
+      "loss": 0.8627,
+      "step": 8250
+    },
+    {
+      "epoch": 10.210265924551639,
+      "grad_norm": 0.421875,
+      "learning_rate": 5.581634025571274e-05,
+      "loss": 0.8577,
+      "step": 8255
+    },
+    {
+      "epoch": 10.216450216450216,
+      "grad_norm": 0.45703125,
+      "learning_rate": 5.5687200708209076e-05,
+      "loss": 0.8604,
+      "step": 8260
+    },
+    {
+      "epoch": 10.222634508348794,
+      "grad_norm": 0.451171875,
+      "learning_rate": 5.555815305289631e-05,
+      "loss": 0.8684,
+      "step": 8265
+    },
+    {
+      "epoch": 10.228818800247371,
+      "grad_norm": 0.44140625,
+      "learning_rate": 5.542919755738275e-05,
+      "loss": 0.8599,
+      "step": 8270
+    },
+    {
+      "epoch": 10.235003092145949,
+      "grad_norm": 0.515625,
+      "learning_rate": 5.5300334489085595e-05,
+      "loss": 0.8535,
+      "step": 8275
+    },
+    {
+      "epoch": 10.241187384044528,
+      "grad_norm": 0.421875,
+      "learning_rate": 5.5171564115230254e-05,
+      "loss": 0.8628,
+      "step": 8280
+    },
+    {
+      "epoch": 10.247371675943105,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.504288670285008e-05,
+      "loss": 0.8571,
+      "step": 8285
+    },
+    {
+      "epoch": 10.253555967841683,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.491430251878551e-05,
+      "loss": 0.8539,
+      "step": 8290
+    },
+    {
+      "epoch": 10.25974025974026,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.4785811829683764e-05,
+      "loss": 0.8602,
+      "step": 8295
+    },
+    {
+      "epoch": 10.265924551638838,
+      "grad_norm": 0.44921875,
+      "learning_rate": 5.4657414901998095e-05,
+      "loss": 0.8619,
+      "step": 8300
+    },
+    {
+      "epoch": 10.272108843537415,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.4529112001987314e-05,
+      "loss": 0.8605,
+      "step": 8305
+    },
+    {
+      "epoch": 10.278293135435993,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.4400903395715366e-05,
+      "loss": 0.8549,
+      "step": 8310
+    },
+    {
+      "epoch": 10.28447742733457,
+      "grad_norm": 0.486328125,
+      "learning_rate": 5.427278934905049e-05,
+      "loss": 0.8649,
+      "step": 8315
+    },
+    {
+      "epoch": 10.290661719233148,
+      "grad_norm": 0.451171875,
+      "learning_rate": 5.4144770127665024e-05,
+      "loss": 0.8582,
+      "step": 8320
+    },
+    {
+      "epoch": 10.296846011131725,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.401684599703445e-05,
+      "loss": 0.8613,
+      "step": 8325
+    },
+    {
+      "epoch": 10.303030303030303,
+      "grad_norm": 0.44140625,
+      "learning_rate": 5.388901722243724e-05,
+      "loss": 0.8498,
+      "step": 8330
+    },
+    {
+      "epoch": 10.30921459492888,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.376128406895408e-05,
+      "loss": 0.8547,
+      "step": 8335
+    },
+    {
+      "epoch": 10.315398886827458,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.363364680146725e-05,
+      "loss": 0.8501,
+      "step": 8340
+    },
+    {
+      "epoch": 10.321583178726035,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.350610568466039e-05,
+      "loss": 0.8616,
+      "step": 8345
+    },
+    {
+      "epoch": 10.327767470624613,
+      "grad_norm": 0.44140625,
+      "learning_rate": 5.3378660983017536e-05,
+      "loss": 0.8573,
+      "step": 8350
+    },
+    {
+      "epoch": 10.33395176252319,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.325131296082298e-05,
+      "loss": 0.8596,
+      "step": 8355
+    },
+    {
+      "epoch": 10.34013605442177,
+      "grad_norm": 0.423828125,
+      "learning_rate": 5.31240618821604e-05,
+      "loss": 0.8661,
+      "step": 8360
+    },
+    {
+      "epoch": 10.346320346320347,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.2996908010912437e-05,
+      "loss": 0.8436,
+      "step": 8365
+    },
+    {
+      "epoch": 10.352504638218925,
+      "grad_norm": 0.4296875,
+      "learning_rate": 5.286985161076029e-05,
+      "loss": 0.8591,
+      "step": 8370
+    },
+    {
+      "epoch": 10.358688930117502,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.274289294518283e-05,
+      "loss": 0.8619,
+      "step": 8375
+    },
+    {
+      "epoch": 10.36487322201608,
+      "grad_norm": 0.45703125,
+      "learning_rate": 5.2616032277456463e-05,
+      "loss": 0.8588,
+      "step": 8380
+    },
+    {
+      "epoch": 10.371057513914657,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.248926987065417e-05,
+      "loss": 0.8616,
+      "step": 8385
+    },
+    {
+      "epoch": 10.377241805813235,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.236260598764535e-05,
+      "loss": 0.8631,
+      "step": 8390
+    },
+    {
+      "epoch": 10.383426097711812,
+      "grad_norm": 0.44140625,
+      "learning_rate": 5.223604089109495e-05,
+      "loss": 0.8628,
+      "step": 8395
+    },
+    {
+      "epoch": 10.38961038961039,
+      "grad_norm": 0.453125,
+      "learning_rate": 5.210957484346314e-05,
+      "loss": 0.8553,
+      "step": 8400
+    },
+    {
+      "epoch": 10.395794681508967,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.198320810700472e-05,
+      "loss": 0.8551,
+      "step": 8405
+    },
+    {
+      "epoch": 10.401978973407545,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.185694094376843e-05,
+      "loss": 0.8565,
+      "step": 8410
+    },
+    {
+      "epoch": 10.408163265306122,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.173077361559665e-05,
+      "loss": 0.8714,
+      "step": 8415
+    },
+    {
+      "epoch": 10.4143475572047,
+      "grad_norm": 0.4609375,
+      "learning_rate": 5.160470638412461e-05,
+      "loss": 0.8585,
+      "step": 8420
+    },
+    {
+      "epoch": 10.420531849103277,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.1478739510780104e-05,
+      "loss": 0.8488,
+      "step": 8425
+    },
+    {
+      "epoch": 10.426716141001855,
+      "grad_norm": 0.462890625,
+      "learning_rate": 5.135287325678271e-05,
+      "loss": 0.8691,
+      "step": 8430
+    },
+    {
+      "epoch": 10.432900432900432,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.122710788314331e-05,
+      "loss": 0.8578,
+      "step": 8435
+    },
+    {
+      "epoch": 10.439084724799011,
+      "grad_norm": 0.44921875,
+      "learning_rate": 5.1101443650663764e-05,
+      "loss": 0.8591,
+      "step": 8440
+    },
+    {
+      "epoch": 10.445269016697589,
+      "grad_norm": 0.4296875,
+      "learning_rate": 5.0975880819936004e-05,
+      "loss": 0.8588,
+      "step": 8445
+    },
+    {
+      "epoch": 10.451453308596166,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.085041965134183e-05,
+      "loss": 0.8542,
+      "step": 8450
+    },
+    {
+      "epoch": 10.457637600494744,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.072506040505208e-05,
+      "loss": 0.8595,
+      "step": 8455
+    },
+    {
+      "epoch": 10.463821892393321,
+      "grad_norm": 0.44921875,
+      "learning_rate": 5.059980334102637e-05,
+      "loss": 0.8456,
+      "step": 8460
+    },
+    {
+      "epoch": 10.470006184291899,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.04746487190124e-05,
+      "loss": 0.855,
+      "step": 8465
+    },
+    {
+      "epoch": 10.476190476190476,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.034959679854532e-05,
+      "loss": 0.8573,
+      "step": 8470
+    },
+    {
+      "epoch": 10.482374768089054,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.022464783894744e-05,
+      "loss": 0.8549,
+      "step": 8475
+    },
+    {
+      "epoch": 10.488559059987631,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.009980209932743e-05,
+      "loss": 0.8708,
+      "step": 8480
+    },
+    {
+      "epoch": 10.494743351886209,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.9975059838580083e-05,
+      "loss": 0.8615,
+      "step": 8485
+    },
+    {
+      "epoch": 10.500927643784786,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.985042131538545e-05,
+      "loss": 0.8629,
+      "step": 8490
+    },
+    {
+      "epoch": 10.507111935683364,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.9725886788208474e-05,
+      "loss": 0.8657,
+      "step": 8495
+    },
+    {
+      "epoch": 10.513296227581941,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.960145651529856e-05,
+      "loss": 0.8473,
+      "step": 8500
+    },
+    {
+      "epoch": 10.519480519480519,
+      "grad_norm": 0.419921875,
+      "learning_rate": 4.9477130754688775e-05,
+      "loss": 0.8669,
+      "step": 8505
+    },
+    {
+      "epoch": 10.525664811379096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.9352909764195576e-05,
+      "loss": 0.8594,
+      "step": 8510
+    },
+    {
+      "epoch": 10.531849103277676,
+      "grad_norm": 0.46484375,
+      "learning_rate": 4.922879380141805e-05,
+      "loss": 0.8501,
+      "step": 8515
+    },
+    {
+      "epoch": 10.538033395176253,
+      "grad_norm": 0.466796875,
+      "learning_rate": 4.9104783123737566e-05,
+      "loss": 0.8597,
+      "step": 8520
+    },
+    {
+      "epoch": 10.54421768707483,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.898087798831716e-05,
+      "loss": 0.8558,
+      "step": 8525
+    },
+    {
+      "epoch": 10.550401978973408,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.885707865210093e-05,
+      "loss": 0.8599,
+      "step": 8530
+    },
+    {
+      "epoch": 10.556586270871986,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.873338537181368e-05,
+      "loss": 0.8601,
+      "step": 8535
+    },
+    {
+      "epoch": 10.562770562770563,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.860979840396016e-05,
+      "loss": 0.8651,
+      "step": 8540
+    },
+    {
+      "epoch": 10.56895485466914,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.8486318004824794e-05,
+      "loss": 0.8599,
+      "step": 8545
+    },
+    {
+      "epoch": 10.575139146567718,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.836294443047088e-05,
+      "loss": 0.8638,
+      "step": 8550
+    },
+    {
+      "epoch": 10.581323438466296,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.823967793674033e-05,
+      "loss": 0.8507,
+      "step": 8555
+    },
+    {
+      "epoch": 10.587507730364873,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.8116518779252885e-05,
+      "loss": 0.8627,
+      "step": 8560
+    },
+    {
+      "epoch": 10.59369202226345,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.7993467213405706e-05,
+      "loss": 0.8496,
+      "step": 8565
+    },
+    {
+      "epoch": 10.599876314162028,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.787052349437295e-05,
+      "loss": 0.8644,
+      "step": 8570
+    },
+    {
+      "epoch": 10.606060606060606,
+      "grad_norm": 0.45703125,
+      "learning_rate": 4.774768787710501e-05,
+      "loss": 0.8613,
+      "step": 8575
+    },
+    {
+      "epoch": 10.612244897959183,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.762496061632814e-05,
+      "loss": 0.8637,
+      "step": 8580
+    },
+    {
+      "epoch": 10.61842918985776,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.7502341966544e-05,
+      "loss": 0.8614,
+      "step": 8585
+    },
+    {
+      "epoch": 10.624613481756338,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.7379832182028814e-05,
+      "loss": 0.8562,
+      "step": 8590
+    },
+    {
+      "epoch": 10.630797773654917,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.725743151683325e-05,
+      "loss": 0.8573,
+      "step": 8595
+    },
+    {
+      "epoch": 10.636982065553495,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.713514022478155e-05,
+      "loss": 0.8611,
+      "step": 8600
+    },
+    {
+      "epoch": 10.643166357452072,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.701295855947126e-05,
+      "loss": 0.8604,
+      "step": 8605
+    },
+    {
+      "epoch": 10.64935064935065,
+      "grad_norm": 0.423828125,
+      "learning_rate": 4.689088677427249e-05,
+      "loss": 0.8561,
+      "step": 8610
+    },
+    {
+      "epoch": 10.655534941249227,
+      "grad_norm": 0.46484375,
+      "learning_rate": 4.676892512232758e-05,
+      "loss": 0.8619,
+      "step": 8615
+    },
+    {
+      "epoch": 10.661719233147805,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.6647073856550415e-05,
+      "loss": 0.8636,
+      "step": 8620
+    },
+    {
+      "epoch": 10.667903525046382,
+      "grad_norm": 0.48046875,
+      "learning_rate": 4.652533322962597e-05,
+      "loss": 0.8545,
+      "step": 8625
+    },
+    {
+      "epoch": 10.67408781694496,
+      "grad_norm": 0.484375,
+      "learning_rate": 4.6403703494009875e-05,
+      "loss": 0.8494,
+      "step": 8630
+    },
+    {
+      "epoch": 10.680272108843537,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.6282184901927674e-05,
+      "loss": 0.8562,
+      "step": 8635
+    },
+    {
+      "epoch": 10.686456400742115,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.6160777705374524e-05,
+      "loss": 0.8659,
+      "step": 8640
+    },
+    {
+      "epoch": 10.692640692640692,
+      "grad_norm": 0.46484375,
+      "learning_rate": 4.603948215611461e-05,
+      "loss": 0.8543,
+      "step": 8645
+    },
+    {
+      "epoch": 10.69882498453927,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.591829850568046e-05,
+      "loss": 0.8608,
+      "step": 8650
+    },
+    {
+      "epoch": 10.705009276437847,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.579722700537268e-05,
+      "loss": 0.8566,
+      "step": 8655
+    },
+    {
+      "epoch": 10.711193568336425,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.567626790625921e-05,
+      "loss": 0.8532,
+      "step": 8660
+    },
+    {
+      "epoch": 10.717377860235002,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.555542145917501e-05,
+      "loss": 0.8549,
+      "step": 8665
+    },
+    {
+      "epoch": 10.723562152133582,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.543468791472131e-05,
+      "loss": 0.8586,
+      "step": 8670
+    },
+    {
+      "epoch": 10.729746444032159,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.5314067523265333e-05,
+      "loss": 0.8684,
+      "step": 8675
+    },
+    {
+      "epoch": 10.735930735930737,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.519356053493958e-05,
+      "loss": 0.8558,
+      "step": 8680
+    },
+    {
+      "epoch": 10.742115027829314,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.5073167199641367e-05,
+      "loss": 0.8604,
+      "step": 8685
+    },
+    {
+      "epoch": 10.748299319727892,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.495288776703241e-05,
+      "loss": 0.855,
+      "step": 8690
+    },
+    {
+      "epoch": 10.754483611626469,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.483272248653811e-05,
+      "loss": 0.8472,
+      "step": 8695
+    },
+    {
+      "epoch": 10.760667903525047,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.471267160734731e-05,
+      "loss": 0.8542,
+      "step": 8700
+    },
+    {
+      "epoch": 10.766852195423624,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.459273537841141e-05,
+      "loss": 0.8558,
+      "step": 8705
+    },
+    {
+      "epoch": 10.773036487322202,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.447291404844424e-05,
+      "loss": 0.849,
+      "step": 8710
+    },
+    {
+      "epoch": 10.779220779220779,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.43532078659213e-05,
+      "loss": 0.8582,
+      "step": 8715
+    },
+    {
+      "epoch": 10.785405071119357,
+      "grad_norm": 0.451171875,
+      "learning_rate": 4.4233617079079236e-05,
+      "loss": 0.8583,
+      "step": 8720
+    },
+    {
+      "epoch": 10.791589363017934,
+      "grad_norm": 0.466796875,
+      "learning_rate": 4.4114141935915534e-05,
+      "loss": 0.8683,
+      "step": 8725
+    },
+    {
+      "epoch": 10.797773654916512,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.399478268418771e-05,
+      "loss": 0.87,
+      "step": 8730
+    },
+    {
+      "epoch": 10.803957946815089,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.3875539571413106e-05,
+      "loss": 0.8518,
+      "step": 8735
+    },
+    {
+      "epoch": 10.810142238713667,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.375641284486808e-05,
+      "loss": 0.8577,
+      "step": 8740
+    },
+    {
+      "epoch": 10.816326530612244,
+      "grad_norm": 0.4921875,
+      "learning_rate": 4.36374027515878e-05,
+      "loss": 0.8692,
+      "step": 8745
+    },
+    {
+      "epoch": 10.822510822510823,
+      "grad_norm": 0.451171875,
+      "learning_rate": 4.3518509538365425e-05,
+      "loss": 0.866,
+      "step": 8750
+    },
+    {
+      "epoch": 10.8286951144094,
+      "grad_norm": 0.462890625,
+      "learning_rate": 4.3399733451751776e-05,
+      "loss": 0.859,
+      "step": 8755
+    },
+    {
+      "epoch": 10.834879406307978,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.328107473805487e-05,
+      "loss": 0.8584,
+      "step": 8760
+    },
+    {
+      "epoch": 10.841063698206556,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.3162533643339185e-05,
+      "loss": 0.8585,
+      "step": 8765
+    },
+    {
+      "epoch": 10.847247990105133,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.3044110413425395e-05,
+      "loss": 0.867,
+      "step": 8770
+    },
+    {
+      "epoch": 10.85343228200371,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.2925805293889786e-05,
+      "loss": 0.854,
+      "step": 8775
+    },
+    {
+      "epoch": 10.859616573902288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.2807618530063565e-05,
+      "loss": 0.8525,
+      "step": 8780
+    },
+    {
+      "epoch": 10.865800865800866,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.268955036703267e-05,
+      "loss": 0.8495,
+      "step": 8785
+    },
+    {
+      "epoch": 10.871985157699443,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.257160104963696e-05,
+      "loss": 0.863,
+      "step": 8790
+    },
+    {
+      "epoch": 10.87816944959802,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.245377082246995e-05,
+      "loss": 0.8614,
+      "step": 8795
+    },
+    {
+      "epoch": 10.884353741496598,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.23360599298781e-05,
+      "loss": 0.8674,
+      "step": 8800
+    },
+    {
+      "epoch": 10.890538033395176,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.2218468615960484e-05,
+      "loss": 0.8588,
+      "step": 8805
+    },
+    {
+      "epoch": 10.896722325293753,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.210099712456822e-05,
+      "loss": 0.8665,
+      "step": 8810
+    },
+    {
+      "epoch": 10.90290661719233,
+      "grad_norm": 0.451171875,
+      "learning_rate": 4.1983645699303786e-05,
+      "loss": 0.8527,
+      "step": 8815
+    },
+    {
+      "epoch": 10.909090909090908,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.1866414583520877e-05,
+      "loss": 0.8574,
+      "step": 8820
+    },
+    {
+      "epoch": 10.915275200989488,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.174930402032354e-05,
+      "loss": 0.8598,
+      "step": 8825
+    },
+    {
+      "epoch": 10.921459492888065,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.163231425256595e-05,
+      "loss": 0.8656,
+      "step": 8830
+    },
+    {
+      "epoch": 10.927643784786643,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.1515445522851784e-05,
+      "loss": 0.8586,
+      "step": 8835
+    },
+    {
+      "epoch": 10.93382807668522,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.139869807353357e-05,
+      "loss": 0.8615,
+      "step": 8840
+    },
+    {
+      "epoch": 10.940012368583798,
+      "grad_norm": 0.447265625,
+      "learning_rate": 4.128207214671255e-05,
+      "loss": 0.8655,
+      "step": 8845
+    },
+    {
+      "epoch": 10.946196660482375,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.1165567984237764e-05,
+      "loss": 0.8549,
+      "step": 8850
+    },
+    {
+      "epoch": 10.952380952380953,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.1049185827705904e-05,
+      "loss": 0.859,
+      "step": 8855
+    },
+    {
+      "epoch": 10.95856524427953,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.0932925918460516e-05,
+      "loss": 0.8618,
+      "step": 8860
+    },
+    {
+      "epoch": 10.964749536178108,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.081678849759181e-05,
+      "loss": 0.8544,
+      "step": 8865
+    },
+    {
+      "epoch": 10.970933828076685,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.070077380593579e-05,
+      "loss": 0.8579,
+      "step": 8870
+    },
+    {
+      "epoch": 10.977118119975263,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.058488208407415e-05,
+      "loss": 0.8697,
+      "step": 8875
+    },
+    {
+      "epoch": 10.98330241187384,
+      "grad_norm": 0.423828125,
+      "learning_rate": 4.046911357233343e-05,
+      "loss": 0.8579,
+      "step": 8880
+    },
+    {
+      "epoch": 10.989486703772418,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.035346851078471e-05,
+      "loss": 0.8545,
+      "step": 8885
+    },
+    {
+      "epoch": 10.995670995670995,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.02379471392431e-05,
+      "loss": 0.8555,
+      "step": 8890
+    },
+    {
+      "epoch": 10.999381570810142,
+      "eval_loss": 2.519637107849121,
+      "eval_runtime": 0.6376,
+      "eval_samples_per_second": 15.684,
+      "eval_steps_per_second": 1.568,
+      "step": 8893
+    },
+    {
+      "epoch": 11.001855287569573,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.0122549697267244e-05,
+      "loss": 0.8561,
+      "step": 8895
+    },
+    {
+      "epoch": 11.00803957946815,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.000727642415867e-05,
+      "loss": 0.8568,
+      "step": 8900
+    },
+    {
+      "epoch": 11.01422387136673,
+      "grad_norm": 0.470703125,
+      "learning_rate": 3.9892127558961546e-05,
+      "loss": 0.8552,
+      "step": 8905
+    },
+    {
+      "epoch": 11.020408163265307,
+      "grad_norm": 0.44921875,
+      "learning_rate": 3.977710334046193e-05,
+      "loss": 0.8449,
+      "step": 8910
+    },
+    {
+      "epoch": 11.026592455163884,
+      "grad_norm": 0.453125,
+      "learning_rate": 3.9662204007187534e-05,
+      "loss": 0.8568,
+      "step": 8915
+    },
+    {
+      "epoch": 11.032776747062462,
+      "grad_norm": 0.44921875,
+      "learning_rate": 3.954742979740695e-05,
+      "loss": 0.8545,
+      "step": 8920
+    },
+    {
+      "epoch": 11.03896103896104,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.943278094912946e-05,
+      "loss": 0.8591,
+      "step": 8925
+    },
+    {
+      "epoch": 11.045145330859617,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.9318257700104174e-05,
+      "loss": 0.8521,
+      "step": 8930
+    },
+    {
+      "epoch": 11.051329622758194,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.920386028781995e-05,
+      "loss": 0.8529,
+      "step": 8935
+    },
+    {
+      "epoch": 11.057513914656772,
+      "grad_norm": 0.466796875,
+      "learning_rate": 3.9089588949504655e-05,
+      "loss": 0.8505,
+      "step": 8940
+    },
+    {
+      "epoch": 11.06369820655535,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.897544392212453e-05,
+      "loss": 0.8584,
+      "step": 8945
+    },
+    {
+      "epoch": 11.069882498453927,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.8861425442384135e-05,
+      "loss": 0.8614,
+      "step": 8950
+    },
+    {
+      "epoch": 11.076066790352504,
+      "grad_norm": 0.4375,
+      "learning_rate": 3.874753374672542e-05,
+      "loss": 0.8564,
+      "step": 8955
+    },
+    {
+      "epoch": 11.082251082251082,
+      "grad_norm": 0.4375,
+      "learning_rate": 3.863376907132752e-05,
+      "loss": 0.8547,
+      "step": 8960
+    },
+    {
+      "epoch": 11.08843537414966,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.8520131652106186e-05,
+      "loss": 0.8568,
+      "step": 8965
+    },
+    {
+      "epoch": 11.094619666048237,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.840662172471315e-05,
+      "loss": 0.8578,
+      "step": 8970
+    },
+    {
+      "epoch": 11.100803957946814,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.8293239524535915e-05,
+      "loss": 0.8555,
+      "step": 8975
+    },
+    {
+      "epoch": 11.106988249845394,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.8179985286696986e-05,
+      "loss": 0.8503,
+      "step": 8980
+    },
+    {
+      "epoch": 11.113172541743971,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.806685924605361e-05,
+      "loss": 0.8591,
+      "step": 8985
+    },
+    {
+      "epoch": 11.119356833642549,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.7953861637197085e-05,
+      "loss": 0.8569,
+      "step": 8990
+    },
+    {
+      "epoch": 11.125541125541126,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.784099269445247e-05,
+      "loss": 0.8432,
+      "step": 8995
+    },
+    {
+      "epoch": 11.131725417439704,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.772825265187802e-05,
+      "loss": 0.8495,
+      "step": 9000
+    },
+    {
+      "epoch": 11.137909709338281,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.7615641743264586e-05,
+      "loss": 0.8622,
+      "step": 9005
+    },
+    {
+      "epoch": 11.144094001236859,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.75031602021353e-05,
+      "loss": 0.8522,
+      "step": 9010
+    },
+    {
+      "epoch": 11.150278293135436,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.739080826174498e-05,
+      "loss": 0.863,
+      "step": 9015
+    },
+    {
+      "epoch": 11.156462585034014,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.727858615507974e-05,
+      "loss": 0.8574,
+      "step": 9020
+    },
+    {
+      "epoch": 11.162646876932591,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.716649411485649e-05,
+      "loss": 0.8516,
+      "step": 9025
+    },
+    {
+      "epoch": 11.168831168831169,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.705453237352227e-05,
+      "loss": 0.8488,
+      "step": 9030
+    },
+    {
+      "epoch": 11.175015460729746,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.694270116325409e-05,
+      "loss": 0.8661,
+      "step": 9035
+    },
+    {
+      "epoch": 11.181199752628324,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.683100071595813e-05,
+      "loss": 0.8578,
+      "step": 9040
+    },
+    {
+      "epoch": 11.187384044526901,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.6719431263269533e-05,
+      "loss": 0.8548,
+      "step": 9045
+    },
+    {
+      "epoch": 11.193568336425479,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.660799303655166e-05,
+      "loss": 0.8551,
+      "step": 9050
+    },
+    {
+      "epoch": 11.199752628324056,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.6496686266895874e-05,
+      "loss": 0.8744,
+      "step": 9055
+    },
+    {
+      "epoch": 11.205936920222635,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.638551118512089e-05,
+      "loss": 0.8612,
+      "step": 9060
+    },
+    {
+      "epoch": 11.212121212121213,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.6274468021772323e-05,
+      "loss": 0.8442,
+      "step": 9065
+    },
+    {
+      "epoch": 11.21830550401979,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.616355700712221e-05,
+      "loss": 0.8621,
+      "step": 9070
+    },
+    {
+      "epoch": 11.224489795918368,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.605277837116854e-05,
+      "loss": 0.8561,
+      "step": 9075
+    },
+    {
+      "epoch": 11.230674087816945,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.594213234363486e-05,
+      "loss": 0.8508,
+      "step": 9080
+    },
+    {
+      "epoch": 11.236858379715523,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.583161915396971e-05,
+      "loss": 0.8477,
+      "step": 9085
+    },
+    {
+      "epoch": 11.2430426716141,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.5721239031346066e-05,
+      "loss": 0.8554,
+      "step": 9090
+    },
+    {
+      "epoch": 11.249226963512678,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.561099220466111e-05,
+      "loss": 0.8456,
+      "step": 9095
+    },
+    {
+      "epoch": 11.255411255411255,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.550087890253544e-05,
+      "loss": 0.8625,
+      "step": 9100
+    },
+    {
+      "epoch": 11.261595547309833,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.539089935331294e-05,
+      "loss": 0.8504,
+      "step": 9105
+    },
+    {
+      "epoch": 11.26777983920841,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.5281053785059925e-05,
+      "loss": 0.8567,
+      "step": 9110
+    },
+    {
+      "epoch": 11.273964131106988,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.5171342425565055e-05,
+      "loss": 0.8728,
+      "step": 9115
+    },
+    {
+      "epoch": 11.280148423005565,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.506176550233863e-05,
+      "loss": 0.855,
+      "step": 9120
+    },
+    {
+      "epoch": 11.286332714904143,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.495232324261206e-05,
+      "loss": 0.8537,
+      "step": 9125
+    },
+    {
+      "epoch": 11.29251700680272,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.484301587333772e-05,
+      "loss": 0.8524,
+      "step": 9130
+    },
+    {
+      "epoch": 11.2987012987013,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.473384362118794e-05,
+      "loss": 0.8689,
+      "step": 9135
+    },
+    {
+      "epoch": 11.304885590599877,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.462480671255515e-05,
+      "loss": 0.8683,
+      "step": 9140
+    },
+    {
+      "epoch": 11.311069882498455,
+      "grad_norm": 0.447265625,
+      "learning_rate": 3.4515905373551016e-05,
+      "loss": 0.863,
+      "step": 9145
+    },
+    {
+      "epoch": 11.317254174397032,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.440713983000601e-05,
+      "loss": 0.8485,
+      "step": 9150
+    },
+    {
+      "epoch": 11.32343846629561,
+      "grad_norm": 0.41796875,
+      "learning_rate": 3.42985103074691e-05,
+      "loss": 0.8589,
+      "step": 9155
+    },
+    {
+      "epoch": 11.329622758194187,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.419001703120709e-05,
+      "loss": 0.8566,
+      "step": 9160
+    },
+    {
+      "epoch": 11.335807050092765,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.4081660226204357e-05,
+      "loss": 0.8617,
+      "step": 9165
+    },
+    {
+      "epoch": 11.341991341991342,
+      "grad_norm": 0.45703125,
+      "learning_rate": 3.397344011716216e-05,
+      "loss": 0.8579,
+      "step": 9170
+    },
+    {
+      "epoch": 11.34817563388992,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.386535692849838e-05,
+      "loss": 0.8511,
+      "step": 9175
+    },
+    {
+      "epoch": 11.354359925788497,
+      "grad_norm": 0.478515625,
+      "learning_rate": 3.3757410884346894e-05,
+      "loss": 0.8612,
+      "step": 9180
+    },
+    {
+      "epoch": 11.360544217687075,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.364960220855723e-05,
+      "loss": 0.8714,
+      "step": 9185
+    },
+    {
+      "epoch": 11.366728509585652,
+      "grad_norm": 0.447265625,
+      "learning_rate": 3.354193112469407e-05,
+      "loss": 0.8527,
+      "step": 9190
+    },
+    {
+      "epoch": 11.37291280148423,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.34343978560367e-05,
+      "loss": 0.859,
+      "step": 9195
+    },
+    {
+      "epoch": 11.379097093382807,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.332700262557864e-05,
+      "loss": 0.857,
+      "step": 9200
+    },
+    {
+      "epoch": 11.385281385281385,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.321974565602722e-05,
+      "loss": 0.861,
+      "step": 9205
+    },
+    {
+      "epoch": 11.391465677179962,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.3112627169802946e-05,
+      "loss": 0.8597,
+      "step": 9210
+    },
+    {
+      "epoch": 11.397649969078541,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.300564738903926e-05,
+      "loss": 0.8571,
+      "step": 9215
+    },
+    {
+      "epoch": 11.403834260977119,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.289880653558188e-05,
+      "loss": 0.8648,
+      "step": 9220
+    },
+    {
+      "epoch": 11.410018552875696,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.2792104830988515e-05,
+      "loss": 0.8598,
+      "step": 9225
+    },
+    {
+      "epoch": 11.416202844774274,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.2685542496528185e-05,
+      "loss": 0.8615,
+      "step": 9230
+    },
+    {
+      "epoch": 11.422387136672851,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.257911975318109e-05,
+      "loss": 0.8524,
+      "step": 9235
+    },
+    {
+      "epoch": 11.428571428571429,
+      "grad_norm": 0.466796875,
+      "learning_rate": 3.2472836821637744e-05,
+      "loss": 0.8651,
+      "step": 9240
+    },
+    {
+      "epoch": 11.434755720470006,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.236669392229888e-05,
+      "loss": 0.8592,
+      "step": 9245
+    },
+    {
+      "epoch": 11.440940012368584,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.2260691275274835e-05,
+      "loss": 0.8623,
+      "step": 9250
+    },
+    {
+      "epoch": 11.447124304267161,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.2154829100385e-05,
+      "loss": 0.8588,
+      "step": 9255
+    },
+    {
+      "epoch": 11.453308596165739,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.204910761715763e-05,
+      "loss": 0.8525,
+      "step": 9260
+    },
+    {
+      "epoch": 11.459492888064316,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.194352704482899e-05,
+      "loss": 0.8494,
+      "step": 9265
+    },
+    {
+      "epoch": 11.465677179962894,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.1838087602343344e-05,
+      "loss": 0.8531,
+      "step": 9270
+    },
+    {
+      "epoch": 11.471861471861471,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.173278950835227e-05,
+      "loss": 0.8532,
+      "step": 9275
+    },
+    {
+      "epoch": 11.478045763760049,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.162763298121408e-05,
+      "loss": 0.8603,
+      "step": 9280
+    },
+    {
+      "epoch": 11.484230055658626,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.1522618238993725e-05,
+      "loss": 0.8561,
+      "step": 9285
+    },
+    {
+      "epoch": 11.490414347557206,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.1417745499461934e-05,
+      "loss": 0.8618,
+      "step": 9290
+    },
+    {
+      "epoch": 11.496598639455783,
+      "grad_norm": 0.44921875,
+      "learning_rate": 3.131301498009514e-05,
+      "loss": 0.8649,
+      "step": 9295
+    },
+    {
+      "epoch": 11.50278293135436,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.120842689807468e-05,
+      "loss": 0.8618,
+      "step": 9300
+    },
+    {
+      "epoch": 11.508967223252938,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.110398147028666e-05,
+      "loss": 0.8535,
+      "step": 9305
+    },
+    {
+      "epoch": 11.515151515151516,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.099967891332132e-05,
+      "loss": 0.8454,
+      "step": 9310
+    },
+    {
+      "epoch": 11.521335807050093,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.089551944347255e-05,
+      "loss": 0.8551,
+      "step": 9315
+    },
+    {
+      "epoch": 11.52752009894867,
+      "grad_norm": 0.447265625,
+      "learning_rate": 3.079150327673766e-05,
+      "loss": 0.8603,
+      "step": 9320
+    },
+    {
+      "epoch": 11.533704390847248,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.0687630628816656e-05,
+      "loss": 0.8571,
+      "step": 9325
+    },
+    {
+      "epoch": 11.539888682745826,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.058390171511196e-05,
+      "loss": 0.8548,
+      "step": 9330
+    },
+    {
+      "epoch": 11.546072974644403,
+      "grad_norm": 0.4140625,
+      "learning_rate": 3.0480316750728012e-05,
+      "loss": 0.8556,
+      "step": 9335
+    },
+    {
+      "epoch": 11.55225726654298,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.0376875950470617e-05,
+      "loss": 0.8581,
+      "step": 9340
+    },
+    {
+      "epoch": 11.558441558441558,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.0273579528846762e-05,
+      "loss": 0.8545,
+      "step": 9345
+    },
+    {
+      "epoch": 11.564625850340136,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.0170427700063873e-05,
+      "loss": 0.8601,
+      "step": 9350
+    },
+    {
+      "epoch": 11.570810142238713,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.0067420678029702e-05,
+      "loss": 0.8641,
+      "step": 9355
+    },
+    {
+      "epoch": 11.57699443413729,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.996455867635155e-05,
+      "loss": 0.8537,
+      "step": 9360
+    },
+    {
+      "epoch": 11.583178726035868,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.9861841908336098e-05,
+      "loss": 0.8596,
+      "step": 9365
+    },
+    {
+      "epoch": 11.589363017934446,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.9759270586988865e-05,
+      "loss": 0.8543,
+      "step": 9370
+    },
+    {
+      "epoch": 11.595547309833025,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.9656844925013637e-05,
+      "loss": 0.8673,
+      "step": 9375
+    },
+    {
+      "epoch": 11.601731601731602,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.9554565134812294e-05,
+      "loss": 0.8609,
+      "step": 9380
+    },
+    {
+      "epoch": 11.60791589363018,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.9452431428484062e-05,
+      "loss": 0.853,
+      "step": 9385
+    },
+    {
+      "epoch": 11.614100185528757,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.9350444017825385e-05,
+      "loss": 0.8552,
+      "step": 9390
+    },
+    {
+      "epoch": 11.620284477427335,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.9248603114329222e-05,
+      "loss": 0.8633,
+      "step": 9395
+    },
+    {
+      "epoch": 11.626468769325912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.9146908929184713e-05,
+      "loss": 0.8658,
+      "step": 9400
+    },
+    {
+      "epoch": 11.63265306122449,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.9045361673276872e-05,
+      "loss": 0.8517,
+      "step": 9405
+    },
+    {
+      "epoch": 11.638837353123067,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.894396155718585e-05,
+      "loss": 0.8599,
+      "step": 9410
+    },
+    {
+      "epoch": 11.645021645021645,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.8842708791186835e-05,
+      "loss": 0.8648,
+      "step": 9415
+    },
+    {
+      "epoch": 11.651205936920222,
+      "grad_norm": 0.4140625,
+      "learning_rate": 2.874160358524931e-05,
+      "loss": 0.8556,
+      "step": 9420
+    },
+    {
+      "epoch": 11.6573902288188,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.8640646149036898e-05,
+      "loss": 0.8598,
+      "step": 9425
+    },
+    {
+      "epoch": 11.663574520717377,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.853983669190664e-05,
+      "loss": 0.8665,
+      "step": 9430
+    },
+    {
+      "epoch": 11.669758812615955,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.8439175422908824e-05,
+      "loss": 0.8515,
+      "step": 9435
+    },
+    {
+      "epoch": 11.675943104514532,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.8338662550786443e-05,
+      "loss": 0.8591,
+      "step": 9440
+    },
+    {
+      "epoch": 11.682127396413112,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.823829828397465e-05,
+      "loss": 0.8331,
+      "step": 9445
+    },
+    {
+      "epoch": 11.688311688311689,
+      "grad_norm": 0.44921875,
+      "learning_rate": 2.8138082830600554e-05,
+      "loss": 0.8499,
+      "step": 9450
+    },
+    {
+      "epoch": 11.694495980210267,
+      "grad_norm": 0.443359375,
+      "learning_rate": 2.8038016398482593e-05,
+      "loss": 0.857,
+      "step": 9455
+    },
+    {
+      "epoch": 11.700680272108844,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.7938099195130153e-05,
+      "loss": 0.8609,
+      "step": 9460
+    },
+    {
+      "epoch": 11.706864564007422,
+      "grad_norm": 0.44140625,
+      "learning_rate": 2.7838331427743282e-05,
+      "loss": 0.853,
+      "step": 9465
+    },
+    {
+      "epoch": 11.713048855905999,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.7738713303211982e-05,
+      "loss": 0.8501,
+      "step": 9470
+    },
+    {
+      "epoch": 11.719233147804577,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.763924502811609e-05,
+      "loss": 0.8579,
+      "step": 9475
+    },
+    {
+      "epoch": 11.725417439703154,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.753992680872457e-05,
+      "loss": 0.8647,
+      "step": 9480
+    },
+    {
+      "epoch": 11.731601731601732,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.7440758850995318e-05,
+      "loss": 0.8558,
+      "step": 9485
+    },
+    {
+      "epoch": 11.737786023500309,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.7341741360574548e-05,
+      "loss": 0.8538,
+      "step": 9490
+    },
+    {
+      "epoch": 11.743970315398887,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.7242874542796482e-05,
+      "loss": 0.8588,
+      "step": 9495
+    },
+    {
+      "epoch": 11.750154607297464,
+      "grad_norm": 0.451171875,
+      "learning_rate": 2.7144158602682924e-05,
+      "loss": 0.8642,
+      "step": 9500
+    },
+    {
+      "epoch": 11.756338899196042,
+      "grad_norm": 0.451171875,
+      "learning_rate": 2.704559374494272e-05,
+      "loss": 0.8658,
+      "step": 9505
+    },
+    {
+      "epoch": 11.762523191094619,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.6947180173971508e-05,
+      "loss": 0.8672,
+      "step": 9510
+    },
+    {
+      "epoch": 11.768707482993197,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.6848918093851104e-05,
+      "loss": 0.8565,
+      "step": 9515
+    },
+    {
+      "epoch": 11.774891774891774,
+      "grad_norm": 0.4140625,
+      "learning_rate": 2.6750807708349267e-05,
+      "loss": 0.8646,
+      "step": 9520
+    },
+    {
+      "epoch": 11.781076066790352,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.665284922091912e-05,
+      "loss": 0.8566,
+      "step": 9525
+    },
+    {
+      "epoch": 11.78726035868893,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.6555042834698773e-05,
+      "loss": 0.853,
+      "step": 9530
+    },
+    {
+      "epoch": 11.793444650587508,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.6457388752511025e-05,
+      "loss": 0.8578,
+      "step": 9535
+    },
+    {
+      "epoch": 11.799628942486086,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.6359887176862718e-05,
+      "loss": 0.8506,
+      "step": 9540
+    },
+    {
+      "epoch": 11.805813234384663,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.626253830994455e-05,
+      "loss": 0.8466,
+      "step": 9545
+    },
+    {
+      "epoch": 11.81199752628324,
+      "grad_norm": 0.455078125,
+      "learning_rate": 2.6165342353630428e-05,
+      "loss": 0.8668,
+      "step": 9550
+    },
+    {
+      "epoch": 11.818181818181818,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.6068299509477266e-05,
+      "loss": 0.8625,
+      "step": 9555
+    },
+    {
+      "epoch": 11.824366110080396,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.5971409978724458e-05,
+      "loss": 0.8743,
+      "step": 9560
+    },
+    {
+      "epoch": 11.830550401978973,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.5874673962293373e-05,
+      "loss": 0.8559,
+      "step": 9565
+    },
+    {
+      "epoch": 11.83673469387755,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.577809166078716e-05,
+      "loss": 0.8542,
+      "step": 9570
+    },
+    {
+      "epoch": 11.842918985776128,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.5681663274490107e-05,
+      "loss": 0.8476,
+      "step": 9575
+    },
+    {
+      "epoch": 11.849103277674706,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.558538900336741e-05,
+      "loss": 0.8544,
+      "step": 9580
+    },
+    {
+      "epoch": 11.855287569573283,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.548926904706459e-05,
+      "loss": 0.8608,
+      "step": 9585
+    },
+    {
+      "epoch": 11.86147186147186,
+      "grad_norm": 0.46484375,
+      "learning_rate": 2.5393303604907205e-05,
+      "loss": 0.8591,
+      "step": 9590
+    },
+    {
+      "epoch": 11.867656153370438,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.529749287590042e-05,
+      "loss": 0.8516,
+      "step": 9595
+    },
+    {
+      "epoch": 11.873840445269018,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.5201837058728505e-05,
+      "loss": 0.8549,
+      "step": 9600
+    },
+    {
+      "epoch": 11.880024737167595,
+      "grad_norm": 0.443359375,
+      "learning_rate": 2.5106336351754578e-05,
+      "loss": 0.8498,
+      "step": 9605
+    },
+    {
+      "epoch": 11.886209029066173,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.5010990953019975e-05,
+      "loss": 0.8661,
+      "step": 9610
+    },
+    {
+      "epoch": 11.89239332096475,
+      "grad_norm": 0.455078125,
+      "learning_rate": 2.4915801060244092e-05,
+      "loss": 0.8668,
+      "step": 9615
+    },
+    {
+      "epoch": 11.898577612863328,
+      "grad_norm": 0.392578125,
+      "learning_rate": 2.4820766870823807e-05,
+      "loss": 0.8547,
+      "step": 9620
+    },
+    {
+      "epoch": 11.904761904761905,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.4725888581833058e-05,
+      "loss": 0.852,
+      "step": 9625
+    },
+    {
+      "epoch": 11.910946196660483,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.4631166390022574e-05,
+      "loss": 0.8616,
+      "step": 9630
+    },
+    {
+      "epoch": 11.91713048855906,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.4536600491819318e-05,
+      "loss": 0.8677,
+      "step": 9635
+    },
+    {
+      "epoch": 11.923314780457638,
+      "grad_norm": 0.408203125,
+      "learning_rate": 2.4442191083326195e-05,
+      "loss": 0.8516,
+      "step": 9640
+    },
+    {
+      "epoch": 11.929499072356215,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.4347938360321566e-05,
+      "loss": 0.8636,
+      "step": 9645
+    },
+    {
+      "epoch": 11.935683364254793,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.425384251825882e-05,
+      "loss": 0.8504,
+      "step": 9650
+    },
+    {
+      "epoch": 11.94186765615337,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.4159903752266156e-05,
+      "loss": 0.8598,
+      "step": 9655
+    },
+    {
+      "epoch": 11.948051948051948,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.4066122257145894e-05,
+      "loss": 0.8374,
+      "step": 9660
+    },
+    {
+      "epoch": 11.954236239950525,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.3972498227374342e-05,
+      "loss": 0.8554,
+      "step": 9665
+    },
+    {
+      "epoch": 11.960420531849103,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.387903185710115e-05,
+      "loss": 0.8525,
+      "step": 9670
+    },
+    {
+      "epoch": 11.96660482374768,
+      "grad_norm": 0.462890625,
+      "learning_rate": 2.3785723340149134e-05,
+      "loss": 0.8618,
+      "step": 9675
+    },
+    {
+      "epoch": 11.972789115646258,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.3692572870013718e-05,
+      "loss": 0.8645,
+      "step": 9680
+    },
+    {
+      "epoch": 11.978973407544837,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.359958063986256e-05,
+      "loss": 0.8472,
+      "step": 9685
+    },
+    {
+      "epoch": 11.985157699443414,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.3506746842535242e-05,
+      "loss": 0.8537,
+      "step": 9690
+    },
+    {
+      "epoch": 11.991341991341992,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.3414071670542703e-05,
+      "loss": 0.8494,
+      "step": 9695
+    },
+    {
+      "epoch": 11.99752628324057,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.3321555316067045e-05,
+      "loss": 0.8442,
+      "step": 9700
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 2.5192575454711914,
+      "eval_runtime": 0.5399,
+      "eval_samples_per_second": 18.52,
+      "eval_steps_per_second": 1.852,
+      "step": 9702
+    },
+    {
+      "epoch": 12.003710575139147,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.3229197970960924e-05,
+      "loss": 0.8504,
+      "step": 9705
+    },
+    {
+      "epoch": 12.009894867037724,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.313699982674736e-05,
+      "loss": 0.8496,
+      "step": 9710
+    },
+    {
+      "epoch": 12.016079158936302,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.3044961074619165e-05,
+      "loss": 0.8635,
+      "step": 9715
+    },
+    {
+      "epoch": 12.02226345083488,
+      "grad_norm": 0.46484375,
+      "learning_rate": 2.295308190543859e-05,
+      "loss": 0.8556,
+      "step": 9720
+    },
+    {
+      "epoch": 12.028447742733457,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.2861362509737072e-05,
+      "loss": 0.8588,
+      "step": 9725
+    },
+    {
+      "epoch": 12.034632034632034,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.276980307771458e-05,
+      "loss": 0.858,
+      "step": 9730
+    },
+    {
+      "epoch": 12.040816326530612,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.26784037992395e-05,
+      "loss": 0.848,
+      "step": 9735
+    },
+    {
+      "epoch": 12.04700061842919,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.2587164863847975e-05,
+      "loss": 0.8624,
+      "step": 9740
+    },
+    {
+      "epoch": 12.053184910327767,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.249608646074375e-05,
+      "loss": 0.8623,
+      "step": 9745
+    },
+    {
+      "epoch": 12.059369202226344,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.2405168778797646e-05,
+      "loss": 0.8504,
+      "step": 9750
+    },
+    {
+      "epoch": 12.065553494124922,
+      "grad_norm": 0.478515625,
+      "learning_rate": 2.2314412006547125e-05,
+      "loss": 0.8403,
+      "step": 9755
+    },
+    {
+      "epoch": 12.071737786023501,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.222381633219608e-05,
+      "loss": 0.859,
+      "step": 9760
+    },
+    {
+      "epoch": 12.077922077922079,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.2133381943614207e-05,
+      "loss": 0.8639,
+      "step": 9765
+    },
+    {
+      "epoch": 12.084106369820656,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.204310902833685e-05,
+      "loss": 0.859,
+      "step": 9770
+    },
+    {
+      "epoch": 12.090290661719234,
+      "grad_norm": 0.451171875,
+      "learning_rate": 2.1952997773564467e-05,
+      "loss": 0.8598,
+      "step": 9775
+    },
+    {
+      "epoch": 12.096474953617811,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.1863048366162208e-05,
+      "loss": 0.8472,
+      "step": 9780
+    },
+    {
+      "epoch": 12.102659245516389,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.1773260992659728e-05,
+      "loss": 0.8557,
+      "step": 9785
+    },
+    {
+      "epoch": 12.108843537414966,
+      "grad_norm": 0.443359375,
+      "learning_rate": 2.1683635839250537e-05,
+      "loss": 0.8597,
+      "step": 9790
+    },
+    {
+      "epoch": 12.115027829313544,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.159417309179189e-05,
+      "loss": 0.8594,
+      "step": 9795
+    },
+    {
+      "epoch": 12.121212121212121,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.15048729358041e-05,
+      "loss": 0.859,
+      "step": 9800
+    },
+    {
+      "epoch": 12.127396413110699,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.141573555647042e-05,
+      "loss": 0.8495,
+      "step": 9805
+    },
+    {
+      "epoch": 12.133580705009276,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.1326761138636553e-05,
+      "loss": 0.8567,
+      "step": 9810
+    },
+    {
+      "epoch": 12.139764996907854,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.123794986681017e-05,
+      "loss": 0.8475,
+      "step": 9815
+    },
+    {
+      "epoch": 12.145949288806431,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.114930192516076e-05,
+      "loss": 0.8616,
+      "step": 9820
+    },
+    {
+      "epoch": 12.152133580705009,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.106081749751897e-05,
+      "loss": 0.8602,
+      "step": 9825
+    },
+    {
+      "epoch": 12.158317872603586,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.097249676737648e-05,
+      "loss": 0.8583,
+      "step": 9830
+    },
+    {
+      "epoch": 12.164502164502165,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.0884339917885433e-05,
+      "loss": 0.8533,
+      "step": 9835
+    },
+    {
+      "epoch": 12.170686456400743,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.0796347131858186e-05,
+      "loss": 0.8532,
+      "step": 9840
+    },
+    {
+      "epoch": 12.17687074829932,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.0708518591766825e-05,
+      "loss": 0.8622,
+      "step": 9845
+    },
+    {
+      "epoch": 12.183055040197898,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.0620854479742834e-05,
+      "loss": 0.8548,
+      "step": 9850
+    },
+    {
+      "epoch": 12.189239332096475,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.05333549775768e-05,
+      "loss": 0.8512,
+      "step": 9855
+    },
+    {
+      "epoch": 12.195423623995053,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.044602026671786e-05,
+      "loss": 0.8632,
+      "step": 9860
+    },
+    {
+      "epoch": 12.20160791589363,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.0358850528273455e-05,
+      "loss": 0.856,
+      "step": 9865
+    },
+    {
+      "epoch": 12.207792207792208,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.027184594300898e-05,
+      "loss": 0.856,
+      "step": 9870
+    },
+    {
+      "epoch": 12.213976499690785,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.018500669134723e-05,
+      "loss": 0.8541,
+      "step": 9875
+    },
+    {
+      "epoch": 12.220160791589363,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.0098332953368272e-05,
+      "loss": 0.86,
+      "step": 9880
+    },
+    {
+      "epoch": 12.22634508348794,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.0011824908808808e-05,
+      "loss": 0.8557,
+      "step": 9885
+    },
+    {
+      "epoch": 12.232529375386518,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.9925482737062085e-05,
+      "loss": 0.8758,
+      "step": 9890
+    },
+    {
+      "epoch": 12.238713667285095,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.9839306617177243e-05,
+      "loss": 0.8628,
+      "step": 9895
+    },
+    {
+      "epoch": 12.244897959183673,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.9753296727859195e-05,
+      "loss": 0.854,
+      "step": 9900
+    },
+    {
+      "epoch": 12.25108225108225,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.966745324746806e-05,
+      "loss": 0.8571,
+      "step": 9905
+    },
+    {
+      "epoch": 12.257266542980828,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.9581776354018854e-05,
+      "loss": 0.8629,
+      "step": 9910
+    },
+    {
+      "epoch": 12.263450834879407,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.9496266225181248e-05,
+      "loss": 0.8587,
+      "step": 9915
+    },
+    {
+      "epoch": 12.269635126777985,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.941092303827896e-05,
+      "loss": 0.8526,
+      "step": 9920
+    },
+    {
+      "epoch": 12.275819418676562,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.9325746970289627e-05,
+      "loss": 0.8631,
+      "step": 9925
+    },
+    {
+      "epoch": 12.28200371057514,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.9240738197844278e-05,
+      "loss": 0.8637,
+      "step": 9930
+    },
+    {
+      "epoch": 12.288188002473717,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.9155896897226988e-05,
+      "loss": 0.8544,
+      "step": 9935
+    },
+    {
+      "epoch": 12.294372294372295,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.9071223244374614e-05,
+      "loss": 0.8676,
+      "step": 9940
+    },
+    {
+      "epoch": 12.300556586270872,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.89867174148763e-05,
+      "loss": 0.8686,
+      "step": 9945
+    },
+    {
+      "epoch": 12.30674087816945,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.8902379583973208e-05,
+      "loss": 0.8476,
+      "step": 9950
+    },
+    {
+      "epoch": 12.312925170068027,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.8818209926558082e-05,
+      "loss": 0.8516,
+      "step": 9955
+    },
+    {
+      "epoch": 12.319109461966605,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.8734208617174988e-05,
+      "loss": 0.8567,
+      "step": 9960
+    },
+    {
+      "epoch": 12.325293753865182,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.8650375830018785e-05,
+      "loss": 0.8539,
+      "step": 9965
+    },
+    {
+      "epoch": 12.33147804576376,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.856671173893497e-05,
+      "loss": 0.8477,
+      "step": 9970
+    },
+    {
+      "epoch": 12.337662337662337,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.8483216517419142e-05,
+      "loss": 0.8546,
+      "step": 9975
+    },
+    {
+      "epoch": 12.343846629560915,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.839989033861673e-05,
+      "loss": 0.8508,
+      "step": 9980
+    },
+    {
+      "epoch": 12.350030921459492,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.8316733375322637e-05,
+      "loss": 0.8376,
+      "step": 9985
+    },
+    {
+      "epoch": 12.35621521335807,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.8233745799980817e-05,
+      "loss": 0.8456,
+      "step": 9990
+    },
+    {
+      "epoch": 12.362399505256649,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.8150927784684e-05,
+      "loss": 0.8578,
+      "step": 9995
+    },
+    {
+      "epoch": 12.368583797155226,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.8068279501173335e-05,
+      "loss": 0.8569,
+      "step": 10000
+    },
+    {
+      "epoch": 12.374768089053804,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.7985801120837865e-05,
+      "loss": 0.8492,
+      "step": 10005
+    },
+    {
+      "epoch": 12.380952380952381,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.790349281471445e-05,
+      "loss": 0.8583,
+      "step": 10010
+    },
+    {
+      "epoch": 12.387136672850959,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.782135475348714e-05,
+      "loss": 0.8493,
+      "step": 10015
+    },
+    {
+      "epoch": 12.393320964749536,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.773938710748706e-05,
+      "loss": 0.8574,
+      "step": 10020
+    },
+    {
+      "epoch": 12.399505256648114,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.765759004669183e-05,
+      "loss": 0.8448,
+      "step": 10025
+    },
+    {
+      "epoch": 12.405689548546691,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.757596374072543e-05,
+      "loss": 0.8523,
+      "step": 10030
+    },
+    {
+      "epoch": 12.411873840445269,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.7494508358857677e-05,
+      "loss": 0.8663,
+      "step": 10035
+    },
+    {
+      "epoch": 12.418058132343846,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.741322407000391e-05,
+      "loss": 0.8499,
+      "step": 10040
+    },
+    {
+      "epoch": 12.424242424242424,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.7332111042724775e-05,
+      "loss": 0.8595,
+      "step": 10045
+    },
+    {
+      "epoch": 12.430426716141001,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.7251169445225657e-05,
+      "loss": 0.8644,
+      "step": 10050
+    },
+    {
+      "epoch": 12.436611008039579,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.7170399445356532e-05,
+      "loss": 0.8636,
+      "step": 10055
+    },
+    {
+      "epoch": 12.442795299938156,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.70898012106115e-05,
+      "loss": 0.8547,
+      "step": 10060
+    },
+    {
+      "epoch": 12.448979591836734,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.700937490812844e-05,
+      "loss": 0.8547,
+      "step": 10065
+    },
+    {
+      "epoch": 12.455163883735313,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.692912070468874e-05,
+      "loss": 0.8514,
+      "step": 10070
+    },
+    {
+      "epoch": 12.46134817563389,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.684903876671685e-05,
+      "loss": 0.849,
+      "step": 10075
+    },
+    {
+      "epoch": 12.467532467532468,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.676912926028007e-05,
+      "loss": 0.8497,
+      "step": 10080
+    },
+    {
+      "epoch": 12.473716759431046,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.668939235108802e-05,
+      "loss": 0.8668,
+      "step": 10085
+    },
+    {
+      "epoch": 12.479901051329623,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.660982820449247e-05,
+      "loss": 0.8532,
+      "step": 10090
+    },
+    {
+      "epoch": 12.4860853432282,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.6530436985486996e-05,
+      "loss": 0.8464,
+      "step": 10095
+    },
+    {
+      "epoch": 12.492269635126778,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.6451218858706374e-05,
+      "loss": 0.8589,
+      "step": 10100
+    },
+    {
+      "epoch": 12.498453927025356,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.637217398842663e-05,
+      "loss": 0.8584,
+      "step": 10105
+    },
+    {
+      "epoch": 12.504638218923933,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.6293302538564382e-05,
+      "loss": 0.8587,
+      "step": 10110
+    },
+    {
+      "epoch": 12.51082251082251,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.6214604672676704e-05,
+      "loss": 0.8626,
+      "step": 10115
+    },
+    {
+      "epoch": 12.517006802721088,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.6136080553960687e-05,
+      "loss": 0.8634,
+      "step": 10120
+    },
+    {
+      "epoch": 12.523191094619666,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.6057730345253065e-05,
+      "loss": 0.8508,
+      "step": 10125
+    },
+    {
+      "epoch": 12.529375386518243,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.5979554209030024e-05,
+      "loss": 0.8541,
+      "step": 10130
+    },
+    {
+      "epoch": 12.53555967841682,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.5901552307406653e-05,
+      "loss": 0.8523,
+      "step": 10135
+    },
+    {
+      "epoch": 12.541743970315398,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.5823724802136865e-05,
+      "loss": 0.8674,
+      "step": 10140
+    },
+    {
+      "epoch": 12.547928262213976,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.5746071854612797e-05,
+      "loss": 0.859,
+      "step": 10145
+    },
+    {
+      "epoch": 12.554112554112555,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.5668593625864715e-05,
+      "loss": 0.8582,
+      "step": 10150
+    },
+    {
+      "epoch": 12.560296846011132,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.5591290276560466e-05,
+      "loss": 0.857,
+      "step": 10155
+    },
+    {
+      "epoch": 12.56648113790971,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.5514161967005337e-05,
+      "loss": 0.8556,
+      "step": 10160
+    },
+    {
+      "epoch": 12.572665429808287,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.543720885714157e-05,
+      "loss": 0.854,
+      "step": 10165
+    },
+    {
+      "epoch": 12.578849721706865,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.536043110654809e-05,
+      "loss": 0.8486,
+      "step": 10170
+    },
+    {
+      "epoch": 12.585034013605442,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.528382887444022e-05,
+      "loss": 0.8531,
+      "step": 10175
+    },
+    {
+      "epoch": 12.59121830550402,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.5207402319669306e-05,
+      "loss": 0.8527,
+      "step": 10180
+    },
+    {
+      "epoch": 12.597402597402597,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.5131151600722337e-05,
+      "loss": 0.8534,
+      "step": 10185
+    },
+    {
+      "epoch": 12.603586889301175,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.505507687572173e-05,
+      "loss": 0.8519,
+      "step": 10190
+    },
+    {
+      "epoch": 12.609771181199752,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.4979178302424867e-05,
+      "loss": 0.8637,
+      "step": 10195
+    },
+    {
+      "epoch": 12.61595547309833,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.4903456038223939e-05,
+      "loss": 0.8663,
+      "step": 10200
+    },
+    {
+      "epoch": 12.622139764996907,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.4827910240145426e-05,
+      "loss": 0.858,
+      "step": 10205
+    },
+    {
+      "epoch": 12.628324056895485,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.4752541064849946e-05,
+      "loss": 0.8537,
+      "step": 10210
+    },
+    {
+      "epoch": 12.634508348794062,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.4677348668631763e-05,
+      "loss": 0.8485,
+      "step": 10215
+    },
+    {
+      "epoch": 12.64069264069264,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.4602333207418651e-05,
+      "loss": 0.8529,
+      "step": 10220
+    },
+    {
+      "epoch": 12.64687693259122,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.4527494836771438e-05,
+      "loss": 0.8639,
+      "step": 10225
+    },
+    {
+      "epoch": 12.653061224489797,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.4452833711883628e-05,
+      "loss": 0.8531,
+      "step": 10230
+    },
+    {
+      "epoch": 12.659245516388374,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.4378349987581307e-05,
+      "loss": 0.8619,
+      "step": 10235
+    },
+    {
+      "epoch": 12.665429808286952,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.4304043818322565e-05,
+      "loss": 0.8556,
+      "step": 10240
+    },
+    {
+      "epoch": 12.67161410018553,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.4229915358197377e-05,
+      "loss": 0.8543,
+      "step": 10245
+    },
+    {
+      "epoch": 12.677798392084107,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.4155964760927176e-05,
+      "loss": 0.8553,
+      "step": 10250
+    },
+    {
+      "epoch": 12.683982683982684,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.4082192179864518e-05,
+      "loss": 0.8439,
+      "step": 10255
+    },
+    {
+      "epoch": 12.690166975881262,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.4008597767992871e-05,
+      "loss": 0.8558,
+      "step": 10260
+    },
+    {
+      "epoch": 12.69635126777984,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.3935181677926156e-05,
+      "loss": 0.858,
+      "step": 10265
+    },
+    {
+      "epoch": 12.702535559678417,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.3861944061908583e-05,
+      "loss": 0.8561,
+      "step": 10270
+    },
+    {
+      "epoch": 12.708719851576994,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.3788885071814172e-05,
+      "loss": 0.8531,
+      "step": 10275
+    },
+    {
+      "epoch": 12.714904143475572,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.3716004859146592e-05,
+      "loss": 0.8465,
+      "step": 10280
+    },
+    {
+      "epoch": 12.72108843537415,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.3643303575038779e-05,
+      "loss": 0.8433,
+      "step": 10285
+    },
+    {
+      "epoch": 12.727272727272727,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.3570781370252582e-05,
+      "loss": 0.8566,
+      "step": 10290
+    },
+    {
+      "epoch": 12.733457019171304,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.3498438395178492e-05,
+      "loss": 0.8578,
+      "step": 10295
+    },
+    {
+      "epoch": 12.739641311069882,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.3426274799835337e-05,
+      "loss": 0.8606,
+      "step": 10300
+    },
+    {
+      "epoch": 12.745825602968461,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.3354290733869979e-05,
+      "loss": 0.8597,
+      "step": 10305
+    },
+    {
+      "epoch": 12.752009894867038,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.328248634655701e-05,
+      "loss": 0.8595,
+      "step": 10310
+    },
+    {
+      "epoch": 12.758194186765616,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.3210861786798335e-05,
+      "loss": 0.8615,
+      "step": 10315
+    },
+    {
+      "epoch": 12.764378478664193,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.3139417203123027e-05,
+      "loss": 0.8588,
+      "step": 10320
+    },
+    {
+      "epoch": 12.770562770562771,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.306815274368689e-05,
+      "loss": 0.8572,
+      "step": 10325
+    },
+    {
+      "epoch": 12.776747062461348,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.2997068556272263e-05,
+      "loss": 0.8521,
+      "step": 10330
+    },
+    {
+      "epoch": 12.782931354359926,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.2926164788287543e-05,
+      "loss": 0.852,
+      "step": 10335
+    },
+    {
+      "epoch": 12.789115646258503,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.2855441586767113e-05,
+      "loss": 0.8662,
+      "step": 10340
+    },
+    {
+      "epoch": 12.79529993815708,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.278489909837085e-05,
+      "loss": 0.8589,
+      "step": 10345
+    },
+    {
+      "epoch": 12.801484230055658,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.2714537469383858e-05,
+      "loss": 0.8578,
+      "step": 10350
+    },
+    {
+      "epoch": 12.807668521954236,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.2644356845716287e-05,
+      "loss": 0.8596,
+      "step": 10355
+    },
+    {
+      "epoch": 12.813852813852813,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.2574357372902767e-05,
+      "loss": 0.8585,
+      "step": 10360
+    },
+    {
+      "epoch": 12.82003710575139,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.2504539196102439e-05,
+      "loss": 0.8529,
+      "step": 10365
+    },
+    {
+      "epoch": 12.826221397649968,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.243490246009842e-05,
+      "loss": 0.8488,
+      "step": 10370
+    },
+    {
+      "epoch": 12.832405689548546,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.2365447309297529e-05,
+      "loss": 0.8562,
+      "step": 10375
+    },
+    {
+      "epoch": 12.838589981447125,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.2296173887730123e-05,
+      "loss": 0.8609,
+      "step": 10380
+    },
+    {
+      "epoch": 12.844774273345703,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.2227082339049612e-05,
+      "loss": 0.862,
+      "step": 10385
+    },
+    {
+      "epoch": 12.85095856524428,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.215817280653232e-05,
+      "loss": 0.8521,
+      "step": 10390
+    },
+    {
+      "epoch": 12.857142857142858,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.2089445433077073e-05,
+      "loss": 0.8468,
+      "step": 10395
+    },
+    {
+      "epoch": 12.863327149041435,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.2020900361204968e-05,
+      "loss": 0.855,
+      "step": 10400
+    },
+    {
+      "epoch": 12.869511440940013,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.19525377330591e-05,
+      "loss": 0.8517,
+      "step": 10405
+    },
+    {
+      "epoch": 12.87569573283859,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.1884357690404158e-05,
+      "loss": 0.8631,
+      "step": 10410
+    },
+    {
+      "epoch": 12.881880024737168,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.1816360374626245e-05,
+      "loss": 0.8458,
+      "step": 10415
+    },
+    {
+      "epoch": 12.888064316635745,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.1748545926732535e-05,
+      "loss": 0.8588,
+      "step": 10420
+    },
+    {
+      "epoch": 12.894248608534323,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.1680914487350959e-05,
+      "loss": 0.8574,
+      "step": 10425
+    },
+    {
+      "epoch": 12.9004329004329,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.1613466196729984e-05,
+      "loss": 0.8485,
+      "step": 10430
+    },
+    {
+      "epoch": 12.906617192331478,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.1546201194738227e-05,
+      "loss": 0.8718,
+      "step": 10435
+    },
+    {
+      "epoch": 12.912801484230055,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.1479119620864276e-05,
+      "loss": 0.851,
+      "step": 10440
+    },
+    {
+      "epoch": 12.918985776128633,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.1412221614216278e-05,
+      "loss": 0.846,
+      "step": 10445
+    },
+    {
+      "epoch": 12.92517006802721,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.1345507313521786e-05,
+      "loss": 0.8575,
+      "step": 10450
+    },
+    {
+      "epoch": 12.931354359925788,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.1278976857127311e-05,
+      "loss": 0.8535,
+      "step": 10455
+    },
+    {
+      "epoch": 12.937538651824367,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.1212630382998213e-05,
+      "loss": 0.8522,
+      "step": 10460
+    },
+    {
+      "epoch": 12.943722943722944,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.1146468028718237e-05,
+      "loss": 0.8622,
+      "step": 10465
+    },
+    {
+      "epoch": 12.949907235621522,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.1080489931489391e-05,
+      "loss": 0.8574,
+      "step": 10470
+    },
+    {
+      "epoch": 12.9560915275201,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.101469622813157e-05,
+      "loss": 0.867,
+      "step": 10475
+    },
+    {
+      "epoch": 12.962275819418677,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0949087055082252e-05,
+      "loss": 0.8508,
+      "step": 10480
+    },
+    {
+      "epoch": 12.968460111317254,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0883662548396257e-05,
+      "loss": 0.8491,
+      "step": 10485
+    },
+    {
+      "epoch": 12.974644403215832,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0818422843745512e-05,
+      "loss": 0.8596,
+      "step": 10490
+    },
+    {
+      "epoch": 12.98082869511441,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.0753368076418647e-05,
+      "loss": 0.8611,
+      "step": 10495
+    },
+    {
+      "epoch": 12.987012987012987,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.0688498381320855e-05,
+      "loss": 0.8466,
+      "step": 10500
+    },
+    {
+      "epoch": 12.993197278911564,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.0623813892973455e-05,
+      "loss": 0.8582,
+      "step": 10505
+    },
+    {
+      "epoch": 12.999381570810142,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0559314745513805e-05,
+      "loss": 0.8621,
+      "step": 10510
+    },
+    {
+      "epoch": 12.999381570810142,
+      "eval_loss": 2.520648717880249,
+      "eval_runtime": 0.6408,
+      "eval_samples_per_second": 15.605,
+      "eval_steps_per_second": 1.56,
+      "step": 10510
+    },
+    {
+      "epoch": 13.00556586270872,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.049500107269481e-05,
+      "loss": 0.8552,
+      "step": 10515
+    },
+    {
+      "epoch": 13.011750154607297,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.0430873007884857e-05,
+      "loss": 0.8558,
+      "step": 10520
+    },
+    {
+      "epoch": 13.017934446505874,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.0366930684067333e-05,
+      "loss": 0.8587,
+      "step": 10525
+    },
+    {
+      "epoch": 13.024118738404452,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0303174233840528e-05,
+      "loss": 0.861,
+      "step": 10530
+    },
+    {
+      "epoch": 13.030303030303031,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.0239603789417274e-05,
+      "loss": 0.8619,
+      "step": 10535
+    },
+    {
+      "epoch": 13.036487322201609,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.0176219482624616e-05,
+      "loss": 0.8493,
+      "step": 10540
+    },
+    {
+      "epoch": 13.042671614100186,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.0113021444903726e-05,
+      "loss": 0.8582,
+      "step": 10545
+    },
+    {
+      "epoch": 13.048855905998764,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.0050009807309325e-05,
+      "loss": 0.8486,
+      "step": 10550
+    },
+    {
+      "epoch": 13.055040197897341,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.987184700509755e-06,
+      "loss": 0.8461,
+      "step": 10555
+    },
+    {
+      "epoch": 13.061224489795919,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.924546254786493e-06,
+      "loss": 0.8626,
+      "step": 10560
+    },
+    {
+      "epoch": 13.067408781694496,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.862094600033912e-06,
+      "loss": 0.861,
+      "step": 10565
+    },
+    {
+      "epoch": 13.073593073593074,
+      "grad_norm": 0.416015625,
+      "learning_rate": 9.799829865759069e-06,
+      "loss": 0.8576,
+      "step": 10570
+    },
+    {
+      "epoch": 13.079777365491651,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.737752181081338e-06,
+      "loss": 0.8471,
+      "step": 10575
+    },
+    {
+      "epoch": 13.085961657390229,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.675861674732312e-06,
+      "loss": 0.8518,
+      "step": 10580
+    },
+    {
+      "epoch": 13.092145949288806,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.614158475055302e-06,
+      "loss": 0.859,
+      "step": 10585
+    },
+    {
+      "epoch": 13.098330241187384,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.552642710005299e-06,
+      "loss": 0.8624,
+      "step": 10590
+    },
+    {
+      "epoch": 13.104514533085961,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.491314507148597e-06,
+      "loss": 0.8538,
+      "step": 10595
+    },
+    {
+      "epoch": 13.110698824984539,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.430173993662451e-06,
+      "loss": 0.8567,
+      "step": 10600
+    },
+    {
+      "epoch": 13.116883116883116,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.369221296335006e-06,
+      "loss": 0.8544,
+      "step": 10605
+    },
+    {
+      "epoch": 13.123067408781694,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.308456541564881e-06,
+      "loss": 0.8461,
+      "step": 10610
+    },
+    {
+      "epoch": 13.129251700680273,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.24787985536094e-06,
+      "loss": 0.8612,
+      "step": 10615
+    },
+    {
+      "epoch": 13.13543599257885,
+      "grad_norm": 0.4140625,
+      "learning_rate": 9.187491363342093e-06,
+      "loss": 0.8727,
+      "step": 10620
+    },
+    {
+      "epoch": 13.141620284477428,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.127291190736943e-06,
+      "loss": 0.8599,
+      "step": 10625
+    },
+    {
+      "epoch": 13.147804576376005,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.067279462383615e-06,
+      "loss": 0.8478,
+      "step": 10630
+    },
+    {
+      "epoch": 13.153988868274583,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.0074563027294e-06,
+      "loss": 0.8527,
+      "step": 10635
+    },
+    {
+      "epoch": 13.16017316017316,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.947821835830616e-06,
+      "loss": 0.8508,
+      "step": 10640
+    },
+    {
+      "epoch": 13.166357452071738,
+      "grad_norm": 0.447265625,
+      "learning_rate": 8.88837618535222e-06,
+      "loss": 0.8556,
+      "step": 10645
+    },
+    {
+      "epoch": 13.172541743970315,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.829119474567671e-06,
+      "loss": 0.8484,
+      "step": 10650
+    },
+    {
+      "epoch": 13.178726035868893,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.770051826358594e-06,
+      "loss": 0.8635,
+      "step": 10655
+    },
+    {
+      "epoch": 13.18491032776747,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.711173363214553e-06,
+      "loss": 0.849,
+      "step": 10660
+    },
+    {
+      "epoch": 13.191094619666048,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.652484207232803e-06,
+      "loss": 0.855,
+      "step": 10665
+    },
+    {
+      "epoch": 13.197278911564625,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.593984480118011e-06,
+      "loss": 0.8637,
+      "step": 10670
+    },
+    {
+      "epoch": 13.203463203463203,
+      "grad_norm": 0.419921875,
+      "learning_rate": 8.535674303182061e-06,
+      "loss": 0.855,
+      "step": 10675
+    },
+    {
+      "epoch": 13.20964749536178,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.47755379734373e-06,
+      "loss": 0.8649,
+      "step": 10680
+    },
+    {
+      "epoch": 13.215831787260358,
+      "grad_norm": 0.42578125,
+      "learning_rate": 8.419623083128458e-06,
+      "loss": 0.8694,
+      "step": 10685
+    },
+    {
+      "epoch": 13.222016079158937,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.361882280668165e-06,
+      "loss": 0.8572,
+      "step": 10690
+    },
+    {
+      "epoch": 13.228200371057515,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.304331509700891e-06,
+      "loss": 0.8529,
+      "step": 10695
+    },
+    {
+      "epoch": 13.234384662956092,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.24697088957066e-06,
+      "loss": 0.863,
+      "step": 10700
+    },
+    {
+      "epoch": 13.24056895485467,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.189800539227111e-06,
+      "loss": 0.86,
+      "step": 10705
+    },
+    {
+      "epoch": 13.246753246753247,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.132820577225387e-06,
+      "loss": 0.8673,
+      "step": 10710
+    },
+    {
+      "epoch": 13.252937538651825,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.076031121725746e-06,
+      "loss": 0.8537,
+      "step": 10715
+    },
+    {
+      "epoch": 13.259121830550402,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.019432290493457e-06,
+      "loss": 0.8516,
+      "step": 10720
+    },
+    {
+      "epoch": 13.26530612244898,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.963024200898462e-06,
+      "loss": 0.859,
+      "step": 10725
+    },
+    {
+      "epoch": 13.271490414347557,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.906806969915148e-06,
+      "loss": 0.8595,
+      "step": 10730
+    },
+    {
+      "epoch": 13.277674706246135,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.85078071412213e-06,
+      "loss": 0.8483,
+      "step": 10735
+    },
+    {
+      "epoch": 13.283858998144712,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.794945549701993e-06,
+      "loss": 0.8498,
+      "step": 10740
+    },
+    {
+      "epoch": 13.29004329004329,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.739301592441017e-06,
+      "loss": 0.859,
+      "step": 10745
+    },
+    {
+      "epoch": 13.296227581941867,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.683848957729056e-06,
+      "loss": 0.8533,
+      "step": 10750
+    },
+    {
+      "epoch": 13.302411873840445,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.6285877605591135e-06,
+      "loss": 0.8526,
+      "step": 10755
+    },
+    {
+      "epoch": 13.308596165739022,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.573518115527289e-06,
+      "loss": 0.8536,
+      "step": 10760
+    },
+    {
+      "epoch": 13.3147804576376,
+      "grad_norm": 0.435546875,
+      "learning_rate": 7.5186401368324e-06,
+      "loss": 0.8591,
+      "step": 10765
+    },
+    {
+      "epoch": 13.320964749536179,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.463953938275858e-06,
+      "loss": 0.8542,
+      "step": 10770
+    },
+    {
+      "epoch": 13.327149041434756,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.409459633261307e-06,
+      "loss": 0.8495,
+      "step": 10775
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 0.435546875,
+      "learning_rate": 7.355157334794516e-06,
+      "loss": 0.8547,
+      "step": 10780
+    },
+    {
+      "epoch": 13.339517625231911,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.3010471554830766e-06,
+      "loss": 0.8581,
+      "step": 10785
+    },
+    {
+      "epoch": 13.345701917130489,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.247129207536152e-06,
+      "loss": 0.8583,
+      "step": 10790
+    },
+    {
+      "epoch": 13.351886209029066,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.193403602764315e-06,
+      "loss": 0.8665,
+      "step": 10795
+    },
+    {
+      "epoch": 13.358070500927644,
+      "grad_norm": 0.443359375,
+      "learning_rate": 7.1398704525792e-06,
+      "loss": 0.8528,
+      "step": 10800
+    },
+    {
+      "epoch": 13.364254792826221,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.086529867993453e-06,
+      "loss": 0.8538,
+      "step": 10805
+    },
+    {
+      "epoch": 13.370439084724799,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.0333819596203e-06,
+      "loss": 0.8579,
+      "step": 10810
+    },
+    {
+      "epoch": 13.376623376623376,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.980426837673437e-06,
+      "loss": 0.8534,
+      "step": 10815
+    },
+    {
+      "epoch": 13.382807668521954,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.927664611966811e-06,
+      "loss": 0.8529,
+      "step": 10820
+    },
+    {
+      "epoch": 13.388991960420531,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.875095391914299e-06,
+      "loss": 0.8577,
+      "step": 10825
+    },
+    {
+      "epoch": 13.395176252319109,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.8227192865295995e-06,
+      "loss": 0.8588,
+      "step": 10830
+    },
+    {
+      "epoch": 13.401360544217686,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.770536404425887e-06,
+      "loss": 0.8484,
+      "step": 10835
+    },
+    {
+      "epoch": 13.407544836116264,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.718546853815688e-06,
+      "loss": 0.8584,
+      "step": 10840
+    },
+    {
+      "epoch": 13.413729128014843,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.666750742510619e-06,
+      "loss": 0.8484,
+      "step": 10845
+    },
+    {
+      "epoch": 13.41991341991342,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.6151481779211155e-06,
+      "loss": 0.8478,
+      "step": 10850
+    },
+    {
+      "epoch": 13.426097711811998,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.56373926705629e-06,
+      "loss": 0.8502,
+      "step": 10855
+    },
+    {
+      "epoch": 13.432282003710576,
+      "grad_norm": 0.419921875,
+      "learning_rate": 6.512524116523633e-06,
+      "loss": 0.8595,
+      "step": 10860
+    },
+    {
+      "epoch": 13.438466295609153,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.461502832528887e-06,
+      "loss": 0.8675,
+      "step": 10865
+    },
+    {
+      "epoch": 13.44465058750773,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.410675520875742e-06,
+      "loss": 0.8641,
+      "step": 10870
+    },
+    {
+      "epoch": 13.450834879406308,
+      "grad_norm": 0.42578125,
+      "learning_rate": 6.360042286965595e-06,
+      "loss": 0.8618,
+      "step": 10875
+    },
+    {
+      "epoch": 13.457019171304886,
+      "grad_norm": 0.4609375,
+      "learning_rate": 6.30960323579749e-06,
+      "loss": 0.8606,
+      "step": 10880
+    },
+    {
+      "epoch": 13.463203463203463,
+      "grad_norm": 0.43359375,
+      "learning_rate": 6.2593584719676805e-06,
+      "loss": 0.8486,
+      "step": 10885
+    },
+    {
+      "epoch": 13.46938775510204,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.209308099669597e-06,
+      "loss": 0.855,
+      "step": 10890
+    },
+    {
+      "epoch": 13.475572047000618,
+      "grad_norm": 0.427734375,
+      "learning_rate": 6.159452222693507e-06,
+      "loss": 0.8553,
+      "step": 10895
+    },
+    {
+      "epoch": 13.481756338899196,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.109790944426397e-06,
+      "loss": 0.8564,
+      "step": 10900
+    },
+    {
+      "epoch": 13.487940630797773,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.0603243678516995e-06,
+      "loss": 0.8541,
+      "step": 10905
+    },
+    {
+      "epoch": 13.49412492269635,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.011052595549038e-06,
+      "loss": 0.8528,
+      "step": 10910
+    },
+    {
+      "epoch": 13.500309214594928,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.961975729694158e-06,
+      "loss": 0.8538,
+      "step": 10915
+    },
+    {
+      "epoch": 13.506493506493506,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.913093872058528e-06,
+      "loss": 0.8546,
+      "step": 10920
+    },
+    {
+      "epoch": 13.512677798392085,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.864407124009297e-06,
+      "loss": 0.8573,
+      "step": 10925
+    },
+    {
+      "epoch": 13.518862090290662,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.81591558650898e-06,
+      "loss": 0.8529,
+      "step": 10930
+    },
+    {
+      "epoch": 13.52504638218924,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.767619360115295e-06,
+      "loss": 0.8528,
+      "step": 10935
+    },
+    {
+      "epoch": 13.531230674087817,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.719518544980929e-06,
+      "loss": 0.8482,
+      "step": 10940
+    },
+    {
+      "epoch": 13.537414965986395,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.6716132408533355e-06,
+      "loss": 0.852,
+      "step": 10945
+    },
+    {
+      "epoch": 13.543599257884972,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.623903547074549e-06,
+      "loss": 0.8517,
+      "step": 10950
+    },
+    {
+      "epoch": 13.54978354978355,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.5763895625809415e-06,
+      "loss": 0.8595,
+      "step": 10955
+    },
+    {
+      "epoch": 13.555967841682127,
+      "grad_norm": 0.451171875,
+      "learning_rate": 5.529071385903084e-06,
+      "loss": 0.8475,
+      "step": 10960
+    },
+    {
+      "epoch": 13.562152133580705,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.481949115165452e-06,
+      "loss": 0.8665,
+      "step": 10965
+    },
+    {
+      "epoch": 13.568336425479282,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.43502284808628e-06,
+      "loss": 0.8565,
+      "step": 10970
+    },
+    {
+      "epoch": 13.57452071737786,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.38829268197738e-06,
+      "loss": 0.8591,
+      "step": 10975
+    },
+    {
+      "epoch": 13.580705009276437,
+      "grad_norm": 0.447265625,
+      "learning_rate": 5.341758713743828e-06,
+      "loss": 0.8511,
+      "step": 10980
+    },
+    {
+      "epoch": 13.586889301175015,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.295421039883941e-06,
+      "loss": 0.8608,
+      "step": 10985
+    },
+    {
+      "epoch": 13.593073593073592,
+      "grad_norm": 0.4609375,
+      "learning_rate": 5.249279756488878e-06,
+      "loss": 0.8418,
+      "step": 10990
+    },
+    {
+      "epoch": 13.59925788497217,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.2033349592426335e-06,
+      "loss": 0.8517,
+      "step": 10995
+    },
+    {
+      "epoch": 13.60544217687075,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.157586743421672e-06,
+      "loss": 0.8641,
+      "step": 11000
+    },
+    {
+      "epoch": 13.611626468769327,
+      "grad_norm": 0.451171875,
+      "learning_rate": 5.112035203894827e-06,
+      "loss": 0.846,
+      "step": 11005
+    },
+    {
+      "epoch": 13.617810760667904,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.066680435123106e-06,
+      "loss": 0.8578,
+      "step": 11010
+    },
+    {
+      "epoch": 13.623995052566482,
+      "grad_norm": 0.447265625,
+      "learning_rate": 5.021522531159428e-06,
+      "loss": 0.8662,
+      "step": 11015
+    },
+    {
+      "epoch": 13.63017934446506,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.976561585648509e-06,
+      "loss": 0.8479,
+      "step": 11020
+    },
+    {
+      "epoch": 13.636363636363637,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.931797691826601e-06,
+      "loss": 0.8492,
+      "step": 11025
+    },
+    {
+      "epoch": 13.642547928262214,
+      "grad_norm": 0.423828125,
+      "learning_rate": 4.887230942521337e-06,
+      "loss": 0.8523,
+      "step": 11030
+    },
+    {
+      "epoch": 13.648732220160792,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.842861430151557e-06,
+      "loss": 0.8563,
+      "step": 11035
+    },
+    {
+      "epoch": 13.65491651205937,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.798689246727006e-06,
+      "loss": 0.8535,
+      "step": 11040
+    },
+    {
+      "epoch": 13.661100803957947,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.754714483848333e-06,
+      "loss": 0.8502,
+      "step": 11045
+    },
+    {
+      "epoch": 13.667285095856524,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.710937232706691e-06,
+      "loss": 0.8562,
+      "step": 11050
+    },
+    {
+      "epoch": 13.673469387755102,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.667357584083721e-06,
+      "loss": 0.8523,
+      "step": 11055
+    },
+    {
+      "epoch": 13.67965367965368,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.623975628351273e-06,
+      "loss": 0.8623,
+      "step": 11060
+    },
+    {
+      "epoch": 13.685837971552257,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.5807914554712005e-06,
+      "loss": 0.8535,
+      "step": 11065
+    },
+    {
+      "epoch": 13.692022263450834,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.537805154995278e-06,
+      "loss": 0.8433,
+      "step": 11070
+    },
+    {
+      "epoch": 13.698206555349412,
+      "grad_norm": 0.419921875,
+      "learning_rate": 4.49501681606489e-06,
+      "loss": 0.8533,
+      "step": 11075
+    },
+    {
+      "epoch": 13.70439084724799,
+      "grad_norm": 0.419921875,
+      "learning_rate": 4.452426527410947e-06,
+      "loss": 0.851,
+      "step": 11080
+    },
+    {
+      "epoch": 13.710575139146568,
+      "grad_norm": 0.423828125,
+      "learning_rate": 4.4100343773536225e-06,
+      "loss": 0.8634,
+      "step": 11085
+    },
+    {
+      "epoch": 13.716759431045146,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.36784045380223e-06,
+      "loss": 0.8549,
+      "step": 11090
+    },
+    {
+      "epoch": 13.722943722943723,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.325844844255023e-06,
+      "loss": 0.8541,
+      "step": 11095
+    },
+    {
+      "epoch": 13.729128014842301,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.2840476357989825e-06,
+      "loss": 0.8611,
+      "step": 11100
+    },
+    {
+      "epoch": 13.735312306740878,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.242448915109698e-06,
+      "loss": 0.8532,
+      "step": 11105
+    },
+    {
+      "epoch": 13.741496598639456,
+      "grad_norm": 0.423828125,
+      "learning_rate": 4.20104876845111e-06,
+      "loss": 0.8538,
+      "step": 11110
+    },
+    {
+      "epoch": 13.747680890538033,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.159847281675411e-06,
+      "loss": 0.8525,
+      "step": 11115
+    },
+    {
+      "epoch": 13.753865182436611,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.118844540222788e-06,
+      "loss": 0.8455,
+      "step": 11120
+    },
+    {
+      "epoch": 13.760049474335188,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.078040629121327e-06,
+      "loss": 0.8592,
+      "step": 11125
+    },
+    {
+      "epoch": 13.766233766233766,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.037435632986786e-06,
+      "loss": 0.86,
+      "step": 11130
+    },
+    {
+      "epoch": 13.772418058132343,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.997029636022387e-06,
+      "loss": 0.8488,
+      "step": 11135
+    },
+    {
+      "epoch": 13.778602350030921,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.95682272201876e-06,
+      "loss": 0.8485,
+      "step": 11140
+    },
+    {
+      "epoch": 13.784786641929498,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.916814974353633e-06,
+      "loss": 0.8525,
+      "step": 11145
+    },
+    {
+      "epoch": 13.790970933828076,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.877006475991729e-06,
+      "loss": 0.8634,
+      "step": 11150
+    },
+    {
+      "epoch": 13.797155225726655,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.837397309484636e-06,
+      "loss": 0.8615,
+      "step": 11155
+    },
+    {
+      "epoch": 13.803339517625233,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.797987556970495e-06,
+      "loss": 0.8533,
+      "step": 11160
+    },
+    {
+      "epoch": 13.80952380952381,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.75877730017401e-06,
+      "loss": 0.8519,
+      "step": 11165
+    },
+    {
+      "epoch": 13.815708101422388,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.7197666204060955e-06,
+      "loss": 0.8673,
+      "step": 11170
+    },
+    {
+      "epoch": 13.821892393320965,
+      "grad_norm": 0.4140625,
+      "learning_rate": 3.6809555985639068e-06,
+      "loss": 0.8561,
+      "step": 11175
+    },
+    {
+      "epoch": 13.828076685219543,
+      "grad_norm": 0.447265625,
+      "learning_rate": 3.6423443151304526e-06,
+      "loss": 0.8519,
+      "step": 11180
+    },
+    {
+      "epoch": 13.83426097711812,
+      "grad_norm": 0.4375,
+      "learning_rate": 3.6039328501746293e-06,
+      "loss": 0.8641,
+      "step": 11185
+    },
+    {
+      "epoch": 13.840445269016698,
+      "grad_norm": 0.41796875,
+      "learning_rate": 3.565721283350931e-06,
+      "loss": 0.8552,
+      "step": 11190
+    },
+    {
+      "epoch": 13.846629560915275,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.527709693899306e-06,
+      "loss": 0.8747,
+      "step": 11195
+    },
+    {
+      "epoch": 13.852813852813853,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.4898981606450333e-06,
+      "loss": 0.8507,
+      "step": 11200
+    },
+    {
+      "epoch": 13.85899814471243,
+      "grad_norm": 0.44921875,
+      "learning_rate": 3.452286761998491e-06,
+      "loss": 0.8636,
+      "step": 11205
+    },
+    {
+      "epoch": 13.865182436611008,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.414875575955101e-06,
+      "loss": 0.8498,
+      "step": 11210
+    },
+    {
+      "epoch": 13.871366728509585,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.3776646800950605e-06,
+      "loss": 0.8534,
+      "step": 11215
+    },
+    {
+      "epoch": 13.877551020408163,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.3406541515832003e-06,
+      "loss": 0.8435,
+      "step": 11220
+    },
+    {
+      "epoch": 13.88373531230674,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.303844067168904e-06,
+      "loss": 0.8524,
+      "step": 11225
+    },
+    {
+      "epoch": 13.889919604205318,
+      "grad_norm": 0.4375,
+      "learning_rate": 3.267234503185823e-06,
+      "loss": 0.8505,
+      "step": 11230
+    },
+    {
+      "epoch": 13.896103896103895,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.2308255355518403e-06,
+      "loss": 0.8557,
+      "step": 11235
+    },
+    {
+      "epoch": 13.902288188002474,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.1946172397688267e-06,
+      "loss": 0.8562,
+      "step": 11240
+    },
+    {
+      "epoch": 13.908472479901052,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.158609690922554e-06,
+      "loss": 0.8555,
+      "step": 11245
+    },
+    {
+      "epoch": 13.91465677179963,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.1228029636824475e-06,
+      "loss": 0.8618,
+      "step": 11250
+    },
+    {
+      "epoch": 13.920841063698207,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.0871971323015336e-06,
+      "loss": 0.8604,
+      "step": 11255
+    },
+    {
+      "epoch": 13.927025355596784,
+      "grad_norm": 0.41015625,
+      "learning_rate": 3.051792270616216e-06,
+      "loss": 0.8598,
+      "step": 11260
+    },
+    {
+      "epoch": 13.933209647495362,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.0165884520461316e-06,
+      "loss": 0.8531,
+      "step": 11265
+    },
+    {
+      "epoch": 13.93939393939394,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.981585749594051e-06,
+      "loss": 0.8508,
+      "step": 11270
+    },
+    {
+      "epoch": 13.945578231292517,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.9467842358456345e-06,
+      "loss": 0.8466,
+      "step": 11275
+    },
+    {
+      "epoch": 13.951762523191094,
+      "grad_norm": 0.44140625,
+      "learning_rate": 2.912183982969385e-06,
+      "loss": 0.8548,
+      "step": 11280
+    },
+    {
+      "epoch": 13.957946815089672,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.8777850627164205e-06,
+      "loss": 0.8616,
+      "step": 11285
+    },
+    {
+      "epoch": 13.96413110698825,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.8435875464203343e-06,
+      "loss": 0.8578,
+      "step": 11290
+    },
+    {
+      "epoch": 13.970315398886827,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.809591504997111e-06,
+      "loss": 0.8674,
+      "step": 11295
+    },
+    {
+      "epoch": 13.976499690785404,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.7757970089449024e-06,
+      "loss": 0.8573,
+      "step": 11300
+    },
+    {
+      "epoch": 13.982683982683982,
+      "grad_norm": 0.44921875,
+      "learning_rate": 2.742204128343917e-06,
+      "loss": 0.8582,
+      "step": 11305
+    },
+    {
+      "epoch": 13.988868274582561,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.708812932856253e-06,
+      "loss": 0.8554,
+      "step": 11310
+    },
+    {
+      "epoch": 13.995052566481139,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.6756234917258205e-06,
+      "loss": 0.8586,
+      "step": 11315
+    },
+    {
+      "epoch": 14.0,
+      "eval_loss": 2.5210444927215576,
+      "eval_runtime": 0.5394,
+      "eval_samples_per_second": 18.54,
+      "eval_steps_per_second": 1.854,
+      "step": 11319
+    },
+    {
+      "epoch": 14.001236858379716,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.6426358737781098e-06,
+      "loss": 0.8497,
+      "step": 11320
+    },
+    {
+      "epoch": 14.007421150278294,
+      "grad_norm": 0.44921875,
+      "learning_rate": 2.6098501474200787e-06,
+      "loss": 0.8456,
+      "step": 11325
+    },
+    {
+      "epoch": 14.013605442176871,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.577266380640053e-06,
+      "loss": 0.8513,
+      "step": 11330
+    },
+    {
+      "epoch": 14.019789734075449,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.5448846410075166e-06,
+      "loss": 0.8512,
+      "step": 11335
+    },
+    {
+      "epoch": 14.025974025974026,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.5127049956730207e-06,
+      "loss": 0.8564,
+      "step": 11340
+    },
+    {
+      "epoch": 14.032158317872604,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.480727511368064e-06,
+      "loss": 0.8522,
+      "step": 11345
+    },
+    {
+      "epoch": 14.038342609771181,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.448952254404846e-06,
+      "loss": 0.8645,
+      "step": 11350
+    },
+    {
+      "epoch": 14.044526901669759,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.4173792906762804e-06,
+      "loss": 0.8509,
+      "step": 11355
+    },
+    {
+      "epoch": 14.050711193568336,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.3860086856557383e-06,
+      "loss": 0.857,
+      "step": 11360
+    },
+    {
+      "epoch": 14.056895485466914,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.35484050439696e-06,
+      "loss": 0.8457,
+      "step": 11365
+    },
+    {
+      "epoch": 14.063079777365491,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.3238748115339324e-06,
+      "loss": 0.8502,
+      "step": 11370
+    },
+    {
+      "epoch": 14.069264069264069,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.293111671280712e-06,
+      "loss": 0.8608,
+      "step": 11375
+    },
+    {
+      "epoch": 14.075448361162646,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.2625511474313685e-06,
+      "loss": 0.851,
+      "step": 11380
+    },
+    {
+      "epoch": 14.081632653061224,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.232193303359742e-06,
+      "loss": 0.8599,
+      "step": 11385
+    },
+    {
+      "epoch": 14.087816944959803,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.2020382020194074e-06,
+      "loss": 0.8584,
+      "step": 11390
+    },
+    {
+      "epoch": 14.09400123685838,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.1720859059434993e-06,
+      "loss": 0.8594,
+      "step": 11395
+    },
+    {
+      "epoch": 14.100185528756958,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.1423364772445887e-06,
+      "loss": 0.8604,
+      "step": 11400
+    },
+    {
+      "epoch": 14.106369820655535,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.112789977614582e-06,
+      "loss": 0.8407,
+      "step": 11405
+    },
+    {
+      "epoch": 14.112554112554113,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.0834464683245346e-06,
+      "loss": 0.8531,
+      "step": 11410
+    },
+    {
+      "epoch": 14.11873840445269,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.0543060102245717e-06,
+      "loss": 0.8524,
+      "step": 11415
+    },
+    {
+      "epoch": 14.124922696351268,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.025368663743743e-06,
+      "loss": 0.8522,
+      "step": 11420
+    },
+    {
+      "epoch": 14.131106988249845,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.9966344888899147e-06,
+      "loss": 0.8648,
+      "step": 11425
+    },
+    {
+      "epoch": 14.137291280148423,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.968103545249611e-06,
+      "loss": 0.8546,
+      "step": 11430
+    },
+    {
+      "epoch": 14.143475572047,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.9397758919879495e-06,
+      "loss": 0.8616,
+      "step": 11435
+    },
+    {
+      "epoch": 14.149659863945578,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.91165158784844e-06,
+      "loss": 0.8539,
+      "step": 11440
+    },
+    {
+      "epoch": 14.155844155844155,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.8837306911529184e-06,
+      "loss": 0.8599,
+      "step": 11445
+    },
+    {
+      "epoch": 14.162028447742733,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.8560132598014368e-06,
+      "loss": 0.8608,
+      "step": 11450
+    },
+    {
+      "epoch": 14.16821273964131,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.8284993512720505e-06,
+      "loss": 0.8558,
+      "step": 11455
+    },
+    {
+      "epoch": 14.174397031539888,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.8011890226208527e-06,
+      "loss": 0.8564,
+      "step": 11460
+    },
+    {
+      "epoch": 14.180581323438465,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.7740823304817188e-06,
+      "loss": 0.8581,
+      "step": 11465
+    },
+    {
+      "epoch": 14.186765615337045,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.7471793310662287e-06,
+      "loss": 0.866,
+      "step": 11470
+    },
+    {
+      "epoch": 14.192949907235622,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.7204800801636e-06,
+      "loss": 0.8628,
+      "step": 11475
+    },
+    {
+      "epoch": 14.1991341991342,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.6939846331405108e-06,
+      "loss": 0.8594,
+      "step": 11480
+    },
+    {
+      "epoch": 14.205318491032777,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.6676930449410099e-06,
+      "loss": 0.855,
+      "step": 11485
+    },
+    {
+      "epoch": 14.211502782931355,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.6416053700863964e-06,
+      "loss": 0.8688,
+      "step": 11490
+    },
+    {
+      "epoch": 14.217687074829932,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.6157216626751292e-06,
+      "loss": 0.8607,
+      "step": 11495
+    },
+    {
+      "epoch": 14.22387136672851,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.5900419763826614e-06,
+      "loss": 0.8577,
+      "step": 11500
+    },
+    {
+      "epoch": 14.230055658627087,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.5645663644614172e-06,
+      "loss": 0.8483,
+      "step": 11505
+    },
+    {
+      "epoch": 14.236239950525665,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.5392948797405827e-06,
+      "loss": 0.8501,
+      "step": 11510
+    },
+    {
+      "epoch": 14.242424242424242,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.5142275746260593e-06,
+      "loss": 0.8419,
+      "step": 11515
+    },
+    {
+      "epoch": 14.24860853432282,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.489364501100332e-06,
+      "loss": 0.8652,
+      "step": 11520
+    },
+    {
+      "epoch": 14.254792826221397,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.4647057107223583e-06,
+      "loss": 0.8626,
+      "step": 11525
+    },
+    {
+      "epoch": 14.260977118119975,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.4402512546275114e-06,
+      "loss": 0.8516,
+      "step": 11530
+    },
+    {
+      "epoch": 14.267161410018552,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.4160011835273934e-06,
+      "loss": 0.8496,
+      "step": 11535
+    },
+    {
+      "epoch": 14.27334570191713,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.3919555477097668e-06,
+      "loss": 0.8558,
+      "step": 11540
+    },
+    {
+      "epoch": 14.279529993815707,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.3681143970385003e-06,
+      "loss": 0.8568,
+      "step": 11545
+    },
+    {
+      "epoch": 14.285714285714286,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.344477780953346e-06,
+      "loss": 0.8614,
+      "step": 11550
+    },
+    {
+      "epoch": 14.291898577612864,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.3210457484699733e-06,
+      "loss": 0.8518,
+      "step": 11555
+    },
+    {
+      "epoch": 14.298082869511441,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.2978183481797801e-06,
+      "loss": 0.8459,
+      "step": 11560
+    },
+    {
+      "epoch": 14.304267161410019,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.274795628249792e-06,
+      "loss": 0.8547,
+      "step": 11565
+    },
+    {
+      "epoch": 14.310451453308596,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.251977636422641e-06,
+      "loss": 0.8592,
+      "step": 11570
+    },
+    {
+      "epoch": 14.316635745207174,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.2293644200163544e-06,
+      "loss": 0.8447,
+      "step": 11575
+    },
+    {
+      "epoch": 14.322820037105751,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.2069560259243328e-06,
+      "loss": 0.8548,
+      "step": 11580
+    },
+    {
+      "epoch": 14.329004329004329,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.1847525006152493e-06,
+      "loss": 0.8512,
+      "step": 11585
+    },
+    {
+      "epoch": 14.335188620902906,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.1627538901329172e-06,
+      "loss": 0.856,
+      "step": 11590
+    },
+    {
+      "epoch": 14.341372912801484,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.1409602400962227e-06,
+      "loss": 0.8518,
+      "step": 11595
+    },
+    {
+      "epoch": 14.347557204700061,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.1193715956990258e-06,
+      "loss": 0.8417,
+      "step": 11600
+    },
+    {
+      "epoch": 14.353741496598639,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.0979880017100596e-06,
+      "loss": 0.861,
+      "step": 11605
+    },
+    {
+      "epoch": 14.359925788497216,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.076809502472831e-06,
+      "loss": 0.8562,
+      "step": 11610
+    },
+    {
+      "epoch": 14.366110080395794,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.055836141905553e-06,
+      "loss": 0.8512,
+      "step": 11615
+    },
+    {
+      "epoch": 14.372294372294371,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.035067963501024e-06,
+      "loss": 0.873,
+      "step": 11620
+    },
+    {
+      "epoch": 14.37847866419295,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.014505010326583e-06,
+      "loss": 0.8564,
+      "step": 11625
+    },
+    {
+      "epoch": 14.384662956091528,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.94147325023953e-07,
+      "loss": 0.8537,
+      "step": 11630
+    },
+    {
+      "epoch": 14.390847247990106,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.739949498091982e-07,
+      "loss": 0.8618,
+      "step": 11635
+    },
+    {
+      "epoch": 14.397031539888683,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.540479264726676e-07,
+      "loss": 0.8567,
+      "step": 11640
+    },
+    {
+      "epoch": 14.40321583178726,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.343062963787952e-07,
+      "loss": 0.8583,
+      "step": 11645
+    },
+    {
+      "epoch": 14.409400123685838,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.147701004661446e-07,
+      "loss": 0.8569,
+      "step": 11650
+    },
+    {
+      "epoch": 14.415584415584416,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.954393792472649e-07,
+      "loss": 0.8431,
+      "step": 11655
+    },
+    {
+      "epoch": 14.421768707482993,
+      "grad_norm": 0.419921875,
+      "learning_rate": 8.763141728085789e-07,
+      "loss": 0.856,
+      "step": 11660
+    },
+    {
+      "epoch": 14.42795299938157,
+      "grad_norm": 0.41796875,
+      "learning_rate": 8.573945208103618e-07,
+      "loss": 0.8554,
+      "step": 11665
+    },
+    {
+      "epoch": 14.434137291280148,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.386804624865851e-07,
+      "loss": 0.8636,
+      "step": 11670
+    },
+    {
+      "epoch": 14.440321583178726,
+      "grad_norm": 0.42578125,
+      "learning_rate": 8.201720366449283e-07,
+      "loss": 0.8605,
+      "step": 11675
+    },
+    {
+      "epoch": 14.446505875077303,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.018692816666118e-07,
+      "loss": 0.8529,
+      "step": 11680
+    },
+    {
+      "epoch": 14.45269016697588,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.837722355063637e-07,
+      "loss": 0.8594,
+      "step": 11685
+    },
+    {
+      "epoch": 14.458874458874458,
+      "grad_norm": 0.435546875,
+      "learning_rate": 7.658809356923424e-07,
+      "loss": 0.8562,
+      "step": 11690
+    },
+    {
+      "epoch": 14.465058750773036,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.481954193260143e-07,
+      "loss": 0.8563,
+      "step": 11695
+    },
+    {
+      "epoch": 14.471243042671613,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.307157230821426e-07,
+      "loss": 0.8515,
+      "step": 11700
+    },
+    {
+      "epoch": 14.477427334570192,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.134418832086653e-07,
+      "loss": 0.858,
+      "step": 11705
+    },
+    {
+      "epoch": 14.48361162646877,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.963739355266286e-07,
+      "loss": 0.8608,
+      "step": 11710
+    },
+    {
+      "epoch": 14.489795918367347,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.7951191543012e-07,
+      "loss": 0.8662,
+      "step": 11715
+    },
+    {
+      "epoch": 14.495980210265925,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.628558578862021e-07,
+      "loss": 0.86,
+      "step": 11720
+    },
+    {
+      "epoch": 14.502164502164502,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.464057974348014e-07,
+      "loss": 0.8587,
+      "step": 11725
+    },
+    {
+      "epoch": 14.50834879406308,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.301617681886863e-07,
+      "loss": 0.859,
+      "step": 11730
+    },
+    {
+      "epoch": 14.514533085961657,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.141238038333885e-07,
+      "loss": 0.8607,
+      "step": 11735
+    },
+    {
+      "epoch": 14.520717377860235,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.982919376270823e-07,
+      "loss": 0.865,
+      "step": 11740
+    },
+    {
+      "epoch": 14.526901669758812,
+      "grad_norm": 0.4296875,
+      "learning_rate": 5.826662024005835e-07,
+      "loss": 0.864,
+      "step": 11745
+    },
+    {
+      "epoch": 14.53308596165739,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.672466305572388e-07,
+      "loss": 0.8474,
+      "step": 11750
+    },
+    {
+      "epoch": 14.539270253555967,
+      "grad_norm": 0.421875,
+      "learning_rate": 5.52033254072859e-07,
+      "loss": 0.8571,
+      "step": 11755
+    },
+    {
+      "epoch": 14.545454545454545,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.370261044956971e-07,
+      "loss": 0.852,
+      "step": 11760
+    },
+    {
+      "epoch": 14.551638837353122,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.222252129463146e-07,
+      "loss": 0.8602,
+      "step": 11765
+    },
+    {
+      "epoch": 14.5578231292517,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.07630610117582e-07,
+      "loss": 0.8637,
+      "step": 11770
+    },
+    {
+      "epoch": 14.564007421150277,
+      "grad_norm": 0.419921875,
+      "learning_rate": 4.932423262745456e-07,
+      "loss": 0.8446,
+      "step": 11775
+    },
+    {
+      "epoch": 14.570191713048857,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.790603912544489e-07,
+      "loss": 0.8551,
+      "step": 11780
+    },
+    {
+      "epoch": 14.576376004947434,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.6508483446661144e-07,
+      "loss": 0.8601,
+      "step": 11785
+    },
+    {
+      "epoch": 14.582560296846012,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.5131568489236166e-07,
+      "loss": 0.862,
+      "step": 11790
+    },
+    {
+      "epoch": 14.58874458874459,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.377529710850259e-07,
+      "loss": 0.8592,
+      "step": 11795
+    },
+    {
+      "epoch": 14.594928880643167,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.2439672116982855e-07,
+      "loss": 0.8491,
+      "step": 11800
+    },
+    {
+      "epoch": 14.601113172541744,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.112469628438365e-07,
+      "loss": 0.8512,
+      "step": 11805
+    },
+    {
+      "epoch": 14.607297464440322,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.983037233759368e-07,
+      "loss": 0.8562,
+      "step": 11810
+    },
+    {
+      "epoch": 14.6134817563389,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.8556702960673706e-07,
+      "loss": 0.8498,
+      "step": 11815
+    },
+    {
+      "epoch": 14.619666048237477,
+      "grad_norm": 0.400390625,
+      "learning_rate": 3.73036907948543e-07,
+      "loss": 0.8548,
+      "step": 11820
+    },
+    {
+      "epoch": 14.625850340136054,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.6071338438524726e-07,
+      "loss": 0.8642,
+      "step": 11825
+    },
+    {
+      "epoch": 14.632034632034632,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.485964844723744e-07,
+      "loss": 0.8464,
+      "step": 11830
+    },
+    {
+      "epoch": 14.63821892393321,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.366862333369358e-07,
+      "loss": 0.8568,
+      "step": 11835
+    },
+    {
+      "epoch": 14.644403215831787,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.2498265567739717e-07,
+      "loss": 0.8662,
+      "step": 11840
+    },
+    {
+      "epoch": 14.650587507730364,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.134857757636889e-07,
+      "loss": 0.8558,
+      "step": 11845
+    },
+    {
+      "epoch": 14.656771799628942,
+      "grad_norm": 0.404296875,
+      "learning_rate": 3.0219561743707326e-07,
+      "loss": 0.8455,
+      "step": 11850
+    },
+    {
+      "epoch": 14.66295609152752,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.9111220411014437e-07,
+      "loss": 0.8558,
+      "step": 11855
+    },
+    {
+      "epoch": 14.669140383426098,
+      "grad_norm": 0.4609375,
+      "learning_rate": 2.8023555876673937e-07,
+      "loss": 0.85,
+      "step": 11860
+    },
+    {
+      "epoch": 14.675324675324676,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.6956570396197143e-07,
+      "loss": 0.848,
+      "step": 11865
+    },
+    {
+      "epoch": 14.681508967223253,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.5910266182207486e-07,
+      "loss": 0.8651,
+      "step": 11870
+    },
+    {
+      "epoch": 14.687693259121831,
+      "grad_norm": 0.4140625,
+      "learning_rate": 2.4884645404443795e-07,
+      "loss": 0.8467,
+      "step": 11875
+    },
+    {
+      "epoch": 14.693877551020408,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.3879710189753656e-07,
+      "loss": 0.8597,
+      "step": 11880
+    },
+    {
+      "epoch": 14.700061842918986,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.289546262208786e-07,
+      "loss": 0.8578,
+      "step": 11885
+    },
+    {
+      "epoch": 14.706246134817563,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.1931904742495957e-07,
+      "loss": 0.8528,
+      "step": 11890
+    },
+    {
+      "epoch": 14.712430426716141,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.098903854912515e-07,
+      "loss": 0.8549,
+      "step": 11895
+    },
+    {
+      "epoch": 14.718614718614718,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.0066865997212525e-07,
+      "loss": 0.8537,
+      "step": 11900
+    },
+    {
+      "epoch": 14.724799010513296,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.9165388999082822e-07,
+      "loss": 0.8617,
+      "step": 11905
+    },
+    {
+      "epoch": 14.730983302411873,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.8284609424142895e-07,
+      "loss": 0.858,
+      "step": 11910
+    },
+    {
+      "epoch": 14.737167594310451,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.7424529098881703e-07,
+      "loss": 0.8557,
+      "step": 11915
+    },
+    {
+      "epoch": 14.743351886209028,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.6585149806860324e-07,
+      "loss": 0.8588,
+      "step": 11920
+    },
+    {
+      "epoch": 14.749536178107606,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.5766473288715278e-07,
+      "loss": 0.8595,
+      "step": 11925
+    },
+    {
+      "epoch": 14.755720470006183,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.4968501242148547e-07,
+      "loss": 0.8498,
+      "step": 11930
+    },
+    {
+      "epoch": 14.761904761904763,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.4191235321928676e-07,
+      "loss": 0.8676,
+      "step": 11935
+    },
+    {
+      "epoch": 14.76808905380334,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.3434677139885222e-07,
+      "loss": 0.8537,
+      "step": 11940
+    },
+    {
+      "epoch": 14.774273345701918,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.2698828264904317e-07,
+      "loss": 0.8622,
+      "step": 11945
+    },
+    {
+      "epoch": 14.780457637600495,
+      "grad_norm": 0.4609375,
+      "learning_rate": 1.1983690222929778e-07,
+      "loss": 0.8581,
+      "step": 11950
+    },
+    {
+      "epoch": 14.786641929499073,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.1289264496953111e-07,
+      "loss": 0.8554,
+      "step": 11955
+    },
+    {
+      "epoch": 14.79282622139765,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0615552527017958e-07,
+      "loss": 0.8502,
+      "step": 11960
+    },
+    {
+      "epoch": 14.799010513296228,
+      "grad_norm": 0.419921875,
+      "learning_rate": 9.962555710212318e-08,
+      "loss": 0.8599,
+      "step": 11965
+    },
+    {
+      "epoch": 14.805194805194805,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.330275400666332e-08,
+      "loss": 0.8524,
+      "step": 11970
+    },
+    {
+      "epoch": 14.811379097093383,
+      "grad_norm": 0.42578125,
+      "learning_rate": 8.718712909548953e-08,
+      "loss": 0.8665,
+      "step": 11975
+    },
+    {
+      "epoch": 14.81756338899196,
+      "grad_norm": 0.42578125,
+      "learning_rate": 8.127869505069053e-08,
+      "loss": 0.8573,
+      "step": 11980
+    },
+    {
+      "epoch": 14.823747680890538,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.557746412468758e-08,
+      "loss": 0.8544,
+      "step": 11985
+    },
+    {
+      "epoch": 14.829931972789115,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.00834481402013e-08,
+      "loss": 0.8494,
+      "step": 11990
+    },
+    {
+      "epoch": 14.836116264687693,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.479665849027372e-08,
+      "loss": 0.8539,
+      "step": 11995
+    },
+    {
+      "epoch": 14.84230055658627,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.971710613821291e-08,
+      "loss": 0.8534,
+      "step": 12000
+    },
+    {
+      "epoch": 14.848484848484848,
+      "grad_norm": 0.423828125,
+      "learning_rate": 5.484480161755956e-08,
+      "loss": 0.8428,
+      "step": 12005
+    },
+    {
+      "epoch": 14.854669140383425,
+      "grad_norm": 0.455078125,
+      "learning_rate": 5.0179755032109253e-08,
+      "loss": 0.8615,
+      "step": 12010
+    },
+    {
+      "epoch": 14.860853432282005,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.572197605583473e-08,
+      "loss": 0.8527,
+      "step": 12015
+    },
+    {
+      "epoch": 14.867037724180582,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.147147393290807e-08,
+      "loss": 0.861,
+      "step": 12020
+    },
+    {
+      "epoch": 14.87322201607916,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.742825747766743e-08,
+      "loss": 0.8517,
+      "step": 12025
+    },
+    {
+      "epoch": 14.879406307977737,
+      "grad_norm": 0.408203125,
+      "learning_rate": 3.359233507459481e-08,
+      "loss": 0.8382,
+      "step": 12030
+    },
+    {
+      "epoch": 14.885590599876314,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.9963714678316045e-08,
+      "loss": 0.8497,
+      "step": 12035
+    },
+    {
+      "epoch": 14.891774891774892,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.6542403813545334e-08,
+      "loss": 0.8573,
+      "step": 12040
+    },
+    {
+      "epoch": 14.89795918367347,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.3328409575129608e-08,
+      "loss": 0.8541,
+      "step": 12045
+    },
+    {
+      "epoch": 14.904143475572047,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.0321738627981923e-08,
+      "loss": 0.856,
+      "step": 12050
+    },
+    {
+      "epoch": 14.910327767470624,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.7522397207070383e-08,
+      "loss": 0.8598,
+      "step": 12055
+    },
+    {
+      "epoch": 14.916512059369202,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.4930391117451426e-08,
+      "loss": 0.8526,
+      "step": 12060
+    },
+    {
+      "epoch": 14.92269635126778,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.2545725734192103e-08,
+      "loss": 0.8637,
+      "step": 12065
+    },
+    {
+      "epoch": 14.928880643166357,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0368406002436715e-08,
+      "loss": 0.8585,
+      "step": 12070
+    },
+    {
+      "epoch": 14.935064935064934,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.398436437317969e-09,
+      "loss": 0.8521,
+      "step": 12075
+    },
+    {
+      "epoch": 14.941249226963512,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.635821124001406e-09,
+      "loss": 0.8555,
+      "step": 12080
+    },
+    {
+      "epoch": 14.94743351886209,
+      "grad_norm": 0.423828125,
+      "learning_rate": 5.080563717629882e-09,
+      "loss": 0.8507,
+      "step": 12085
+    },
+    {
+      "epoch": 14.953617810760669,
+      "grad_norm": 0.431640625,
+      "learning_rate": 3.732667443390181e-09,
+      "loss": 0.8526,
+      "step": 12090
+    },
+    {
+      "epoch": 14.959802102659246,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.5921350964352997e-09,
+      "loss": 0.8538,
+      "step": 12095
+    },
+    {
+      "epoch": 14.965986394557824,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.6589690418955528e-09,
+      "loss": 0.8587,
+      "step": 12100
+    },
+    {
+      "epoch": 14.972170686456401,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.33171214889672e-10,
+      "loss": 0.8486,
+      "step": 12105
+    },
+    {
+      "epoch": 14.978354978354979,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.147431205359098e-10,
+      "loss": 0.8622,
+      "step": 12110
+    },
+    {
+      "epoch": 14.984539270253556,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0368583388542519e-10,
+      "loss": 0.8548,
+      "step": 12115
+    },
+    {
+      "epoch": 14.990723562152134,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0,
+      "loss": 0.8485,
+      "step": 12120
+    },
+    {
+      "epoch": 14.990723562152134,
+      "eval_loss": 2.520606517791748,
+      "eval_runtime": 0.5387,
+      "eval_samples_per_second": 18.562,
+      "eval_steps_per_second": 1.856,
+      "step": 12120
+    },
+    {
+      "epoch": 14.990723562152134,
+      "step": 12120,
+      "total_flos": 7.118964864348848e+18,
+      "train_loss": 0.9255497357436139,
+      "train_runtime": 41659.6506,
+      "train_samples_per_second": 13.97,
+      "train_steps_per_second": 0.291
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 12120,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 15,
+  "save_steps": 100,
+  "total_flos": 7.118964864348848e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}