diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,3666 @@
+{
+  "best_global_step": 4000,
+  "best_metric": 1.430210828781128,
+  "best_model_checkpoint": "output/Tiger-generation/2025-07-18-18.00.46/checkpoint-4000",
+  "epoch": 0.8167432363450741,
+  "eval_steps": 1000,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002041858090862685,
+      "grad_norm": 40.58543014526367,
+      "learning_rate": 1.2244897959183673e-07,
+      "loss": 2.4785,
+      "mean_token_accuracy": 0.525111411511898,
+      "num_tokens": 69505.0,
+      "step": 10
+    },
+    {
+      "epoch": 0.00408371618172537,
+      "grad_norm": 42.50286102294922,
+      "learning_rate": 2.5850340136054423e-07,
+      "loss": 2.4665,
+      "mean_token_accuracy": 0.5225918605923653,
+      "num_tokens": 139531.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.006125574272588055,
+      "grad_norm": 32.75865936279297,
+      "learning_rate": 3.945578231292517e-07,
+      "loss": 2.4605,
+      "mean_token_accuracy": 0.5167869120836258,
+      "num_tokens": 210870.0,
+      "step": 30
+    },
+    {
+      "epoch": 0.00816743236345074,
+      "grad_norm": 37.280799865722656,
+      "learning_rate": 5.306122448979592e-07,
+      "loss": 2.481,
+      "mean_token_accuracy": 0.5240069389343261,
+      "num_tokens": 281885.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.010209290454313425,
+      "grad_norm": 35.95231628417969,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 2.4516,
+      "mean_token_accuracy": 0.5276862233877182,
+      "num_tokens": 351322.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.01225114854517611,
+      "grad_norm": 35.390323638916016,
+      "learning_rate": 8.027210884353742e-07,
+      "loss": 2.4629,
+      "mean_token_accuracy": 0.5194140955805778,
+      "num_tokens": 423235.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.014293006636038795,
+      "grad_norm": 32.571537017822266,
+      "learning_rate": 9.387755102040817e-07,
+      "loss": 2.326,
+      "mean_token_accuracy": 0.5447167232632637,
+      "num_tokens": 494426.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.01633486472690148,
+      "grad_norm": 51.0481071472168,
+      "learning_rate": 1.0748299319727893e-06,
+      "loss": 2.3607,
+      "mean_token_accuracy": 0.5329553991556167,
+      "num_tokens": 565410.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.018376722817764167,
+      "grad_norm": 40.323486328125,
+      "learning_rate": 1.2108843537414967e-06,
+      "loss": 2.32,
+      "mean_token_accuracy": 0.5369406580924988,
+      "num_tokens": 637547.0,
+      "step": 90
+    },
+    {
+      "epoch": 0.02041858090862685,
+      "grad_norm": 41.64793014526367,
+      "learning_rate": 1.3469387755102043e-06,
+      "loss": 2.2872,
+      "mean_token_accuracy": 0.5405925557017326,
+      "num_tokens": 709374.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.022460438999489536,
+      "grad_norm": 48.55975341796875,
+      "learning_rate": 1.4829931972789117e-06,
+      "loss": 2.2847,
+      "mean_token_accuracy": 0.5352288991212845,
+      "num_tokens": 779218.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.02450229709035222,
+      "grad_norm": 38.39033889770508,
+      "learning_rate": 1.6190476190476193e-06,
+      "loss": 2.2336,
+      "mean_token_accuracy": 0.5404456198215485,
+      "num_tokens": 846940.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.026544155181214904,
+      "grad_norm": 23.284969329833984,
+      "learning_rate": 1.7551020408163267e-06,
+      "loss": 2.1848,
+      "mean_token_accuracy": 0.5559335589408875,
+      "num_tokens": 921181.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.02858601327207759,
+      "grad_norm": 31.96380615234375,
+      "learning_rate": 1.8911564625850343e-06,
+      "loss": 2.168,
+      "mean_token_accuracy": 0.5423629283905029,
+      "num_tokens": 990533.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.030627871362940276,
+      "grad_norm": 38.387664794921875,
+      "learning_rate": 2.0272108843537416e-06,
+      "loss": 2.0985,
+      "mean_token_accuracy": 0.5486705094575882,
+      "num_tokens": 1060250.0,
+      "step": 150
+    },
+    {
+      "epoch": 0.03266972945380296,
+      "grad_norm": 30.54590606689453,
+      "learning_rate": 2.1632653061224495e-06,
+      "loss": 2.1418,
+      "mean_token_accuracy": 0.5525166988372803,
+      "num_tokens": 1132379.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.03471158754466565,
+      "grad_norm": 32.12227249145508,
+      "learning_rate": 2.299319727891157e-06,
+      "loss": 2.1179,
+      "mean_token_accuracy": 0.5542925208806991,
+      "num_tokens": 1208198.0,
+      "step": 170
+    },
+    {
+      "epoch": 0.036753445635528334,
+      "grad_norm": 38.035926818847656,
+      "learning_rate": 2.4353741496598642e-06,
+      "loss": 2.0651,
+      "mean_token_accuracy": 0.5671319425106048,
+      "num_tokens": 1279054.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.03879530372639101,
+      "grad_norm": 25.86829376220703,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 2.1026,
+      "mean_token_accuracy": 0.5436463728547096,
+      "num_tokens": 1347862.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.0408371618172537,
+      "grad_norm": 40.36919021606445,
+      "learning_rate": 2.707482993197279e-06,
+      "loss": 1.98,
+      "mean_token_accuracy": 0.5548223063349724,
+      "num_tokens": 1417594.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.042879019908116385,
+      "grad_norm": 28.774658203125,
+      "learning_rate": 2.843537414965987e-06,
+      "loss": 1.965,
+      "mean_token_accuracy": 0.559009425342083,
+      "num_tokens": 1489549.0,
+      "step": 210
+    },
+    {
+      "epoch": 0.04492087799897907,
+      "grad_norm": 21.973716735839844,
+      "learning_rate": 2.979591836734694e-06,
+      "loss": 1.9638,
+      "mean_token_accuracy": 0.5593225166201592,
+      "num_tokens": 1555401.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.04696273608984176,
+      "grad_norm": 36.374786376953125,
+      "learning_rate": 3.1156462585034016e-06,
+      "loss": 1.9379,
+      "mean_token_accuracy": 0.5614111497998238,
+      "num_tokens": 1627148.0,
+      "step": 230
+    },
+    {
+      "epoch": 0.04900459418070444,
+      "grad_norm": 30.232772827148438,
+      "learning_rate": 3.251700680272109e-06,
+      "loss": 1.9311,
+      "mean_token_accuracy": 0.5598069429397583,
+      "num_tokens": 1696743.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.05104645227156713,
+      "grad_norm": 33.978477478027344,
+      "learning_rate": 3.3877551020408168e-06,
+      "loss": 1.8232,
+      "mean_token_accuracy": 0.5703684508800506,
+      "num_tokens": 1769142.0,
+      "step": 250
+    },
+    {
+      "epoch": 0.05308831036242981,
+      "grad_norm": 22.255043029785156,
+      "learning_rate": 3.523809523809524e-06,
+      "loss": 1.6812,
+      "mean_token_accuracy": 0.5865129962563514,
+      "num_tokens": 1843700.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.055130168453292494,
+      "grad_norm": 24.115663528442383,
+      "learning_rate": 3.6598639455782316e-06,
+      "loss": 1.7255,
+      "mean_token_accuracy": 0.5739775270223617,
+      "num_tokens": 1916733.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.05717202654415518,
+      "grad_norm": 26.15631866455078,
+      "learning_rate": 3.795918367346939e-06,
+      "loss": 1.7444,
+      "mean_token_accuracy": 0.5637126505374909,
+      "num_tokens": 1987487.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.059213884635017866,
+      "grad_norm": 23.212749481201172,
+      "learning_rate": 3.931972789115646e-06,
+      "loss": 1.7016,
+      "mean_token_accuracy": 0.5802351862192154,
+      "num_tokens": 2061962.0,
+      "step": 290
+    },
+    {
+      "epoch": 0.06125574272588055,
+      "grad_norm": 24.3252010345459,
+      "learning_rate": 4.068027210884354e-06,
+      "loss": 1.7064,
+      "mean_token_accuracy": 0.5761498302221298,
+      "num_tokens": 2132628.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.06329760081674324,
+      "grad_norm": 20.47394561767578,
+      "learning_rate": 4.204081632653061e-06,
+      "loss": 1.6628,
+      "mean_token_accuracy": 0.5744958698749543,
+      "num_tokens": 2202325.0,
+      "step": 310
+    },
+    {
+      "epoch": 0.06533945890760592,
+      "grad_norm": 19.049283981323242,
+      "learning_rate": 4.340136054421769e-06,
+      "loss": 1.61,
+      "mean_token_accuracy": 0.598686158657074,
+      "num_tokens": 2272383.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.06738131699846861,
+      "grad_norm": 14.891511917114258,
+      "learning_rate": 4.476190476190477e-06,
+      "loss": 1.6495,
+      "mean_token_accuracy": 0.5836592674255371,
+      "num_tokens": 2343570.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.0694231750893313,
+      "grad_norm": 23.719728469848633,
+      "learning_rate": 4.612244897959184e-06,
+      "loss": 1.5801,
+      "mean_token_accuracy": 0.597952377796173,
+      "num_tokens": 2412952.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.07146503318019398,
+      "grad_norm": 24.924657821655273,
+      "learning_rate": 4.7482993197278915e-06,
+      "loss": 1.5314,
+      "mean_token_accuracy": 0.5992678642272949,
+      "num_tokens": 2486087.0,
+      "step": 350
+    },
+    {
+      "epoch": 0.07350689127105667,
+      "grad_norm": 19.125892639160156,
+      "learning_rate": 4.884353741496599e-06,
+      "loss": 1.5699,
+      "mean_token_accuracy": 0.5989518314599991,
+      "num_tokens": 2560105.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.07554874936191934,
+      "grad_norm": 24.626413345336914,
+      "learning_rate": 5.020408163265307e-06,
+      "loss": 1.5849,
+      "mean_token_accuracy": 0.5961738631129265,
+      "num_tokens": 2635351.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.07759060745278203,
+      "grad_norm": 16.234474182128906,
+      "learning_rate": 5.156462585034014e-06,
+      "loss": 1.6359,
+      "mean_token_accuracy": 0.59544438123703,
+      "num_tokens": 2708644.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.07963246554364471,
+      "grad_norm": 15.393643379211426,
+      "learning_rate": 5.292517006802722e-06,
+      "loss": 1.6003,
+      "mean_token_accuracy": 0.597255888581276,
+      "num_tokens": 2782814.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.0816743236345074,
+      "grad_norm": 18.73783302307129,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 1.6231,
+      "mean_token_accuracy": 0.5859753489494324,
+      "num_tokens": 2850906.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.08371618172537008,
+      "grad_norm": 14.788663864135742,
+      "learning_rate": 5.564625850340136e-06,
+      "loss": 1.5492,
+      "mean_token_accuracy": 0.6039781719446182,
+      "num_tokens": 2923967.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.08575803981623277,
+      "grad_norm": 16.760385513305664,
+      "learning_rate": 5.700680272108844e-06,
+      "loss": 1.5715,
+      "mean_token_accuracy": 0.6139408111572265,
+      "num_tokens": 2992994.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.08779989790709546,
+      "grad_norm": 23.059823989868164,
+      "learning_rate": 5.8367346938775515e-06,
+      "loss": 1.5412,
+      "mean_token_accuracy": 0.601115295290947,
+      "num_tokens": 3063958.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.08984175599795814,
+      "grad_norm": 21.973384857177734,
+      "learning_rate": 5.972789115646259e-06,
+      "loss": 1.5855,
+      "mean_token_accuracy": 0.6108441263437271,
+      "num_tokens": 3132005.0,
+      "step": 440
+    },
+    {
+      "epoch": 0.09188361408882083,
+      "grad_norm": 17.117069244384766,
+      "learning_rate": 6.108843537414967e-06,
+      "loss": 1.576,
+      "mean_token_accuracy": 0.5952845871448517,
+      "num_tokens": 3204170.0,
+      "step": 450
+    },
+    {
+      "epoch": 0.09392547217968351,
+      "grad_norm": 18.47603988647461,
+      "learning_rate": 6.244897959183675e-06,
+      "loss": 1.5333,
+      "mean_token_accuracy": 0.6100009500980377,
+      "num_tokens": 3275122.0,
+      "step": 460
+    },
+    {
+      "epoch": 0.0959673302705462,
+      "grad_norm": 20.058645248413086,
+      "learning_rate": 6.380952380952381e-06,
+      "loss": 1.4875,
+      "mean_token_accuracy": 0.6178710669279098,
+      "num_tokens": 3348601.0,
+      "step": 470
+    },
+    {
+      "epoch": 0.09800918836140889,
+      "grad_norm": 17.164993286132812,
+      "learning_rate": 6.517006802721089e-06,
+      "loss": 1.5212,
+      "mean_token_accuracy": 0.6061257958412171,
+      "num_tokens": 3420761.0,
+      "step": 480
+    },
+    {
+      "epoch": 0.10005104645227157,
+      "grad_norm": 15.247114181518555,
+      "learning_rate": 6.653061224489797e-06,
+      "loss": 1.5402,
+      "mean_token_accuracy": 0.6126458764076232,
+      "num_tokens": 3490967.0,
+      "step": 490
+    },
+    {
+      "epoch": 0.10209290454313426,
+      "grad_norm": 18.676435470581055,
+      "learning_rate": 6.789115646258504e-06,
+      "loss": 1.4995,
+      "mean_token_accuracy": 0.6140502035617829,
+      "num_tokens": 3564295.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.10413476263399694,
+      "grad_norm": 19.05228042602539,
+      "learning_rate": 6.925170068027211e-06,
+      "loss": 1.5967,
+      "mean_token_accuracy": 0.5993693619966507,
+      "num_tokens": 3633837.0,
+      "step": 510
+    },
+    {
+      "epoch": 0.10617662072485962,
+      "grad_norm": 16.225095748901367,
+      "learning_rate": 7.061224489795919e-06,
+      "loss": 1.4627,
+      "mean_token_accuracy": 0.6195647984743118,
+      "num_tokens": 3707429.0,
+      "step": 520
+    },
+    {
+      "epoch": 0.1082184788157223,
+      "grad_norm": 17.119951248168945,
+      "learning_rate": 7.197278911564627e-06,
+      "loss": 1.4916,
+      "mean_token_accuracy": 0.6147974878549576,
+      "num_tokens": 3782215.0,
+      "step": 530
+    },
+    {
+      "epoch": 0.11026033690658499,
+      "grad_norm": 14.728631973266602,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.5828,
+      "mean_token_accuracy": 0.6027207791805267,
+      "num_tokens": 3856106.0,
+      "step": 540
+    },
+    {
+      "epoch": 0.11230219499744767,
+      "grad_norm": 13.182671546936035,
+      "learning_rate": 7.469387755102041e-06,
+      "loss": 1.5749,
+      "mean_token_accuracy": 0.6024670600891113,
+      "num_tokens": 3925174.0,
+      "step": 550
+    },
+    {
+      "epoch": 0.11434405308831036,
+      "grad_norm": 17.364179611206055,
+      "learning_rate": 7.605442176870749e-06,
+      "loss": 1.5689,
+      "mean_token_accuracy": 0.606898409128189,
+      "num_tokens": 3993338.0,
+      "step": 560
+    },
+    {
+      "epoch": 0.11638591117917305,
+      "grad_norm": 16.174150466918945,
+      "learning_rate": 7.741496598639456e-06,
+      "loss": 1.5548,
+      "mean_token_accuracy": 0.6065890669822693,
+      "num_tokens": 4063428.0,
+      "step": 570
+    },
+    {
+      "epoch": 0.11842776927003573,
+      "grad_norm": 17.375446319580078,
+      "learning_rate": 7.877551020408164e-06,
+      "loss": 1.4776,
+      "mean_token_accuracy": 0.614559343457222,
+      "num_tokens": 4135382.0,
+      "step": 580
+    },
+    {
+      "epoch": 0.12046962736089842,
+      "grad_norm": 17.446441650390625,
+      "learning_rate": 8.013605442176871e-06,
+      "loss": 1.5583,
+      "mean_token_accuracy": 0.6063664674758911,
+      "num_tokens": 4211679.0,
+      "step": 590
+    },
+    {
+      "epoch": 0.1225114854517611,
+      "grad_norm": 17.279420852661133,
+      "learning_rate": 8.14965986394558e-06,
+      "loss": 1.4835,
+      "mean_token_accuracy": 0.6287278205156326,
+      "num_tokens": 4277878.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.12455334354262379,
+      "grad_norm": 16.633319854736328,
+      "learning_rate": 8.285714285714287e-06,
+      "loss": 1.5295,
+      "mean_token_accuracy": 0.6134639114141465,
+      "num_tokens": 4348459.0,
+      "step": 610
+    },
+    {
+      "epoch": 0.12659520163348648,
+      "grad_norm": 15.54926872253418,
+      "learning_rate": 8.421768707482994e-06,
+      "loss": 1.5368,
+      "mean_token_accuracy": 0.612190106511116,
+      "num_tokens": 4423381.0,
+      "step": 620
+    },
+    {
+      "epoch": 0.12863705972434916,
+      "grad_norm": 15.600358963012695,
+      "learning_rate": 8.557823129251701e-06,
+      "loss": 1.4858,
+      "mean_token_accuracy": 0.6275644600391388,
+      "num_tokens": 4493430.0,
+      "step": 630
+    },
+    {
+      "epoch": 0.13067891781521185,
+      "grad_norm": 15.166326522827148,
+      "learning_rate": 8.69387755102041e-06,
+      "loss": 1.5038,
+      "mean_token_accuracy": 0.6171506643295288,
+      "num_tokens": 4562553.0,
+      "step": 640
+    },
+    {
+      "epoch": 0.13272077590607453,
+      "grad_norm": 19.53791618347168,
+      "learning_rate": 8.829931972789117e-06,
+      "loss": 1.578,
+      "mean_token_accuracy": 0.6041344046592713,
+      "num_tokens": 4627736.0,
+      "step": 650
+    },
+    {
+      "epoch": 0.13476263399693722,
+      "grad_norm": 15.902642250061035,
+      "learning_rate": 8.965986394557823e-06,
+      "loss": 1.5097,
+      "mean_token_accuracy": 0.6086627811193466,
+      "num_tokens": 4701324.0,
+      "step": 660
+    },
+    {
+      "epoch": 0.1368044920877999,
+      "grad_norm": 16.800201416015625,
+      "learning_rate": 9.102040816326532e-06,
+      "loss": 1.5276,
+      "mean_token_accuracy": 0.6193519622087479,
+      "num_tokens": 4773920.0,
+      "step": 670
+    },
+    {
+      "epoch": 0.1388463501786626,
+      "grad_norm": 15.870624542236328,
+      "learning_rate": 9.238095238095239e-06,
+      "loss": 1.5096,
+      "mean_token_accuracy": 0.6179488390684128,
+      "num_tokens": 4846912.0,
+      "step": 680
+    },
+    {
+      "epoch": 0.14088820826952528,
+      "grad_norm": 15.414668083190918,
+      "learning_rate": 9.374149659863946e-06,
+      "loss": 1.5386,
+      "mean_token_accuracy": 0.6122144073247909,
+      "num_tokens": 4921846.0,
+      "step": 690
+    },
+    {
+      "epoch": 0.14293006636038796,
+      "grad_norm": 14.189248085021973,
+      "learning_rate": 9.510204081632653e-06,
+      "loss": 1.5427,
+      "mean_token_accuracy": 0.5970522820949554,
+      "num_tokens": 5000558.0,
+      "step": 700
+    },
+    {
+      "epoch": 0.14497192445125065,
+      "grad_norm": 12.32139778137207,
+      "learning_rate": 9.646258503401362e-06,
+      "loss": 1.4581,
+      "mean_token_accuracy": 0.6264193892478943,
+      "num_tokens": 5070767.0,
+      "step": 710
+    },
+    {
+      "epoch": 0.14701378254211334,
+      "grad_norm": 15.26685619354248,
+      "learning_rate": 9.782312925170069e-06,
+      "loss": 1.5367,
+      "mean_token_accuracy": 0.6124130487442017,
+      "num_tokens": 5142627.0,
+      "step": 720
+    },
+    {
+      "epoch": 0.14905564063297602,
+      "grad_norm": 15.513026237487793,
+      "learning_rate": 9.918367346938776e-06,
+      "loss": 1.5742,
+      "mean_token_accuracy": 0.6043413370847702,
+      "num_tokens": 5213337.0,
+      "step": 730
+    },
+    {
+      "epoch": 0.15109749872383868,
+      "grad_norm": 17.927433013916016,
+      "learning_rate": 9.999999300105633e-06,
+      "loss": 1.5007,
+      "mean_token_accuracy": 0.6198744863271713,
+      "num_tokens": 5284294.0,
+      "step": 740
+    },
+    {
+      "epoch": 0.15313935681470137,
+      "grad_norm": 14.790064811706543,
+      "learning_rate": 9.999991426296246e-06,
+      "loss": 1.4678,
+      "mean_token_accuracy": 0.6264089643955231,
+      "num_tokens": 5353839.0,
+      "step": 750
+    },
+    {
+      "epoch": 0.15518121490556405,
+      "grad_norm": 15.573653221130371,
+      "learning_rate": 9.999974803823333e-06,
+      "loss": 1.5332,
+      "mean_token_accuracy": 0.614776486158371,
+      "num_tokens": 5423944.0,
+      "step": 760
+    },
+    {
+      "epoch": 0.15722307299642674,
+      "grad_norm": 14.701098442077637,
+      "learning_rate": 9.999949432715983e-06,
+      "loss": 1.4416,
+      "mean_token_accuracy": 0.6279941260814667,
+      "num_tokens": 5497356.0,
+      "step": 770
+    },
+    {
+      "epoch": 0.15926493108728942,
+      "grad_norm": 14.240693092346191,
+      "learning_rate": 9.999915313018586e-06,
+      "loss": 1.5334,
+      "mean_token_accuracy": 0.6209155797958374,
+      "num_tokens": 5568918.0,
+      "step": 780
+    },
+    {
+      "epoch": 0.1613067891781521,
+      "grad_norm": 15.73400592803955,
+      "learning_rate": 9.999872444790843e-06,
+      "loss": 1.4824,
+      "mean_token_accuracy": 0.6152042269706726,
+      "num_tokens": 5638113.0,
+      "step": 790
+    },
+    {
+      "epoch": 0.1633486472690148,
+      "grad_norm": 17.401098251342773,
+      "learning_rate": 9.999820828107761e-06,
+      "loss": 1.478,
+      "mean_token_accuracy": 0.6273152261972428,
+      "num_tokens": 5710578.0,
+      "step": 800
+    },
+    {
+      "epoch": 0.16539050535987748,
+      "grad_norm": 13.625336647033691,
+      "learning_rate": 9.999760463059657e-06,
+      "loss": 1.4973,
+      "mean_token_accuracy": 0.6253129422664643,
+      "num_tokens": 5782260.0,
+      "step": 810
+    },
+    {
+      "epoch": 0.16743236345074017,
+      "grad_norm": 15.314433097839355,
+      "learning_rate": 9.999691349752154e-06,
+      "loss": 1.4774,
+      "mean_token_accuracy": 0.6250138610601426,
+      "num_tokens": 5853512.0,
+      "step": 820
+    },
+    {
+      "epoch": 0.16947422154160285,
+      "grad_norm": 14.873124122619629,
+      "learning_rate": 9.999613488306182e-06,
+      "loss": 1.4189,
+      "mean_token_accuracy": 0.621257945895195,
+      "num_tokens": 5924369.0,
+      "step": 830
+    },
+    {
+      "epoch": 0.17151607963246554,
+      "grad_norm": 17.970434188842773,
+      "learning_rate": 9.999526878857977e-06,
+      "loss": 1.5224,
+      "mean_token_accuracy": 0.6051260381937027,
+      "num_tokens": 5998837.0,
+      "step": 840
+    },
+    {
+      "epoch": 0.17355793772332823,
+      "grad_norm": 14.060564041137695,
+      "learning_rate": 9.999431521559081e-06,
+      "loss": 1.4957,
+      "mean_token_accuracy": 0.6203907072544098,
+      "num_tokens": 6069353.0,
+      "step": 850
+    },
+    {
+      "epoch": 0.1755997958141909,
+      "grad_norm": 15.269739151000977,
+      "learning_rate": 9.99932741657635e-06,
+      "loss": 1.571,
+      "mean_token_accuracy": 0.606784787774086,
+      "num_tokens": 6142037.0,
+      "step": 860
+    },
+    {
+      "epoch": 0.1776416539050536,
+      "grad_norm": 13.99073600769043,
+      "learning_rate": 9.999214564091934e-06,
+      "loss": 1.4143,
+      "mean_token_accuracy": 0.630013570189476,
+      "num_tokens": 6215390.0,
+      "step": 870
+    },
+    {
+      "epoch": 0.17968351199591628,
+      "grad_norm": 15.911304473876953,
+      "learning_rate": 9.999092964303299e-06,
+      "loss": 1.4837,
+      "mean_token_accuracy": 0.6139797419309616,
+      "num_tokens": 6288877.0,
+      "step": 880
+    },
+    {
+      "epoch": 0.18172537008677897,
+      "grad_norm": 14.30933666229248,
+      "learning_rate": 9.99896261742321e-06,
+      "loss": 1.477,
+      "mean_token_accuracy": 0.6204395204782486,
+      "num_tokens": 6360567.0,
+      "step": 890
+    },
+    {
+      "epoch": 0.18376722817764166,
+      "grad_norm": 15.878334045410156,
+      "learning_rate": 9.998823523679739e-06,
+      "loss": 1.4763,
+      "mean_token_accuracy": 0.6281387954950333,
+      "num_tokens": 6431133.0,
+      "step": 900
+    },
+    {
+      "epoch": 0.18580908626850434,
+      "grad_norm": 14.967886924743652,
+      "learning_rate": 9.998675683316266e-06,
+      "loss": 1.486,
+      "mean_token_accuracy": 0.6245438635349274,
+      "num_tokens": 6502623.0,
+      "step": 910
+    },
+    {
+      "epoch": 0.18785094435936703,
+      "grad_norm": 16.480722427368164,
+      "learning_rate": 9.99851909659147e-06,
+      "loss": 1.4937,
+      "mean_token_accuracy": 0.625952273607254,
+      "num_tokens": 6571994.0,
+      "step": 920
+    },
+    {
+      "epoch": 0.18989280245022971,
+      "grad_norm": 13.536521911621094,
+      "learning_rate": 9.99835376377934e-06,
+      "loss": 1.4284,
+      "mean_token_accuracy": 0.6304761558771134,
+      "num_tokens": 6645698.0,
+      "step": 930
+    },
+    {
+      "epoch": 0.1919346605410924,
+      "grad_norm": 13.15491008758545,
+      "learning_rate": 9.99817968516916e-06,
+      "loss": 1.4822,
+      "mean_token_accuracy": 0.611976045370102,
+      "num_tokens": 6720389.0,
+      "step": 940
+    },
+    {
+      "epoch": 0.19397651863195509,
+      "grad_norm": 15.53482723236084,
+      "learning_rate": 9.997996861065526e-06,
+      "loss": 1.4947,
+      "mean_token_accuracy": 0.6212883561849594,
+      "num_tokens": 6790853.0,
+      "step": 950
+    },
+    {
+      "epoch": 0.19601837672281777,
+      "grad_norm": 14.705411911010742,
+      "learning_rate": 9.99780529178833e-06,
+      "loss": 1.489,
+      "mean_token_accuracy": 0.6153444766998291,
+      "num_tokens": 6867474.0,
+      "step": 960
+    },
+    {
+      "epoch": 0.19806023481368046,
+      "grad_norm": 12.790360450744629,
+      "learning_rate": 9.997604977672766e-06,
+      "loss": 1.4949,
+      "mean_token_accuracy": 0.6196657717227936,
+      "num_tokens": 6936835.0,
+      "step": 970
+    },
+    {
+      "epoch": 0.20010209290454314,
+      "grad_norm": 13.104793548583984,
+      "learning_rate": 9.997395919069333e-06,
+      "loss": 1.4875,
+      "mean_token_accuracy": 0.6339592069387436,
+      "num_tokens": 7002460.0,
+      "step": 980
+    },
+    {
+      "epoch": 0.20214395099540583,
+      "grad_norm": 14.803412437438965,
+      "learning_rate": 9.997178116343827e-06,
+      "loss": 1.4656,
+      "mean_token_accuracy": 0.6147948563098907,
+      "num_tokens": 7077503.0,
+      "step": 990
+    },
+    {
+      "epoch": 0.20418580908626852,
+      "grad_norm": 13.249740600585938,
+      "learning_rate": 9.996951569877347e-06,
+      "loss": 1.4248,
+      "step": 1000
+    },
+    {
+      "epoch": 0.20418580908626852,
+      "eval_loss": 1.4924129247665405,
+      "eval_mean_token_accuracy": 0.6186948751602743,
+      "eval_num_tokens": 7148003.0,
+      "eval_runtime": 989.6753,
+      "eval_samples_per_second": 4.996,
+      "eval_steps_per_second": 1.249,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2062276671771312,
+      "grad_norm": 14.814459800720215,
+      "learning_rate": 9.996716280066288e-06,
+      "loss": 1.5102,
+      "mean_token_accuracy": 0.628528282046318,
+      "num_tokens": 7219174.0,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2082695252679939,
+      "grad_norm": 13.832090377807617,
+      "learning_rate": 9.996472247322345e-06,
+      "loss": 1.4791,
+      "mean_token_accuracy": 0.6212698370218277,
+      "num_tokens": 7290848.0,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21031138335885655,
+      "grad_norm": 14.368582725524902,
+      "learning_rate": 9.996219472072508e-06,
+      "loss": 1.4015,
+      "mean_token_accuracy": 0.6343317806720734,
+      "num_tokens": 7361488.0,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21235324144971923,
+      "grad_norm": 14.122415542602539,
+      "learning_rate": 9.995957954759073e-06,
+      "loss": 1.5137,
+      "mean_token_accuracy": 0.6238163977861404,
+      "num_tokens": 7432355.0,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21439509954058192,
+      "grad_norm": 13.9231538772583,
+      "learning_rate": 9.995687695839622e-06,
+      "loss": 1.5704,
+      "mean_token_accuracy": 0.6070802628993988,
+      "num_tokens": 7499569.0,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2164369576314446,
+      "grad_norm": 14.253915786743164,
+      "learning_rate": 9.995408695787038e-06,
+      "loss": 1.4669,
+      "mean_token_accuracy": 0.6284116208553314,
+      "num_tokens": 7572280.0,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2184788157223073,
+      "grad_norm": 15.08125114440918,
+      "learning_rate": 9.995120955089496e-06,
+      "loss": 1.4687,
+      "mean_token_accuracy": 0.6308166861534119,
+      "num_tokens": 7644265.0,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22052067381316998,
+      "grad_norm": 15.171319007873535,
+      "learning_rate": 9.994824474250468e-06,
+      "loss": 1.4042,
+      "mean_token_accuracy": 0.6392420619726181,
+      "num_tokens": 7715965.0,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22256253190403266,
+      "grad_norm": 15.727775573730469,
+      "learning_rate": 9.994519253788717e-06,
+      "loss": 1.4386,
+      "mean_token_accuracy": 0.6239751428365707,
+      "num_tokens": 7786521.0,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22460438999489535,
+      "grad_norm": 13.040878295898438,
+      "learning_rate": 9.994205294238297e-06,
+      "loss": 1.4687,
+      "mean_token_accuracy": 0.6325306296348572,
+      "num_tokens": 7856697.0,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22664624808575803,
+      "grad_norm": 14.698892593383789,
+      "learning_rate": 9.993882596148556e-06,
+      "loss": 1.4996,
+      "mean_token_accuracy": 0.6204117625951767,
+      "num_tokens": 7925119.0,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22868810617662072,
+      "grad_norm": 15.8075590133667,
+      "learning_rate": 9.993551160084128e-06,
+      "loss": 1.5177,
+      "mean_token_accuracy": 0.6046328634023667,
+      "num_tokens": 7994808.0,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2307299642674834,
+      "grad_norm": 13.777670860290527,
+      "learning_rate": 9.99321098662494e-06,
+      "loss": 1.443,
+      "mean_token_accuracy": 0.6219522416591644,
+      "num_tokens": 8066587.0,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2327718223583461,
+      "grad_norm": 14.019599914550781,
+      "learning_rate": 9.992862076366203e-06,
+      "loss": 1.4232,
+      "mean_token_accuracy": 0.6349723309278488,
+      "num_tokens": 8140601.0,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23481368044920878,
+      "grad_norm": 13.211136817932129,
+      "learning_rate": 9.992504429918422e-06,
+      "loss": 1.3774,
+      "mean_token_accuracy": 0.6400786489248276,
+      "num_tokens": 8211916.0,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23685553854007146,
+      "grad_norm": 15.216955184936523,
+      "learning_rate": 9.992138047907382e-06,
+      "loss": 1.4249,
+      "mean_token_accuracy": 0.6270944118499756,
+      "num_tokens": 8280425.0,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23889739663093415,
+      "grad_norm": 15.230463027954102,
+      "learning_rate": 9.991762930974154e-06,
+      "loss": 1.4987,
+      "mean_token_accuracy": 0.6189264446496964,
+      "num_tokens": 8348729.0,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24093925472179684,
+      "grad_norm": 13.841259956359863,
+      "learning_rate": 9.991379079775093e-06,
+      "loss": 1.4614,
+      "mean_token_accuracy": 0.6259691953659058,
+      "num_tokens": 8419931.0,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24298111281265952,
+      "grad_norm": 14.62938117980957,
+      "learning_rate": 9.990986494981837e-06,
+      "loss": 1.4555,
+      "mean_token_accuracy": 0.6304128617048264,
+      "num_tokens": 8493325.0,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2450229709035222,
+      "grad_norm": 14.569291114807129,
+      "learning_rate": 9.990585177281312e-06,
+      "loss": 1.4874,
+      "mean_token_accuracy": 0.621327030658722,
+      "num_tokens": 8563202.0,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2470648289943849,
+      "grad_norm": 14.350664138793945,
+      "learning_rate": 9.990175127375708e-06,
+      "loss": 1.4037,
+      "mean_token_accuracy": 0.6393092036247253,
+      "num_tokens": 8636785.0,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24910668708524758,
+      "grad_norm": 14.5271577835083,
+      "learning_rate": 9.989756345982508e-06,
+      "loss": 1.3848,
+      "mean_token_accuracy": 0.6416534215211869,
+      "num_tokens": 8713375.0,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25114854517611024,
+      "grad_norm": 14.138811111450195,
+      "learning_rate": 9.989328833834472e-06,
+      "loss": 1.4711,
+      "mean_token_accuracy": 0.6263648480176925,
+      "num_tokens": 8786119.0,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25319040326697295,
+      "grad_norm": 11.65623664855957,
+      "learning_rate": 9.98889259167963e-06,
+      "loss": 1.435,
+      "mean_token_accuracy": 0.6322738409042359,
+      "num_tokens": 8858592.0,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2552322613578356,
+      "grad_norm": 15.826696395874023,
+      "learning_rate": 9.98844762028129e-06,
+      "loss": 1.441,
+      "mean_token_accuracy": 0.6359235972166062,
+      "num_tokens": 8928180.0,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2572741194486983,
+      "grad_norm": 17.17336082458496,
+      "learning_rate": 9.987993920418036e-06,
+      "loss": 1.4623,
+      "mean_token_accuracy": 0.6244770854711532,
+      "num_tokens": 9001234.0,
+      "step": 1260
+    },
+    {
+      "epoch": 0.259315977539561,
+      "grad_norm": 14.226313591003418,
+      "learning_rate": 9.987531492883722e-06,
+      "loss": 1.4608,
+      "mean_token_accuracy": 0.6295191615819931,
+      "num_tokens": 9077262.0,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2613578356304237,
+      "grad_norm": 14.853294372558594,
+      "learning_rate": 9.987060338487475e-06,
+      "loss": 1.4514,
+      "mean_token_accuracy": 0.6238030642271042,
+      "num_tokens": 9151546.0,
+      "step": 1280
+    },
+    {
+      "epoch": 0.26339969372128635,
+      "grad_norm": 15.472508430480957,
+      "learning_rate": 9.98658045805369e-06,
+      "loss": 1.4623,
+      "mean_token_accuracy": 0.6118121117353439,
+      "num_tokens": 9224771.0,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26544155181214907,
+      "grad_norm": 14.506217002868652,
+      "learning_rate": 9.98609185242203e-06,
+      "loss": 1.4837,
+      "mean_token_accuracy": 0.6259201139211654,
+      "num_tokens": 9296598.0,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2674834099030117,
+      "grad_norm": 15.796782493591309,
+      "learning_rate": 9.985594522447427e-06,
+      "loss": 1.4381,
+      "mean_token_accuracy": 0.6328779578208923,
+      "num_tokens": 9367813.0,
+      "step": 1310
+    },
+    {
+      "epoch": 0.26952526799387444,
+      "grad_norm": 16.872446060180664,
+      "learning_rate": 9.985088469000075e-06,
+      "loss": 1.4738,
+      "mean_token_accuracy": 0.6319139033555985,
+      "num_tokens": 9441995.0,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2715671260847371,
+      "grad_norm": 12.5946683883667,
+      "learning_rate": 9.984573692965439e-06,
+      "loss": 1.4395,
+      "mean_token_accuracy": 0.6213337570428848,
+      "num_tokens": 9512888.0,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2736089841755998,
+      "grad_norm": 15.489473342895508,
+      "learning_rate": 9.984050195244233e-06,
+      "loss": 1.4946,
+      "mean_token_accuracy": 0.62346653342247,
+      "num_tokens": 9583532.0,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27565084226646247,
+      "grad_norm": 16.15251922607422,
+      "learning_rate": 9.983517976752447e-06,
+      "loss": 1.5034,
+      "mean_token_accuracy": 0.6158888429403305,
+      "num_tokens": 9650656.0,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2776927003573252,
+      "grad_norm": 13.289031982421875,
+      "learning_rate": 9.98297703842132e-06,
+      "loss": 1.4061,
+      "mean_token_accuracy": 0.6388078480958939,
+      "num_tokens": 9724038.0,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27973455844818784,
+      "grad_norm": 14.971853256225586,
+      "learning_rate": 9.982427381197349e-06,
+      "loss": 1.3735,
+      "mean_token_accuracy": 0.6322133481502533,
+      "num_tokens": 9792860.0,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28177641653905056,
+      "grad_norm": 15.422203063964844,
+      "learning_rate": 9.981869006042291e-06,
+      "loss": 1.4731,
+      "mean_token_accuracy": 0.6308643728494644,
+      "num_tokens": 9863918.0,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2838182746299132,
+      "grad_norm": 14.272014617919922,
+      "learning_rate": 9.981301913933154e-06,
+      "loss": 1.404,
+      "mean_token_accuracy": 0.6412104159593582,
+      "num_tokens": 9937896.0,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28586013272077593,
+      "grad_norm": 14.743646621704102,
+      "learning_rate": 9.980726105862202e-06,
+      "loss": 1.4821,
+      "mean_token_accuracy": 0.6105817914009094,
+      "num_tokens": 10009650.0,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2879019908116386,
+      "grad_norm": 13.515847206115723,
+      "learning_rate": 9.980141582836944e-06,
+      "loss": 1.4952,
+      "mean_token_accuracy": 0.6141276836395264,
+      "num_tokens": 10080992.0,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2899438489025013,
+      "grad_norm": 12.919459342956543,
+      "learning_rate": 9.979548345880142e-06,
+      "loss": 1.445,
+      "mean_token_accuracy": 0.6350033283233643,
+      "num_tokens": 10152759.0,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29198570699336396,
+      "grad_norm": 14.708041191101074,
+      "learning_rate": 9.978946396029802e-06,
+      "loss": 1.45,
+      "mean_token_accuracy": 0.6281395435333252,
+      "num_tokens": 10222271.0,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29402756508422667,
+      "grad_norm": 13.272540092468262,
+      "learning_rate": 9.97833573433918e-06,
+      "loss": 1.428,
+      "mean_token_accuracy": 0.6370621621608734,
+      "num_tokens": 10294416.0,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29606942317508933,
+      "grad_norm": 12.510384559631348,
+      "learning_rate": 9.97771636187677e-06,
+      "loss": 1.4699,
+      "mean_token_accuracy": 0.6264015674591065,
+      "num_tokens": 10367596.0,
+      "step": 1450
+    },
+    {
+      "epoch": 0.29811128126595204,
+      "grad_norm": 13.551581382751465,
+      "learning_rate": 9.977088279726312e-06,
+      "loss": 1.4212,
+      "mean_token_accuracy": 0.6329251319169998,
+      "num_tokens": 10441159.0,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3001531393568147,
+      "grad_norm": 19.371631622314453,
+      "learning_rate": 9.976451488986785e-06,
+      "loss": 1.4853,
+      "mean_token_accuracy": 0.6176414400339126,
+      "num_tokens": 10507615.0,
+      "step": 1470
+    },
+    {
+      "epoch": 0.30219499744767736,
+      "grad_norm": 14.074532508850098,
+      "learning_rate": 9.975805990772401e-06,
+      "loss": 1.3653,
+      "mean_token_accuracy": 0.6459196925163269,
+      "num_tokens": 10578665.0,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3042368555385401,
+      "grad_norm": 14.489356994628906,
+      "learning_rate": 9.975151786212614e-06,
+      "loss": 1.4668,
+      "mean_token_accuracy": 0.6257261455059051,
+      "num_tokens": 10652077.0,
+      "step": 1490
+    },
+    {
+      "epoch": 0.30627871362940273,
+      "grad_norm": 13.164527893066406,
+      "learning_rate": 9.974488876452107e-06,
+      "loss": 1.3998,
+      "mean_token_accuracy": 0.6408104240894318,
+      "num_tokens": 10723815.0,
+      "step": 1500
+    },
+    {
+      "epoch": 0.30832057172026545,
+      "grad_norm": 13.87382698059082,
+      "learning_rate": 9.973817262650798e-06,
+      "loss": 1.464,
+      "mean_token_accuracy": 0.6288924962282181,
+      "num_tokens": 10794989.0,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3103624298111281,
+      "grad_norm": 14.771330833435059,
+      "learning_rate": 9.973136945983832e-06,
+      "loss": 1.378,
+      "mean_token_accuracy": 0.6336318165063858,
+      "num_tokens": 10867534.0,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3124042879019908,
+      "grad_norm": 15.911246299743652,
+      "learning_rate": 9.972447927641588e-06,
+      "loss": 1.4556,
+      "mean_token_accuracy": 0.625057402253151,
+      "num_tokens": 10936439.0,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3144461459928535,
+      "grad_norm": 14.86609172821045,
+      "learning_rate": 9.971750208829661e-06,
+      "loss": 1.4053,
+      "mean_token_accuracy": 0.642861407995224,
+      "num_tokens": 11005828.0,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3164880040837162,
+      "grad_norm": 15.707890510559082,
+      "learning_rate": 9.971043790768878e-06,
+      "loss": 1.4513,
+      "mean_token_accuracy": 0.6278130769729614,
+      "num_tokens": 11079668.0,
+      "step": 1550
+    },
+    {
+      "epoch": 0.31852986217457885,
+      "grad_norm": 14.676016807556152,
+      "learning_rate": 9.97032867469528e-06,
+      "loss": 1.4306,
+      "mean_token_accuracy": 0.6367242813110352,
+      "num_tokens": 11155772.0,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32057172026544156,
+      "grad_norm": 13.341631889343262,
+      "learning_rate": 9.969604861860133e-06,
+      "loss": 1.4878,
+      "mean_token_accuracy": 0.6201000392436982,
+      "num_tokens": 11227505.0,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3226135783563042,
+      "grad_norm": 13.484333992004395,
+      "learning_rate": 9.968872353529921e-06,
+      "loss": 1.4509,
+      "mean_token_accuracy": 0.6225223064422607,
+      "num_tokens": 11297087.0,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32465543644716693,
+      "grad_norm": 14.83188533782959,
+      "learning_rate": 9.968131150986337e-06,
+      "loss": 1.4449,
+      "mean_token_accuracy": 0.6267853021621704,
+      "num_tokens": 11367600.0,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3266972945380296,
+      "grad_norm": 14.633095741271973,
+      "learning_rate": 9.967381255526291e-06,
+      "loss": 1.3779,
+      "mean_token_accuracy": 0.6430699110031128,
+      "num_tokens": 11440350.0,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3287391526288923,
+      "grad_norm": 14.898681640625,
+      "learning_rate": 9.966622668461899e-06,
+      "loss": 1.3796,
+      "mean_token_accuracy": 0.6352171957492828,
+      "num_tokens": 11508529.0,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33078101071975496,
+      "grad_norm": 13.815352439880371,
+      "learning_rate": 9.965855391120492e-06,
+      "loss": 1.4781,
+      "mean_token_accuracy": 0.6252012938261032,
+      "num_tokens": 11584229.0,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3328228688106177,
+      "grad_norm": 13.569513320922852,
+      "learning_rate": 9.965079424844599e-06,
+      "loss": 1.3532,
+      "mean_token_accuracy": 0.6486948579549789,
+      "num_tokens": 11653163.0,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33486472690148034,
+      "grad_norm": 14.037538528442383,
+      "learning_rate": 9.964294770991958e-06,
+      "loss": 1.4509,
+      "mean_token_accuracy": 0.6179214477539062,
+      "num_tokens": 11722399.0,
+      "step": 1640
+    },
+    {
+      "epoch": 0.33690658499234305,
+      "grad_norm": 14.93041706085205,
+      "learning_rate": 9.963501430935505e-06,
+      "loss": 1.4345,
+      "mean_token_accuracy": 0.6286086499691009,
+      "num_tokens": 11795038.0,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3389484430832057,
+      "grad_norm": 14.412076950073242,
+      "learning_rate": 9.962699406063374e-06,
+      "loss": 1.4148,
+      "mean_token_accuracy": 0.6350490599870682,
+      "num_tokens": 11865144.0,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3409903011740684,
+      "grad_norm": 13.61820125579834,
+      "learning_rate": 9.9618886977789e-06,
+      "loss": 1.4613,
+      "mean_token_accuracy": 0.6232977896928787,
+      "num_tokens": 11936312.0,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3430321592649311,
+      "grad_norm": 13.12203598022461,
+      "learning_rate": 9.961069307500606e-06,
+      "loss": 1.3669,
+      "mean_token_accuracy": 0.6328762382268905,
+      "num_tokens": 12005821.0,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3450740173557938,
+      "grad_norm": 13.293590545654297,
+      "learning_rate": 9.960241236662209e-06,
+      "loss": 1.3379,
+      "mean_token_accuracy": 0.6402474969625473,
+      "num_tokens": 12074453.0,
+      "step": 1690
+    },
+    {
+      "epoch": 0.34711587544665645,
+      "grad_norm": 14.68148422241211,
+      "learning_rate": 9.959404486712613e-06,
+      "loss": 1.4089,
+      "mean_token_accuracy": 0.6392355740070343,
+      "num_tokens": 12142676.0,
+      "step": 1700
+    },
+    {
+      "epoch": 0.34915773353751917,
+      "grad_norm": 14.123222351074219,
+      "learning_rate": 9.958559059115911e-06,
+      "loss": 1.4286,
+      "mean_token_accuracy": 0.6322161883115769,
+      "num_tokens": 12213513.0,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3511995916283818,
+      "grad_norm": 15.215718269348145,
+      "learning_rate": 9.957704955351377e-06,
+      "loss": 1.4943,
+      "mean_token_accuracy": 0.6219414353370667,
+      "num_tokens": 12283923.0,
+      "step": 1720
+    },
+    {
+      "epoch": 0.35324144971924454,
+      "grad_norm": 13.124506950378418,
+      "learning_rate": 9.956842176913467e-06,
+      "loss": 1.4513,
+      "mean_token_accuracy": 0.6365837067365646,
+      "num_tokens": 12358198.0,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3552833078101072,
+      "grad_norm": 15.533964157104492,
+      "learning_rate": 9.955970725311814e-06,
+      "loss": 1.4831,
+      "mean_token_accuracy": 0.6205825060606003,
+      "num_tokens": 12428720.0,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3573251659009699,
+      "grad_norm": 15.942651748657227,
+      "learning_rate": 9.95509060207123e-06,
+      "loss": 1.4388,
+      "mean_token_accuracy": 0.6281418234109879,
+      "num_tokens": 12496646.0,
+      "step": 1750
+    },
+    {
+      "epoch": 0.35936702399183257,
+      "grad_norm": 11.91575813293457,
+      "learning_rate": 9.954201808731698e-06,
+      "loss": 1.4251,
+      "mean_token_accuracy": 0.624052032828331,
+      "num_tokens": 12565166.0,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3614088820826952,
+      "grad_norm": 14.103416442871094,
+      "learning_rate": 9.953304346848372e-06,
+      "loss": 1.3823,
+      "mean_token_accuracy": 0.6392663180828094,
+      "num_tokens": 12636558.0,
+      "step": 1770
+    },
+    {
+      "epoch": 0.36345074017355794,
+      "grad_norm": 13.818516731262207,
+      "learning_rate": 9.952398217991569e-06,
+      "loss": 1.4778,
+      "mean_token_accuracy": 0.6283582001924515,
+      "num_tokens": 12708824.0,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3654925982644206,
+      "grad_norm": 13.625807762145996,
+      "learning_rate": 9.951483423746776e-06,
+      "loss": 1.4254,
+      "mean_token_accuracy": 0.6290278375148773,
+      "num_tokens": 12778732.0,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3675344563552833,
+      "grad_norm": 14.069265365600586,
+      "learning_rate": 9.950559965714647e-06,
+      "loss": 1.4622,
+      "mean_token_accuracy": 0.6254006132483483,
+      "num_tokens": 12848482.0,
+      "step": 1800
+    },
+    {
+      "epoch": 0.36957631444614597,
+      "grad_norm": 14.092951774597168,
+      "learning_rate": 9.949627845510984e-06,
+      "loss": 1.4371,
+      "mean_token_accuracy": 0.6362253725528717,
+      "num_tokens": 12919341.0,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3716181725370087,
+      "grad_norm": 14.89527702331543,
+      "learning_rate": 9.948687064766753e-06,
+      "loss": 1.4774,
+      "mean_token_accuracy": 0.6274341195821762,
+      "num_tokens": 12990953.0,
+      "step": 1820
+    },
+    {
+      "epoch": 0.37366003062787134,
+      "grad_norm": 15.057639122009277,
+      "learning_rate": 9.947737625128068e-06,
+      "loss": 1.3877,
+      "mean_token_accuracy": 0.6474756151437759,
+      "num_tokens": 13061104.0,
+      "step": 1830
+    },
+    {
+      "epoch": 0.37570188871873406,
+      "grad_norm": 13.421624183654785,
+      "learning_rate": 9.946779528256204e-06,
+      "loss": 1.4259,
+      "mean_token_accuracy": 0.636455500125885,
+      "num_tokens": 13132168.0,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3777437468095967,
+      "grad_norm": 13.868426322937012,
+      "learning_rate": 9.945812775827573e-06,
+      "loss": 1.4056,
+      "mean_token_accuracy": 0.635055473446846,
+      "num_tokens": 13202107.0,
+      "step": 1850
+    },
+    {
+      "epoch": 0.37978560490045943,
+      "grad_norm": 13.789758682250977,
+      "learning_rate": 9.944837369533737e-06,
+      "loss": 1.4521,
+      "mean_token_accuracy": 0.6354493737220764,
+      "num_tokens": 13273691.0,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3818274629913221,
+      "grad_norm": 14.410371780395508,
+      "learning_rate": 9.943853311081398e-06,
+      "loss": 1.4376,
+      "mean_token_accuracy": 0.6387281328439712,
+      "num_tokens": 13343207.0,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3838693210821848,
+      "grad_norm": 13.178431510925293,
+      "learning_rate": 9.9428606021924e-06,
+      "loss": 1.4425,
+      "mean_token_accuracy": 0.6428033947944641,
+      "num_tokens": 13413824.0,
+      "step": 1880
+    },
+    {
+      "epoch": 0.38591117917304746,
+      "grad_norm": 13.29395866394043,
+      "learning_rate": 9.941859244603721e-06,
+      "loss": 1.469,
+      "mean_token_accuracy": 0.6198657006025314,
+      "num_tokens": 13484476.0,
+      "step": 1890
+    },
+    {
+      "epoch": 0.38795303726391017,
+      "grad_norm": 16.564151763916016,
+      "learning_rate": 9.940849240067471e-06,
+      "loss": 1.4815,
+      "mean_token_accuracy": 0.6274239957332611,
+      "num_tokens": 13558301.0,
+      "step": 1900
+    },
+    {
+      "epoch": 0.38999489535477283,
+      "grad_norm": 13.39206600189209,
+      "learning_rate": 9.939830590350895e-06,
+      "loss": 1.4434,
+      "mean_token_accuracy": 0.6292670220136642,
+      "num_tokens": 13628067.0,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39203675344563554,
+      "grad_norm": 12.907182693481445,
+      "learning_rate": 9.938803297236354e-06,
+      "loss": 1.3603,
+      "mean_token_accuracy": 0.6486239820718765,
+      "num_tokens": 13701747.0,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3940786115364982,
+      "grad_norm": 14.168998718261719,
+      "learning_rate": 9.937767362521343e-06,
+      "loss": 1.4568,
+      "mean_token_accuracy": 0.6249212712049484,
+      "num_tokens": 13772760.0,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3961204696273609,
+      "grad_norm": 12.438258171081543,
+      "learning_rate": 9.936722788018476e-06,
+      "loss": 1.3811,
+      "mean_token_accuracy": 0.6337268084287644,
+      "num_tokens": 13842794.0,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3981623277182236,
+      "grad_norm": 14.198771476745605,
+      "learning_rate": 9.935669575555478e-06,
+      "loss": 1.4326,
+      "mean_token_accuracy": 0.6176476001739502,
+      "num_tokens": 13916581.0,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4002041858090863,
+      "grad_norm": 15.927424430847168,
+      "learning_rate": 9.934607726975196e-06,
+      "loss": 1.3843,
+      "mean_token_accuracy": 0.645603460073471,
+      "num_tokens": 13990796.0,
+      "step": 1960
+    },
+    {
+      "epoch": 0.40224604389994895,
+      "grad_norm": 13.675074577331543,
+      "learning_rate": 9.933537244135584e-06,
+      "loss": 1.4744,
+      "mean_token_accuracy": 0.6207163512706757,
+      "num_tokens": 14058770.0,
+      "step": 1970
+    },
+    {
+      "epoch": 0.40428790199081166,
+      "grad_norm": 14.225303649902344,
+      "learning_rate": 9.9324581289097e-06,
+      "loss": 1.3557,
+      "mean_token_accuracy": 0.6434049308300018,
+      "num_tokens": 14127780.0,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4063297600816743,
+      "grad_norm": 13.11447811126709,
+      "learning_rate": 9.931370383185717e-06,
+      "loss": 1.4163,
+      "mean_token_accuracy": 0.6322284698486328,
+      "num_tokens": 14199180.0,
+      "step": 1990
+    },
+    {
+      "epoch": 0.40837161817253703,
+      "grad_norm": 13.91115951538086,
+      "learning_rate": 9.930274008866898e-06,
+      "loss": 1.4281,
+      "step": 2000
+    },
+    {
+      "epoch": 0.40837161817253703,
+      "eval_loss": 1.4562746286392212,
+      "eval_mean_token_accuracy": 0.626390626877064,
+      "eval_num_tokens": 14272984.0,
+      "eval_runtime": 992.5565,
+      "eval_samples_per_second": 4.981,
+      "eval_steps_per_second": 1.245,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4104134762633997,
+      "grad_norm": 12.996975898742676,
+      "learning_rate": 9.92916900787161e-06,
+      "loss": 1.4825,
+      "mean_token_accuracy": 0.6286307215690613,
+      "num_tokens": 14345206.0,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4124553343542624,
+      "grad_norm": 14.120156288146973,
+      "learning_rate": 9.928055382133311e-06,
+      "loss": 1.5034,
+      "mean_token_accuracy": 0.6243163645267487,
+      "num_tokens": 14419062.0,
+      "step": 2020
+    },
+    {
+      "epoch": 0.41449719244512506,
+      "grad_norm": 13.214719772338867,
+      "learning_rate": 9.926933133600554e-06,
+      "loss": 1.4539,
+      "mean_token_accuracy": 0.6286552488803864,
+      "num_tokens": 14490124.0,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4165390505359878,
+      "grad_norm": 14.170852661132812,
+      "learning_rate": 9.925802264236974e-06,
+      "loss": 1.437,
+      "mean_token_accuracy": 0.6257667273283005,
+      "num_tokens": 14561490.0,
+      "step": 2040
+    },
+    {
+      "epoch": 0.41858090862685043,
+      "grad_norm": 13.008140563964844,
+      "learning_rate": 9.924662776021297e-06,
+      "loss": 1.4631,
+      "mean_token_accuracy": 0.6202176958322525,
+      "num_tokens": 14634587.0,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4206227667177131,
+      "grad_norm": 12.924283027648926,
+      "learning_rate": 9.923514670947326e-06,
+      "loss": 1.4184,
+      "mean_token_accuracy": 0.6351789355278015,
+      "num_tokens": 14704346.0,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4226646248085758,
+      "grad_norm": 15.573639869689941,
+      "learning_rate": 9.922357951023939e-06,
+      "loss": 1.401,
+      "mean_token_accuracy": 0.62881398499012,
+      "num_tokens": 14775676.0,
+      "step": 2070
+    },
+    {
+      "epoch": 0.42470648289943846,
+      "grad_norm": 13.74057388305664,
+      "learning_rate": 9.92119261827509e-06,
+      "loss": 1.392,
+      "mean_token_accuracy": 0.6317652136087417,
+      "num_tokens": 14846972.0,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4267483409903012,
+      "grad_norm": 17.23077964782715,
+      "learning_rate": 9.920018674739808e-06,
+      "loss": 1.4939,
+      "mean_token_accuracy": 0.6289480417966843,
+      "num_tokens": 14921965.0,
+      "step": 2090
+    },
+    {
+      "epoch": 0.42879019908116384,
+      "grad_norm": 14.922390937805176,
+      "learning_rate": 9.918836122472178e-06,
+      "loss": 1.4451,
+      "mean_token_accuracy": 0.6285185784101486,
+      "num_tokens": 14993212.0,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43083205717202655,
+      "grad_norm": 12.999784469604492,
+      "learning_rate": 9.917644963541359e-06,
+      "loss": 1.402,
+      "mean_token_accuracy": 0.6414052009582519,
+      "num_tokens": 15062067.0,
+      "step": 2110
+    },
+    {
+      "epoch": 0.4328739152628892,
+      "grad_norm": 14.179859161376953,
+      "learning_rate": 9.916445200031559e-06,
+      "loss": 1.4342,
+      "mean_token_accuracy": 0.6401284962892533,
+      "num_tokens": 15134303.0,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4349157733537519,
+      "grad_norm": 13.935656547546387,
+      "learning_rate": 9.915236834042052e-06,
+      "loss": 1.4358,
+      "mean_token_accuracy": 0.6319669187068939,
+      "num_tokens": 15205632.0,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4369576314446146,
+      "grad_norm": 14.011245727539062,
+      "learning_rate": 9.914019867687157e-06,
+      "loss": 1.4334,
+      "mean_token_accuracy": 0.6271157622337341,
+      "num_tokens": 15276462.0,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4389994895354773,
+      "grad_norm": 15.309188842773438,
+      "learning_rate": 9.912794303096244e-06,
+      "loss": 1.4595,
+      "mean_token_accuracy": 0.6256224781274795,
+      "num_tokens": 15347114.0,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44104134762633995,
+      "grad_norm": 12.190061569213867,
+      "learning_rate": 9.911560142413725e-06,
+      "loss": 1.3799,
+      "mean_token_accuracy": 0.6491726189851761,
+      "num_tokens": 15422516.0,
+      "step": 2160
+    },
+    {
+      "epoch": 0.44308320571720267,
+      "grad_norm": 14.980606079101562,
+      "learning_rate": 9.910317387799058e-06,
+      "loss": 1.3266,
+      "mean_token_accuracy": 0.6502772122621536,
+      "num_tokens": 15498472.0,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4451250638080653,
+      "grad_norm": 14.175018310546875,
+      "learning_rate": 9.909066041426733e-06,
+      "loss": 1.4373,
+      "mean_token_accuracy": 0.6358803600072861,
+      "num_tokens": 15566017.0,
+      "step": 2180
+    },
+    {
+      "epoch": 0.44716692189892804,
+      "grad_norm": 12.999357223510742,
+      "learning_rate": 9.907806105486278e-06,
+      "loss": 1.4096,
+      "mean_token_accuracy": 0.6316717475652694,
+      "num_tokens": 15639585.0,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4492087799897907,
+      "grad_norm": 12.991114616394043,
+      "learning_rate": 9.906537582182245e-06,
+      "loss": 1.4713,
+      "mean_token_accuracy": 0.6206406474113464,
+      "num_tokens": 15713767.0,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4512506380806534,
+      "grad_norm": 16.122135162353516,
+      "learning_rate": 9.905260473734215e-06,
+      "loss": 1.3514,
+      "mean_token_accuracy": 0.6418723970651626,
+      "num_tokens": 15786377.0,
+      "step": 2210
+    },
+    {
+      "epoch": 0.45329249617151607,
+      "grad_norm": 15.44117259979248,
+      "learning_rate": 9.903974782376792e-06,
+      "loss": 1.408,
+      "mean_token_accuracy": 0.6281966835260391,
+      "num_tokens": 15854713.0,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4553343542623788,
+      "grad_norm": 13.286972045898438,
+      "learning_rate": 9.902680510359595e-06,
+      "loss": 1.4129,
+      "mean_token_accuracy": 0.6257620751857758,
+      "num_tokens": 15929629.0,
+      "step": 2230
+    },
+    {
+      "epoch": 0.45737621235324144,
+      "grad_norm": 13.401151657104492,
+      "learning_rate": 9.901377659947257e-06,
+      "loss": 1.4552,
+      "mean_token_accuracy": 0.6300186216831207,
+      "num_tokens": 15998948.0,
+      "step": 2240
+    },
+    {
+      "epoch": 0.45941807044410415,
+      "grad_norm": 14.887855529785156,
+      "learning_rate": 9.900066233419423e-06,
+      "loss": 1.4324,
+      "mean_token_accuracy": 0.6363310009241104,
+      "num_tokens": 16068487.0,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4614599285349668,
+      "grad_norm": 15.288812637329102,
+      "learning_rate": 9.898746233070744e-06,
+      "loss": 1.5303,
+      "mean_token_accuracy": 0.6137059539556503,
+      "num_tokens": 16139564.0,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4635017866258295,
+      "grad_norm": 13.743548393249512,
+      "learning_rate": 9.897417661210869e-06,
+      "loss": 1.3417,
+      "mean_token_accuracy": 0.6564855366945267,
+      "num_tokens": 16205885.0,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4655436447166922,
+      "grad_norm": 13.705702781677246,
+      "learning_rate": 9.89608052016445e-06,
+      "loss": 1.4518,
+      "mean_token_accuracy": 0.6158504068851471,
+      "num_tokens": 16277491.0,
+      "step": 2280
+    },
+    {
+      "epoch": 0.4675855028075549,
+      "grad_norm": 13.374293327331543,
+      "learning_rate": 9.894734812271128e-06,
+      "loss": 1.4323,
+      "mean_token_accuracy": 0.6367385804653167,
+      "num_tokens": 16349026.0,
+      "step": 2290
+    },
+    {
+      "epoch": 0.46962736089841756,
+      "grad_norm": 14.050286293029785,
+      "learning_rate": 9.89338053988554e-06,
+      "loss": 1.3748,
+      "mean_token_accuracy": 0.644825917482376,
+      "num_tokens": 16419614.0,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47166921898928027,
+      "grad_norm": 12.7362060546875,
+      "learning_rate": 9.8920177053773e-06,
+      "loss": 1.3568,
+      "mean_token_accuracy": 0.6447948306798935,
+      "num_tokens": 16487160.0,
+      "step": 2310
+    },
+    {
+      "epoch": 0.47371107708014293,
+      "grad_norm": 13.328147888183594,
+      "learning_rate": 9.890646311131011e-06,
+      "loss": 1.4306,
+      "mean_token_accuracy": 0.616517773270607,
+      "num_tokens": 16560163.0,
+      "step": 2320
+    },
+    {
+      "epoch": 0.47575293517100564,
+      "grad_norm": 14.35744571685791,
+      "learning_rate": 9.88926635954625e-06,
+      "loss": 1.4315,
+      "mean_token_accuracy": 0.6324984401464462,
+      "num_tokens": 16633229.0,
+      "step": 2330
+    },
+    {
+      "epoch": 0.4777947932618683,
+      "grad_norm": 14.388071060180664,
+      "learning_rate": 9.887877853037568e-06,
+      "loss": 1.4924,
+      "mean_token_accuracy": 0.6160824686288834,
+      "num_tokens": 16703383.0,
+      "step": 2340
+    },
+    {
+      "epoch": 0.47983665135273096,
+      "grad_norm": 13.642940521240234,
+      "learning_rate": 9.886480794034485e-06,
+      "loss": 1.4676,
+      "mean_token_accuracy": 0.6276875615119935,
+      "num_tokens": 16778245.0,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48187850944359367,
+      "grad_norm": 14.308442115783691,
+      "learning_rate": 9.885075184981484e-06,
+      "loss": 1.3762,
+      "mean_token_accuracy": 0.6397514760494232,
+      "num_tokens": 16851438.0,
+      "step": 2360
+    },
+    {
+      "epoch": 0.48392036753445633,
+      "grad_norm": 15.114699363708496,
+      "learning_rate": 9.883661028338009e-06,
+      "loss": 1.4637,
+      "mean_token_accuracy": 0.6210479766130448,
+      "num_tokens": 16921239.0,
+      "step": 2370
+    },
+    {
+      "epoch": 0.48596222562531904,
+      "grad_norm": 16.276016235351562,
+      "learning_rate": 9.88223832657846e-06,
+      "loss": 1.4999,
+      "mean_token_accuracy": 0.6109961330890655,
+      "num_tokens": 16993286.0,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4880040837161817,
+      "grad_norm": 16.859468460083008,
+      "learning_rate": 9.88080708219219e-06,
+      "loss": 1.4623,
+      "mean_token_accuracy": 0.6310494840145111,
+      "num_tokens": 17066589.0,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4900459418070444,
+      "grad_norm": 15.107973098754883,
+      "learning_rate": 9.879367297683499e-06,
+      "loss": 1.4906,
+      "mean_token_accuracy": 0.6145053148269654,
+      "num_tokens": 17137188.0,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4920877998979071,
+      "grad_norm": 12.906387329101562,
+      "learning_rate": 9.877918975571627e-06,
+      "loss": 1.3995,
+      "mean_token_accuracy": 0.6356921285390854,
+      "num_tokens": 17205087.0,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4941296579887698,
+      "grad_norm": 13.7039213180542,
+      "learning_rate": 9.876462118390758e-06,
+      "loss": 1.35,
+      "mean_token_accuracy": 0.642990505695343,
+      "num_tokens": 17279430.0,
+      "step": 2420
+    },
+    {
+      "epoch": 0.49617151607963245,
+      "grad_norm": 13.48718547821045,
+      "learning_rate": 9.874996728690005e-06,
+      "loss": 1.4008,
+      "mean_token_accuracy": 0.6324507027864457,
+      "num_tokens": 17345739.0,
+      "step": 2430
+    },
+    {
+      "epoch": 0.49821337417049516,
+      "grad_norm": 13.046002388000488,
+      "learning_rate": 9.873522809033414e-06,
+      "loss": 1.4148,
+      "mean_token_accuracy": 0.6250188022851944,
+      "num_tokens": 17415957.0,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5002552322613578,
+      "grad_norm": 12.571805953979492,
+      "learning_rate": 9.872040361999953e-06,
+      "loss": 1.3033,
+      "mean_token_accuracy": 0.6560938596725464,
+      "num_tokens": 17488686.0,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5022970903522205,
+      "grad_norm": 13.674259185791016,
+      "learning_rate": 9.870549390183513e-06,
+      "loss": 1.3663,
+      "mean_token_accuracy": 0.6375747203826905,
+      "num_tokens": 17564148.0,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5043389484430832,
+      "grad_norm": 13.26236343383789,
+      "learning_rate": 9.869049896192902e-06,
+      "loss": 1.4163,
+      "mean_token_accuracy": 0.6320332467556,
+      "num_tokens": 17635358.0,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5063808065339459,
+      "grad_norm": 13.495912551879883,
+      "learning_rate": 9.867541882651839e-06,
+      "loss": 1.4292,
+      "mean_token_accuracy": 0.6307185232639313,
+      "num_tokens": 17708919.0,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5084226646248086,
+      "grad_norm": 15.353795051574707,
+      "learning_rate": 9.866025352198944e-06,
+      "loss": 1.4209,
+      "mean_token_accuracy": 0.6328317284584045,
+      "num_tokens": 17784820.0,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5104645227156712,
+      "grad_norm": 14.236472129821777,
+      "learning_rate": 9.864500307487751e-06,
+      "loss": 1.3915,
+      "mean_token_accuracy": 0.629261365532875,
+      "num_tokens": 17857549.0,
+      "step": 2500
+    },
+    {
+      "epoch": 0.512506380806534,
+      "grad_norm": 13.635055541992188,
+      "learning_rate": 9.86296675118668e-06,
+      "loss": 1.4043,
+      "mean_token_accuracy": 0.6366223454475403,
+      "num_tokens": 17927963.0,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5145482388973966,
+      "grad_norm": 13.6681547164917,
+      "learning_rate": 9.861424685979054e-06,
+      "loss": 1.439,
+      "mean_token_accuracy": 0.6203552514314652,
+      "num_tokens": 18001509.0,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5165900969882593,
+      "grad_norm": 12.835921287536621,
+      "learning_rate": 9.859874114563075e-06,
+      "loss": 1.3936,
+      "mean_token_accuracy": 0.6366373240947724,
+      "num_tokens": 18072963.0,
+      "step": 2530
+    },
+    {
+      "epoch": 0.518631955079122,
+      "grad_norm": 13.316462516784668,
+      "learning_rate": 9.858315039651836e-06,
+      "loss": 1.3831,
+      "mean_token_accuracy": 0.6388492465019227,
+      "num_tokens": 18142046.0,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5206738131699847,
+      "grad_norm": 12.890396118164062,
+      "learning_rate": 9.856747463973303e-06,
+      "loss": 1.5356,
+      "mean_token_accuracy": 0.620140865445137,
+      "num_tokens": 18212201.0,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5227156712608474,
+      "grad_norm": 12.411418914794922,
+      "learning_rate": 9.855171390270325e-06,
+      "loss": 1.3723,
+      "mean_token_accuracy": 0.6438209116458893,
+      "num_tokens": 18283060.0,
+      "step": 2560
+    },
+    {
+      "epoch": 0.52475752935171,
+      "grad_norm": 13.509264945983887,
+      "learning_rate": 9.853586821300608e-06,
+      "loss": 1.429,
+      "mean_token_accuracy": 0.6353526145219803,
+      "num_tokens": 18356527.0,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5267993874425727,
+      "grad_norm": 13.333746910095215,
+      "learning_rate": 9.851993759836734e-06,
+      "loss": 1.4291,
+      "mean_token_accuracy": 0.6351711452007294,
+      "num_tokens": 18426397.0,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5288412455334355,
+      "grad_norm": 13.60482120513916,
+      "learning_rate": 9.850392208666134e-06,
+      "loss": 1.3642,
+      "mean_token_accuracy": 0.6275977313518524,
+      "num_tokens": 18500416.0,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5308831036242981,
+      "grad_norm": 13.40274429321289,
+      "learning_rate": 9.848782170591103e-06,
+      "loss": 1.4108,
+      "mean_token_accuracy": 0.6340374439954758,
+      "num_tokens": 18573354.0,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5329249617151608,
+      "grad_norm": 13.70960521697998,
+      "learning_rate": 9.847163648428783e-06,
+      "loss": 1.3876,
+      "mean_token_accuracy": 0.6381645023822784,
+      "num_tokens": 18642953.0,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5349668198060235,
+      "grad_norm": 13.797345161437988,
+      "learning_rate": 9.845536645011157e-06,
+      "loss": 1.3801,
+      "mean_token_accuracy": 0.6419808328151703,
+      "num_tokens": 18719168.0,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5370086778968862,
+      "grad_norm": 13.198974609375,
+      "learning_rate": 9.843901163185055e-06,
+      "loss": 1.4107,
+      "mean_token_accuracy": 0.6292454957962036,
+      "num_tokens": 18791570.0,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5390505359877489,
+      "grad_norm": 13.272690773010254,
+      "learning_rate": 9.842257205812133e-06,
+      "loss": 1.3883,
+      "mean_token_accuracy": 0.642322301864624,
+      "num_tokens": 18862900.0,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5410923940786115,
+      "grad_norm": 12.19058895111084,
+      "learning_rate": 9.840604775768885e-06,
+      "loss": 1.371,
+      "mean_token_accuracy": 0.6479323506355286,
+      "num_tokens": 18932883.0,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5431342521694742,
+      "grad_norm": 13.2040376663208,
+      "learning_rate": 9.838943875946628e-06,
+      "loss": 1.4196,
+      "mean_token_accuracy": 0.6349193304777145,
+      "num_tokens": 19004875.0,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5451761102603369,
+      "grad_norm": 13.854543685913086,
+      "learning_rate": 9.837274509251494e-06,
+      "loss": 1.4093,
+      "mean_token_accuracy": 0.6391417652368545,
+      "num_tokens": 19076124.0,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5472179683511996,
+      "grad_norm": 13.256423950195312,
+      "learning_rate": 9.835596678604435e-06,
+      "loss": 1.3879,
+      "mean_token_accuracy": 0.6399724900722503,
+      "num_tokens": 19149584.0,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5492598264420623,
+      "grad_norm": 13.840933799743652,
+      "learning_rate": 9.833910386941213e-06,
+      "loss": 1.433,
+      "mean_token_accuracy": 0.6246784031391144,
+      "num_tokens": 19221353.0,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5513016845329249,
+      "grad_norm": 12.89402961730957,
+      "learning_rate": 9.83221563721239e-06,
+      "loss": 1.443,
+      "mean_token_accuracy": 0.635868039727211,
+      "num_tokens": 19295418.0,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5533435426237876,
+      "grad_norm": 12.286266326904297,
+      "learning_rate": 9.830512432383333e-06,
+      "loss": 1.4191,
+      "mean_token_accuracy": 0.6380386412143707,
+      "num_tokens": 19370379.0,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5553854007146504,
+      "grad_norm": 14.805827140808105,
+      "learning_rate": 9.828800775434197e-06,
+      "loss": 1.3819,
+      "mean_token_accuracy": 0.6385849773883819,
+      "num_tokens": 19441372.0,
+      "step": 2720
+    },
+    {
+      "epoch": 0.557427258805513,
+      "grad_norm": 13.304326057434082,
+      "learning_rate": 9.827080669359933e-06,
+      "loss": 1.4274,
+      "mean_token_accuracy": 0.6263547629117966,
+      "num_tokens": 19515686.0,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5594691168963757,
+      "grad_norm": 13.437885284423828,
+      "learning_rate": 9.825352117170269e-06,
+      "loss": 1.3906,
+      "mean_token_accuracy": 0.648447448015213,
+      "num_tokens": 19584899.0,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5615109749872383,
+      "grad_norm": 12.786911964416504,
+      "learning_rate": 9.823615121889716e-06,
+      "loss": 1.3985,
+      "mean_token_accuracy": 0.6387988567352295,
+      "num_tokens": 19657466.0,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5635528330781011,
+      "grad_norm": 13.658928871154785,
+      "learning_rate": 9.821869686557555e-06,
+      "loss": 1.3936,
+      "mean_token_accuracy": 0.6408439576625824,
+      "num_tokens": 19725251.0,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5655946911689638,
+      "grad_norm": 12.969337463378906,
+      "learning_rate": 9.82011581422784e-06,
+      "loss": 1.431,
+      "mean_token_accuracy": 0.628329211473465,
+      "num_tokens": 19793131.0,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5676365492598264,
+      "grad_norm": 12.977407455444336,
+      "learning_rate": 9.818353507969379e-06,
+      "loss": 1.4193,
+      "mean_token_accuracy": 0.6275966912508011,
+      "num_tokens": 19869081.0,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5696784073506891,
+      "grad_norm": 12.836411476135254,
+      "learning_rate": 9.816582770865747e-06,
+      "loss": 1.3117,
+      "mean_token_accuracy": 0.6492154330015183,
+      "num_tokens": 19939561.0,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5717202654415519,
+      "grad_norm": 13.496788024902344,
+      "learning_rate": 9.814803606015263e-06,
+      "loss": 1.394,
+      "mean_token_accuracy": 0.6368442118167877,
+      "num_tokens": 20008753.0,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5737621235324145,
+      "grad_norm": 14.3073148727417,
+      "learning_rate": 9.813016016530996e-06,
+      "loss": 1.4656,
+      "mean_token_accuracy": 0.6255601584911347,
+      "num_tokens": 20079759.0,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5758039816232772,
+      "grad_norm": 13.646188735961914,
+      "learning_rate": 9.811220005540756e-06,
+      "loss": 1.4906,
+      "mean_token_accuracy": 0.6181686758995056,
+      "num_tokens": 20150221.0,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5778458397141398,
+      "grad_norm": 14.640626907348633,
+      "learning_rate": 9.809415576187087e-06,
+      "loss": 1.4371,
+      "mean_token_accuracy": 0.6324725329875946,
+      "num_tokens": 20221920.0,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5798876978050026,
+      "grad_norm": 15.234823226928711,
+      "learning_rate": 9.807602731627262e-06,
+      "loss": 1.4532,
+      "mean_token_accuracy": 0.6349549174308777,
+      "num_tokens": 20292653.0,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5819295558958653,
+      "grad_norm": 13.222195625305176,
+      "learning_rate": 9.805781475033282e-06,
+      "loss": 1.4338,
+      "mean_token_accuracy": 0.6166985005140304,
+      "num_tokens": 20366340.0,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5839714139867279,
+      "grad_norm": 14.15178394317627,
+      "learning_rate": 9.803951809591866e-06,
+      "loss": 1.4232,
+      "mean_token_accuracy": 0.6322162270545959,
+      "num_tokens": 20432957.0,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5860132720775906,
+      "grad_norm": 12.758091926574707,
+      "learning_rate": 9.80211373850444e-06,
+      "loss": 1.4142,
+      "mean_token_accuracy": 0.6438927322626113,
+      "num_tokens": 20505606.0,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5880551301684533,
+      "grad_norm": 13.361724853515625,
+      "learning_rate": 9.800267264987147e-06,
+      "loss": 1.3776,
+      "mean_token_accuracy": 0.6391714990139008,
+      "num_tokens": 20579466.0,
+      "step": 2880
+    },
+    {
+      "epoch": 0.590096988259316,
+      "grad_norm": 13.651033401489258,
+      "learning_rate": 9.798412392270825e-06,
+      "loss": 1.3711,
+      "mean_token_accuracy": 0.6416450500488281,
+      "num_tokens": 20652379.0,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5921388463501787,
+      "grad_norm": 14.636039733886719,
+      "learning_rate": 9.796549123601012e-06,
+      "loss": 1.3782,
+      "mean_token_accuracy": 0.6453828096389771,
+      "num_tokens": 20722216.0,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5941807044410413,
+      "grad_norm": 13.31446361541748,
+      "learning_rate": 9.794677462237936e-06,
+      "loss": 1.4271,
+      "mean_token_accuracy": 0.6281617492437362,
+      "num_tokens": 20794125.0,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5962225625319041,
+      "grad_norm": 14.600592613220215,
+      "learning_rate": 9.792797411456509e-06,
+      "loss": 1.476,
+      "mean_token_accuracy": 0.6172799646854401,
+      "num_tokens": 20863211.0,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5982644206227667,
+      "grad_norm": 14.365285873413086,
+      "learning_rate": 9.790908974546326e-06,
+      "loss": 1.4591,
+      "mean_token_accuracy": 0.631415992975235,
+      "num_tokens": 20933317.0,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6003062787136294,
+      "grad_norm": 13.793408393859863,
+      "learning_rate": 9.789012154811648e-06,
+      "loss": 1.4121,
+      "mean_token_accuracy": 0.6378588736057281,
+      "num_tokens": 21002698.0,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6023481368044921,
+      "grad_norm": 12.982183456420898,
+      "learning_rate": 9.78710695557141e-06,
+      "loss": 1.3554,
+      "mean_token_accuracy": 0.6380349576473237,
+      "num_tokens": 21074284.0,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6043899948953547,
+      "grad_norm": 14.370036125183105,
+      "learning_rate": 9.785193380159208e-06,
+      "loss": 1.4423,
+      "mean_token_accuracy": 0.6347444444894791,
+      "num_tokens": 21148349.0,
+      "step": 2960
+    },
+    {
+      "epoch": 0.6064318529862175,
+      "grad_norm": 12.563482284545898,
+      "learning_rate": 9.783271431923293e-06,
+      "loss": 1.3354,
+      "mean_token_accuracy": 0.6468933254480362,
+      "num_tokens": 21219321.0,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6084737110770801,
+      "grad_norm": 15.911494255065918,
+      "learning_rate": 9.781341114226565e-06,
+      "loss": 1.426,
+      "mean_token_accuracy": 0.6219086617231369,
+      "num_tokens": 21291472.0,
+      "step": 2980
+    },
+    {
+      "epoch": 0.6105155691679428,
+      "grad_norm": 14.266389846801758,
+      "learning_rate": 9.779402430446572e-06,
+      "loss": 1.3806,
+      "mean_token_accuracy": 0.6449051052331924,
+      "num_tokens": 21363629.0,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6125574272588055,
+      "grad_norm": 12.48147964477539,
+      "learning_rate": 9.777455383975496e-06,
+      "loss": 1.4112,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6125574272588055,
+      "eval_loss": 1.440710425376892,
+      "eval_mean_token_accuracy": 0.6296637463097048,
+      "eval_num_tokens": 21437325.0,
+      "eval_runtime": 991.1208,
+      "eval_samples_per_second": 4.988,
+      "eval_steps_per_second": 1.247,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6145992853496682,
+      "grad_norm": 14.174776077270508,
+      "learning_rate": 9.775499978220155e-06,
+      "loss": 1.4097,
+      "mean_token_accuracy": 0.6289676755666733,
+      "num_tokens": 21512874.0,
+      "step": 3010
+    },
+    {
+      "epoch": 0.6166411434405309,
+      "grad_norm": 14.071956634521484,
+      "learning_rate": 9.773536216601994e-06,
+      "loss": 1.4198,
+      "mean_token_accuracy": 0.6249706447124481,
+      "num_tokens": 21586472.0,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6186830015313936,
+      "grad_norm": 12.335860252380371,
+      "learning_rate": 9.771564102557072e-06,
+      "loss": 1.4183,
+      "mean_token_accuracy": 0.632024022936821,
+      "num_tokens": 21657572.0,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6207248596222562,
+      "grad_norm": 13.708938598632812,
+      "learning_rate": 9.769583639536071e-06,
+      "loss": 1.4158,
+      "mean_token_accuracy": 0.6340501457452774,
+      "num_tokens": 21729103.0,
+      "step": 3040
+    },
+    {
+      "epoch": 0.622766717713119,
+      "grad_norm": 12.857871055603027,
+      "learning_rate": 9.767594831004277e-06,
+      "loss": 1.4024,
+      "mean_token_accuracy": 0.6416067212820054,
+      "num_tokens": 21805653.0,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6248085758039816,
+      "grad_norm": 15.368263244628906,
+      "learning_rate": 9.76559768044158e-06,
+      "loss": 1.3593,
+      "mean_token_accuracy": 0.6511017084121704,
+      "num_tokens": 21878088.0,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6268504338948443,
+      "grad_norm": 13.683640480041504,
+      "learning_rate": 9.763592191342467e-06,
+      "loss": 1.4604,
+      "mean_token_accuracy": 0.6294127464294433,
+      "num_tokens": 21947896.0,
+      "step": 3070
+    },
+    {
+      "epoch": 0.628892291985707,
+      "grad_norm": 12.823150634765625,
+      "learning_rate": 9.761578367216008e-06,
+      "loss": 1.4111,
+      "mean_token_accuracy": 0.6234809100627899,
+      "num_tokens": 22020392.0,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6309341500765697,
+      "grad_norm": 14.795945167541504,
+      "learning_rate": 9.75955621158587e-06,
+      "loss": 1.3956,
+      "mean_token_accuracy": 0.6330976217985154,
+      "num_tokens": 22088766.0,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6329760081674324,
+      "grad_norm": 13.43422794342041,
+      "learning_rate": 9.757525727990286e-06,
+      "loss": 1.4851,
+      "mean_token_accuracy": 0.6163936108350754,
+      "num_tokens": 22158579.0,
+      "step": 3100
+    },
+    {
+      "epoch": 0.635017866258295,
+      "grad_norm": 11.95371150970459,
+      "learning_rate": 9.755486919982067e-06,
+      "loss": 1.4307,
+      "mean_token_accuracy": 0.6282651394605636,
+      "num_tokens": 22226235.0,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6370597243491577,
+      "grad_norm": 12.942562103271484,
+      "learning_rate": 9.75343979112859e-06,
+      "loss": 1.3729,
+      "mean_token_accuracy": 0.640126034617424,
+      "num_tokens": 22295629.0,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6391015824400205,
+      "grad_norm": 15.593914031982422,
+      "learning_rate": 9.751384345011787e-06,
+      "loss": 1.442,
+      "mean_token_accuracy": 0.6314625293016434,
+      "num_tokens": 22365531.0,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6411434405308831,
+      "grad_norm": 14.191858291625977,
+      "learning_rate": 9.749320585228148e-06,
+      "loss": 1.4015,
+      "mean_token_accuracy": 0.6351157635450363,
+      "num_tokens": 22433566.0,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6431852986217458,
+      "grad_norm": 15.005651473999023,
+      "learning_rate": 9.747248515388705e-06,
+      "loss": 1.3679,
+      "mean_token_accuracy": 0.646827119588852,
+      "num_tokens": 22505167.0,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6452271567126084,
+      "grad_norm": 13.515469551086426,
+      "learning_rate": 9.745168139119033e-06,
+      "loss": 1.4565,
+      "mean_token_accuracy": 0.6296883016824723,
+      "num_tokens": 22574602.0,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6472690148034712,
+      "grad_norm": 12.68270206451416,
+      "learning_rate": 9.743079460059241e-06,
+      "loss": 1.3882,
+      "mean_token_accuracy": 0.6341635257005691,
+      "num_tokens": 22646932.0,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6493108728943339,
+      "grad_norm": 14.113992691040039,
+      "learning_rate": 9.740982481863967e-06,
+      "loss": 1.473,
+      "mean_token_accuracy": 0.6364928364753724,
+      "num_tokens": 22718533.0,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6513527309851965,
+      "grad_norm": 11.577704429626465,
+      "learning_rate": 9.738877208202363e-06,
+      "loss": 1.4147,
+      "mean_token_accuracy": 0.6371136277914047,
+      "num_tokens": 22790566.0,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6533945890760592,
+      "grad_norm": 12.993800163269043,
+      "learning_rate": 9.73676364275811e-06,
+      "loss": 1.3988,
+      "mean_token_accuracy": 0.636861988902092,
+      "num_tokens": 22853994.0,
+      "step": 3200
+    },
+    {
+      "epoch": 0.655436447166922,
+      "grad_norm": 12.155184745788574,
+      "learning_rate": 9.734641789229381e-06,
+      "loss": 1.3366,
+      "mean_token_accuracy": 0.6418772786855698,
+      "num_tokens": 22925430.0,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6574783052577846,
+      "grad_norm": 15.805740356445312,
+      "learning_rate": 9.732511651328864e-06,
+      "loss": 1.3588,
+      "mean_token_accuracy": 0.6364259988069534,
+      "num_tokens": 22997748.0,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6595201633486473,
+      "grad_norm": 13.210399627685547,
+      "learning_rate": 9.730373232783735e-06,
+      "loss": 1.3825,
+      "mean_token_accuracy": 0.6286744445562362,
+      "num_tokens": 23069478.0,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6615620214395099,
+      "grad_norm": 13.883077621459961,
+      "learning_rate": 9.72822653733566e-06,
+      "loss": 1.3418,
+      "mean_token_accuracy": 0.6431337684392929,
+      "num_tokens": 23137148.0,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6636038795303726,
+      "grad_norm": 12.698500633239746,
+      "learning_rate": 9.726071568740793e-06,
+      "loss": 1.4307,
+      "mean_token_accuracy": 0.634193229675293,
+      "num_tokens": 23212626.0,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6656457376212354,
+      "grad_norm": 12.851318359375,
+      "learning_rate": 9.723908330769756e-06,
+      "loss": 1.3716,
+      "mean_token_accuracy": 0.6405917227268219,
+      "num_tokens": 23284143.0,
+      "step": 3260
+    },
+    {
+      "epoch": 0.667687595712098,
+      "grad_norm": 13.096170425415039,
+      "learning_rate": 9.721736827207644e-06,
+      "loss": 1.3898,
+      "mean_token_accuracy": 0.6286759912967682,
+      "num_tokens": 23357005.0,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6697294538029607,
+      "grad_norm": 12.506603240966797,
+      "learning_rate": 9.719557061854017e-06,
+      "loss": 1.3495,
+      "mean_token_accuracy": 0.6530344665050507,
+      "num_tokens": 23431394.0,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6717713118938233,
+      "grad_norm": 13.819788932800293,
+      "learning_rate": 9.717369038522884e-06,
+      "loss": 1.3564,
+      "mean_token_accuracy": 0.6386108338832855,
+      "num_tokens": 23504003.0,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6738131699846861,
+      "grad_norm": 13.408661842346191,
+      "learning_rate": 9.715172761042711e-06,
+      "loss": 1.3723,
+      "mean_token_accuracy": 0.628694823384285,
+      "num_tokens": 23577951.0,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6758550280755488,
+      "grad_norm": 13.23100757598877,
+      "learning_rate": 9.712968233256401e-06,
+      "loss": 1.4124,
+      "mean_token_accuracy": 0.6347943753004074,
+      "num_tokens": 23647422.0,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6778968861664114,
+      "grad_norm": 14.13464641571045,
+      "learning_rate": 9.710755459021297e-06,
+      "loss": 1.3851,
+      "mean_token_accuracy": 0.6412581115961075,
+      "num_tokens": 23719685.0,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6799387442572741,
+      "grad_norm": 13.437444686889648,
+      "learning_rate": 9.708534442209168e-06,
+      "loss": 1.3589,
+      "mean_token_accuracy": 0.6525840878486633,
+      "num_tokens": 23792553.0,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6819806023481368,
+      "grad_norm": 14.053814888000488,
+      "learning_rate": 9.706305186706207e-06,
+      "loss": 1.4502,
+      "mean_token_accuracy": 0.626780903339386,
+      "num_tokens": 23866377.0,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6840224604389995,
+      "grad_norm": 13.260232925415039,
+      "learning_rate": 9.70406769641302e-06,
+      "loss": 1.3332,
+      "mean_token_accuracy": 0.6482997745275497,
+      "num_tokens": 23936488.0,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6860643185298622,
+      "grad_norm": 13.664006233215332,
+      "learning_rate": 9.701821975244625e-06,
+      "loss": 1.4597,
+      "mean_token_accuracy": 0.6262546479701996,
+      "num_tokens": 24009134.0,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6881061766207248,
+      "grad_norm": 13.392569541931152,
+      "learning_rate": 9.699568027130442e-06,
+      "loss": 1.3125,
+      "mean_token_accuracy": 0.6487543761730195,
+      "num_tokens": 24080618.0,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6901480347115876,
+      "grad_norm": 12.633986473083496,
+      "learning_rate": 9.697305856014281e-06,
+      "loss": 1.3968,
+      "mean_token_accuracy": 0.6403874427080154,
+      "num_tokens": 24154567.0,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6921898928024502,
+      "grad_norm": 14.033143997192383,
+      "learning_rate": 9.695035465854348e-06,
+      "loss": 1.4009,
+      "mean_token_accuracy": 0.6387769609689713,
+      "num_tokens": 24221960.0,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6942317508933129,
+      "grad_norm": 13.039753913879395,
+      "learning_rate": 9.692756860623221e-06,
+      "loss": 1.3506,
+      "mean_token_accuracy": 0.6354148983955383,
+      "num_tokens": 24292743.0,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6962736089841756,
+      "grad_norm": 11.965189933776855,
+      "learning_rate": 9.690470044307861e-06,
+      "loss": 1.4325,
+      "mean_token_accuracy": 0.6360249876976013,
+      "num_tokens": 24362477.0,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6983154670750383,
+      "grad_norm": 12.434362411499023,
+      "learning_rate": 9.688175020909589e-06,
+      "loss": 1.3562,
+      "mean_token_accuracy": 0.6378575742244721,
+      "num_tokens": 24432920.0,
+      "step": 3420
+    },
+    {
+      "epoch": 0.700357325165901,
+      "grad_norm": 12.607736587524414,
+      "learning_rate": 9.685871794444092e-06,
+      "loss": 1.3255,
+      "mean_token_accuracy": 0.646689772605896,
+      "num_tokens": 24503177.0,
+      "step": 3430
+    },
+    {
+      "epoch": 0.7023991832567636,
+      "grad_norm": 12.687234878540039,
+      "learning_rate": 9.683560368941406e-06,
+      "loss": 1.3244,
+      "mean_token_accuracy": 0.6440610259771347,
+      "num_tokens": 24573187.0,
+      "step": 3440
+    },
+    {
+      "epoch": 0.7044410413476263,
+      "grad_norm": 13.45374584197998,
+      "learning_rate": 9.681240748445914e-06,
+      "loss": 1.3499,
+      "mean_token_accuracy": 0.6462150484323501,
+      "num_tokens": 24646749.0,
+      "step": 3450
+    },
+    {
+      "epoch": 0.7064828994384891,
+      "grad_norm": 13.070556640625,
+      "learning_rate": 9.678912937016343e-06,
+      "loss": 1.4068,
+      "mean_token_accuracy": 0.6340273886919021,
+      "num_tokens": 24716266.0,
+      "step": 3460
+    },
+    {
+      "epoch": 0.7085247575293517,
+      "grad_norm": 13.016335487365723,
+      "learning_rate": 9.676576938725742e-06,
+      "loss": 1.4276,
+      "mean_token_accuracy": 0.6312612593173981,
+      "num_tokens": 24782591.0,
+      "step": 3470
+    },
+    {
+      "epoch": 0.7105666156202144,
+      "grad_norm": 12.000085830688477,
+      "learning_rate": 9.674232757661495e-06,
+      "loss": 1.4557,
+      "mean_token_accuracy": 0.6277955800294877,
+      "num_tokens": 24855167.0,
+      "step": 3480
+    },
+    {
+      "epoch": 0.712608473711077,
+      "grad_norm": 13.089092254638672,
+      "learning_rate": 9.671880397925296e-06,
+      "loss": 1.3836,
+      "mean_token_accuracy": 0.644837275147438,
+      "num_tokens": 24929959.0,
+      "step": 3490
+    },
+    {
+      "epoch": 0.7146503318019398,
+      "grad_norm": 12.519915580749512,
+      "learning_rate": 9.669519863633155e-06,
+      "loss": 1.4579,
+      "mean_token_accuracy": 0.6203823953866958,
+      "num_tokens": 25001204.0,
+      "step": 3500
+    },
+    {
+      "epoch": 0.7166921898928025,
+      "grad_norm": 13.97362232208252,
+      "learning_rate": 9.667151158915382e-06,
+      "loss": 1.3887,
+      "mean_token_accuracy": 0.6337443590164185,
+      "num_tokens": 25069613.0,
+      "step": 3510
+    },
+    {
+      "epoch": 0.7187340479836651,
+      "grad_norm": 14.413997650146484,
+      "learning_rate": 9.664774287916587e-06,
+      "loss": 1.4879,
+      "mean_token_accuracy": 0.6237174719572067,
+      "num_tokens": 25140521.0,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7207759060745278,
+      "grad_norm": 14.139748573303223,
+      "learning_rate": 9.662389254795661e-06,
+      "loss": 1.3745,
+      "mean_token_accuracy": 0.6430779010057449,
+      "num_tokens": 25213201.0,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7228177641653905,
+      "grad_norm": 13.43658447265625,
+      "learning_rate": 9.659996063725787e-06,
+      "loss": 1.3662,
+      "mean_token_accuracy": 0.6530295521020889,
+      "num_tokens": 25284506.0,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7248596222562532,
+      "grad_norm": 13.512331008911133,
+      "learning_rate": 9.657594718894414e-06,
+      "loss": 1.4379,
+      "mean_token_accuracy": 0.6244717538356781,
+      "num_tokens": 25355630.0,
+      "step": 3550
+    },
+    {
+      "epoch": 0.7269014803471159,
+      "grad_norm": 12.112098693847656,
+      "learning_rate": 9.65518522450326e-06,
+      "loss": 1.4099,
+      "mean_token_accuracy": 0.6363851726055145,
+      "num_tokens": 25429960.0,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7289433384379785,
+      "grad_norm": 12.75960922241211,
+      "learning_rate": 9.652767584768304e-06,
+      "loss": 1.4137,
+      "mean_token_accuracy": 0.6329428285360337,
+      "num_tokens": 25501298.0,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7309851965288412,
+      "grad_norm": 13.585481643676758,
+      "learning_rate": 9.650341803919779e-06,
+      "loss": 1.3472,
+      "mean_token_accuracy": 0.6461682111024857,
+      "num_tokens": 25573820.0,
+      "step": 3580
+    },
+    {
+      "epoch": 0.733027054619704,
+      "grad_norm": 12.37596607208252,
+      "learning_rate": 9.647907886202157e-06,
+      "loss": 1.3985,
+      "mean_token_accuracy": 0.6342933237552643,
+      "num_tokens": 25645428.0,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7350689127105666,
+      "grad_norm": 13.346282958984375,
+      "learning_rate": 9.645465835874152e-06,
+      "loss": 1.3366,
+      "mean_token_accuracy": 0.6504071831703186,
+      "num_tokens": 25717166.0,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7371107708014293,
+      "grad_norm": 13.590423583984375,
+      "learning_rate": 9.643015657208708e-06,
+      "loss": 1.4408,
+      "mean_token_accuracy": 0.6356685668230057,
+      "num_tokens": 25786912.0,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7391526288922919,
+      "grad_norm": 13.47347640991211,
+      "learning_rate": 9.640557354492988e-06,
+      "loss": 1.4635,
+      "mean_token_accuracy": 0.6147109150886536,
+      "num_tokens": 25856128.0,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7411944869831547,
+      "grad_norm": 13.709653854370117,
+      "learning_rate": 9.638090932028373e-06,
+      "loss": 1.3778,
+      "mean_token_accuracy": 0.6359743535518646,
+      "num_tokens": 25927390.0,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7432363450740174,
+      "grad_norm": 11.859482765197754,
+      "learning_rate": 9.63561639413045e-06,
+      "loss": 1.3549,
+      "mean_token_accuracy": 0.6486242324113846,
+      "num_tokens": 26002907.0,
+      "step": 3640
+    },
+    {
+      "epoch": 0.74527820316488,
+      "grad_norm": 12.219711303710938,
+      "learning_rate": 9.633133745129008e-06,
+      "loss": 1.3967,
+      "mean_token_accuracy": 0.6251788079738617,
+      "num_tokens": 26073293.0,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7473200612557427,
+      "grad_norm": 13.043457984924316,
+      "learning_rate": 9.630642989368024e-06,
+      "loss": 1.4081,
+      "mean_token_accuracy": 0.6368646204471589,
+      "num_tokens": 26145821.0,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7493619193466055,
+      "grad_norm": 14.711187362670898,
+      "learning_rate": 9.628144131205665e-06,
+      "loss": 1.446,
+      "mean_token_accuracy": 0.6285335540771484,
+      "num_tokens": 26213680.0,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7514037774374681,
+      "grad_norm": 12.913311958312988,
+      "learning_rate": 9.62563717501427e-06,
+      "loss": 1.4466,
+      "mean_token_accuracy": 0.6235233783721924,
+      "num_tokens": 26284742.0,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7534456355283308,
+      "grad_norm": 12.046110153198242,
+      "learning_rate": 9.623122125180351e-06,
+      "loss": 1.4176,
+      "mean_token_accuracy": 0.643735259771347,
+      "num_tokens": 26354716.0,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7554874936191934,
+      "grad_norm": 13.422168731689453,
+      "learning_rate": 9.620598986104578e-06,
+      "loss": 1.4649,
+      "mean_token_accuracy": 0.6326315462589264,
+      "num_tokens": 26426125.0,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7575293517100562,
+      "grad_norm": 12.690073013305664,
+      "learning_rate": 9.618067762201784e-06,
+      "loss": 1.4224,
+      "mean_token_accuracy": 0.6294170558452606,
+      "num_tokens": 26497872.0,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7595712098009189,
+      "grad_norm": 12.08823013305664,
+      "learning_rate": 9.615528457900934e-06,
+      "loss": 1.3856,
+      "mean_token_accuracy": 0.6507549166679383,
+      "num_tokens": 26568238.0,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7616130678917815,
+      "grad_norm": 14.470535278320312,
+      "learning_rate": 9.612981077645144e-06,
+      "loss": 1.3912,
+      "mean_token_accuracy": 0.6367805600166321,
+      "num_tokens": 26635873.0,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7636549259826442,
+      "grad_norm": 13.872713088989258,
+      "learning_rate": 9.610425625891656e-06,
+      "loss": 1.4196,
+      "mean_token_accuracy": 0.6281679064035416,
+      "num_tokens": 26708209.0,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7656967840735069,
+      "grad_norm": 13.715387344360352,
+      "learning_rate": 9.607862107111834e-06,
+      "loss": 1.373,
+      "mean_token_accuracy": 0.6392459809780121,
+      "num_tokens": 26781659.0,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7677386421643696,
+      "grad_norm": 13.774560928344727,
+      "learning_rate": 9.605290525791157e-06,
+      "loss": 1.3897,
+      "mean_token_accuracy": 0.6378937274217605,
+      "num_tokens": 26853155.0,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7697805002552323,
+      "grad_norm": 12.794147491455078,
+      "learning_rate": 9.602710886429218e-06,
+      "loss": 1.3528,
+      "mean_token_accuracy": 0.6442150235176086,
+      "num_tokens": 26920655.0,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7718223583460949,
+      "grad_norm": 12.719574928283691,
+      "learning_rate": 9.600123193539698e-06,
+      "loss": 1.3502,
+      "mean_token_accuracy": 0.6572590112686157,
+      "num_tokens": 26992809.0,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7738642164369577,
+      "grad_norm": 13.51803970336914,
+      "learning_rate": 9.597527451650376e-06,
+      "loss": 1.3243,
+      "mean_token_accuracy": 0.6524654120206833,
+      "num_tokens": 27066434.0,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7759060745278203,
+      "grad_norm": 14.247166633605957,
+      "learning_rate": 9.59492366530312e-06,
+      "loss": 1.4145,
+      "mean_token_accuracy": 0.6389972150325776,
+      "num_tokens": 27139751.0,
+      "step": 3800
+    },
+    {
+      "epoch": 0.777947932618683,
+      "grad_norm": 13.419672012329102,
+      "learning_rate": 9.592311839053863e-06,
+      "loss": 1.3721,
+      "mean_token_accuracy": 0.6519317954778672,
+      "num_tokens": 27213247.0,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7799897907095457,
+      "grad_norm": 13.287151336669922,
+      "learning_rate": 9.589691977472612e-06,
+      "loss": 1.3991,
+      "mean_token_accuracy": 0.6346599578857421,
+      "num_tokens": 27288707.0,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7820316488004083,
+      "grad_norm": 12.613758087158203,
+      "learning_rate": 9.587064085143432e-06,
+      "loss": 1.4093,
+      "mean_token_accuracy": 0.6332722306251526,
+      "num_tokens": 27364833.0,
+      "step": 3830
+    },
+    {
+      "epoch": 0.7840735068912711,
+      "grad_norm": 13.21430778503418,
+      "learning_rate": 9.584428166664441e-06,
+      "loss": 1.3291,
+      "mean_token_accuracy": 0.6426202535629273,
+      "num_tokens": 27439232.0,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7861153649821337,
+      "grad_norm": 12.770293235778809,
+      "learning_rate": 9.581784226647799e-06,
+      "loss": 1.422,
+      "mean_token_accuracy": 0.6253783494234085,
+      "num_tokens": 27508261.0,
+      "step": 3850
+    },
+    {
+      "epoch": 0.7881572230729964,
+      "grad_norm": 14.6603422164917,
+      "learning_rate": 9.5791322697197e-06,
+      "loss": 1.3018,
+      "mean_token_accuracy": 0.6558138817548752,
+      "num_tokens": 27579262.0,
+      "step": 3860
+    },
+    {
+      "epoch": 0.7901990811638591,
+      "grad_norm": 13.182867050170898,
+      "learning_rate": 9.576472300520372e-06,
+      "loss": 1.3363,
+      "mean_token_accuracy": 0.6500471413135529,
+      "num_tokens": 27654478.0,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7922409392547218,
+      "grad_norm": 13.26388931274414,
+      "learning_rate": 9.573804323704057e-06,
+      "loss": 1.3474,
+      "mean_token_accuracy": 0.6410001933574676,
+      "num_tokens": 27729450.0,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7942827973455845,
+      "grad_norm": 13.69534969329834,
+      "learning_rate": 9.571128343939006e-06,
+      "loss": 1.407,
+      "mean_token_accuracy": 0.6288270473480224,
+      "num_tokens": 27800068.0,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7963246554364471,
+      "grad_norm": 12.891340255737305,
+      "learning_rate": 9.56844436590748e-06,
+      "loss": 1.2993,
+      "mean_token_accuracy": 0.6599480360746384,
+      "num_tokens": 27872575.0,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7983665135273098,
+      "grad_norm": 13.020922660827637,
+      "learning_rate": 9.565752394305729e-06,
+      "loss": 1.4082,
+      "mean_token_accuracy": 0.6357982963323593,
+      "num_tokens": 27944873.0,
+      "step": 3910
+    },
+    {
+      "epoch": 0.8004083716181726,
+      "grad_norm": 13.040590286254883,
+      "learning_rate": 9.563052433843993e-06,
+      "loss": 1.3802,
+      "mean_token_accuracy": 0.6361476242542267,
+      "num_tokens": 28016338.0,
+      "step": 3920
+    },
+    {
+      "epoch": 0.8024502297090352,
+      "grad_norm": 13.284208297729492,
+      "learning_rate": 9.560344489246489e-06,
+      "loss": 1.409,
+      "mean_token_accuracy": 0.6426727443933486,
+      "num_tokens": 28088927.0,
+      "step": 3930
+    },
+    {
+      "epoch": 0.8044920877998979,
+      "grad_norm": 13.533629417419434,
+      "learning_rate": 9.557628565251404e-06,
+      "loss": 1.3227,
+      "mean_token_accuracy": 0.6476307839155198,
+      "num_tokens": 28159259.0,
+      "step": 3940
+    },
+    {
+      "epoch": 0.8065339458907606,
+      "grad_norm": 16.405038833618164,
+      "learning_rate": 9.554904666610889e-06,
+      "loss": 1.4397,
+      "mean_token_accuracy": 0.6269133925437927,
+      "num_tokens": 28231271.0,
+      "step": 3950
+    },
+    {
+      "epoch": 0.8085758039816233,
+      "grad_norm": 13.900541305541992,
+      "learning_rate": 9.552172798091045e-06,
+      "loss": 1.4283,
+      "mean_token_accuracy": 0.6322760939598083,
+      "num_tokens": 28301168.0,
+      "step": 3960
+    },
+    {
+      "epoch": 0.810617662072486,
+      "grad_norm": 13.871532440185547,
+      "learning_rate": 9.549432964471919e-06,
+      "loss": 1.4501,
+      "mean_token_accuracy": 0.6328884690999985,
+      "num_tokens": 28379814.0,
+      "step": 3970
+    },
+    {
+      "epoch": 0.8126595201633486,
+      "grad_norm": 12.453668594360352,
+      "learning_rate": 9.546685170547496e-06,
+      "loss": 1.3755,
+      "mean_token_accuracy": 0.6485854834318161,
+      "num_tokens": 28453322.0,
+      "step": 3980
+    },
+    {
+      "epoch": 0.8147013782542113,
+      "grad_norm": 12.559453010559082,
+      "learning_rate": 9.54392942112569e-06,
+      "loss": 1.3336,
+      "mean_token_accuracy": 0.6534389823675155,
+      "num_tokens": 28523506.0,
+      "step": 3990
+    },
+    {
+      "epoch": 0.8167432363450741,
+      "grad_norm": 11.901576042175293,
+      "learning_rate": 9.541165721028337e-06,
+      "loss": 1.3697,
+      "step": 4000
+    },
+    {
+      "epoch": 0.8167432363450741,
+      "eval_loss": 1.430210828781128,
+      "eval_mean_token_accuracy": 0.6311295317021774,
+      "eval_num_tokens": 28595276.0,
+      "eval_runtime": 994.082,
+      "eval_samples_per_second": 4.973,
+      "eval_steps_per_second": 1.243,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 24485,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.8128782228832911e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}