diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4174 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 5914,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0033818058843422386,
+      "grad_norm": 1.5605809688568115,
+      "learning_rate": 9.991538331358945e-05,
+      "loss": 1.5931,
+      "step": 10
+    },
+    {
+      "epoch": 0.006763611768684477,
+      "grad_norm": 0.2704204320907593,
+      "learning_rate": 9.974614994076833e-05,
+      "loss": 0.6298,
+      "step": 20
+    },
+    {
+      "epoch": 0.010145417653026716,
+      "grad_norm": 0.20275332033634186,
+      "learning_rate": 9.957691656794721e-05,
+      "loss": 0.6147,
+      "step": 30
+    },
+    {
+      "epoch": 0.013527223537368955,
+      "grad_norm": 0.1859731376171112,
+      "learning_rate": 9.940768319512609e-05,
+      "loss": 0.5784,
+      "step": 40
+    },
+    {
+      "epoch": 0.016909029421711193,
+      "grad_norm": 0.20401914417743683,
+      "learning_rate": 9.923844982230497e-05,
+      "loss": 0.5696,
+      "step": 50
+    },
+    {
+      "epoch": 0.020290835306053433,
+      "grad_norm": 0.21193653345108032,
+      "learning_rate": 9.906921644948385e-05,
+      "loss": 0.565,
+      "step": 60
+    },
+    {
+      "epoch": 0.023672641190395673,
+      "grad_norm": 0.20135045051574707,
+      "learning_rate": 9.889998307666272e-05,
+      "loss": 0.5251,
+      "step": 70
+    },
+    {
+      "epoch": 0.02705444707473791,
+      "grad_norm": 0.20686441659927368,
+      "learning_rate": 9.873074970384161e-05,
+      "loss": 0.5295,
+      "step": 80
+    },
+    {
+      "epoch": 0.03043625295908015,
+      "grad_norm": 0.20559485256671906,
+      "learning_rate": 9.856151633102048e-05,
+      "loss": 0.5272,
+      "step": 90
+    },
+    {
+      "epoch": 0.033818058843422386,
+      "grad_norm": 0.23076851665973663,
+      "learning_rate": 9.839228295819937e-05,
+      "loss": 0.5284,
+      "step": 100
+    },
+    {
+      "epoch": 0.037199864727764625,
+      "grad_norm": 0.21439723670482635,
+      "learning_rate": 9.822304958537824e-05,
+      "loss": 0.5088,
+      "step": 110
+    },
+    {
+      "epoch": 0.040581670612106865,
+      "grad_norm": 0.23800402879714966,
+      "learning_rate": 9.805381621255713e-05,
+      "loss": 0.5123,
+      "step": 120
+    },
+    {
+      "epoch": 0.043963476496449105,
+      "grad_norm": 0.2768985629081726,
+      "learning_rate": 9.7884582839736e-05,
+      "loss": 0.4884,
+      "step": 130
+    },
+    {
+      "epoch": 0.047345282380791345,
+      "grad_norm": 0.26437142491340637,
+      "learning_rate": 9.771534946691489e-05,
+      "loss": 0.4748,
+      "step": 140
+    },
+    {
+      "epoch": 0.05072708826513358,
+      "grad_norm": 0.2534578740596771,
+      "learning_rate": 9.754611609409376e-05,
+      "loss": 0.4766,
+      "step": 150
+    },
+    {
+      "epoch": 0.05410889414947582,
+      "grad_norm": 0.2550220787525177,
+      "learning_rate": 9.737688272127265e-05,
+      "loss": 0.4613,
+      "step": 160
+    },
+    {
+      "epoch": 0.05749070003381806,
+      "grad_norm": 0.2937515377998352,
+      "learning_rate": 9.720764934845152e-05,
+      "loss": 0.4631,
+      "step": 170
+    },
+    {
+      "epoch": 0.0608725059181603,
+      "grad_norm": 0.31693148612976074,
+      "learning_rate": 9.703841597563041e-05,
+      "loss": 0.4454,
+      "step": 180
+    },
+    {
+      "epoch": 0.06425431180250253,
+      "grad_norm": 0.3209972679615021,
+      "learning_rate": 9.686918260280928e-05,
+      "loss": 0.4389,
+      "step": 190
+    },
+    {
+      "epoch": 0.06763611768684477,
+      "grad_norm": 0.26218757033348083,
+      "learning_rate": 9.669994922998817e-05,
+      "loss": 0.4393,
+      "step": 200
+    },
+    {
+      "epoch": 0.07101792357118701,
+      "grad_norm": 0.23638683557510376,
+      "learning_rate": 9.653071585716704e-05,
+      "loss": 0.393,
+      "step": 210
+    },
+    {
+      "epoch": 0.07439972945552925,
+      "grad_norm": 0.2807103395462036,
+      "learning_rate": 9.636148248434593e-05,
+      "loss": 0.4012,
+      "step": 220
+    },
+    {
+      "epoch": 0.07778153533987149,
+      "grad_norm": 0.29169249534606934,
+      "learning_rate": 9.61922491115248e-05,
+      "loss": 0.4146,
+      "step": 230
+    },
+    {
+      "epoch": 0.08116334122421373,
+      "grad_norm": 0.26288753747940063,
+      "learning_rate": 9.602301573870369e-05,
+      "loss": 0.4185,
+      "step": 240
+    },
+    {
+      "epoch": 0.08454514710855597,
+      "grad_norm": 0.2438131868839264,
+      "learning_rate": 9.585378236588256e-05,
+      "loss": 0.3786,
+      "step": 250
+    },
+    {
+      "epoch": 0.08792695299289821,
+      "grad_norm": 0.2577499449253082,
+      "learning_rate": 9.568454899306143e-05,
+      "loss": 0.4075,
+      "step": 260
+    },
+    {
+      "epoch": 0.09130875887724045,
+      "grad_norm": 0.2468397617340088,
+      "learning_rate": 9.551531562024032e-05,
+      "loss": 0.382,
+      "step": 270
+    },
+    {
+      "epoch": 0.09469056476158269,
+      "grad_norm": 0.27210426330566406,
+      "learning_rate": 9.53460822474192e-05,
+      "loss": 0.3853,
+      "step": 280
+    },
+    {
+      "epoch": 0.09807237064592493,
+      "grad_norm": 0.2579999566078186,
+      "learning_rate": 9.517684887459808e-05,
+      "loss": 0.3816,
+      "step": 290
+    },
+    {
+      "epoch": 0.10145417653026716,
+      "grad_norm": 0.24862495064735413,
+      "learning_rate": 9.500761550177695e-05,
+      "loss": 0.3481,
+      "step": 300
+    },
+    {
+      "epoch": 0.1048359824146094,
+      "grad_norm": 0.289042204618454,
+      "learning_rate": 9.483838212895584e-05,
+      "loss": 0.3961,
+      "step": 310
+    },
+    {
+      "epoch": 0.10821778829895164,
+      "grad_norm": 0.2919853627681732,
+      "learning_rate": 9.466914875613471e-05,
+      "loss": 0.3332,
+      "step": 320
+    },
+    {
+      "epoch": 0.11159959418329388,
+      "grad_norm": 0.25639480352401733,
+      "learning_rate": 9.44999153833136e-05,
+      "loss": 0.3644,
+      "step": 330
+    },
+    {
+      "epoch": 0.11498140006763612,
+      "grad_norm": 0.26778921484947205,
+      "learning_rate": 9.433068201049247e-05,
+      "loss": 0.3695,
+      "step": 340
+    },
+    {
+      "epoch": 0.11836320595197836,
+      "grad_norm": 0.24935244023799896,
+      "learning_rate": 9.416144863767136e-05,
+      "loss": 0.3433,
+      "step": 350
+    },
+    {
+      "epoch": 0.1217450118363206,
+      "grad_norm": 0.2668505609035492,
+      "learning_rate": 9.399221526485023e-05,
+      "loss": 0.3689,
+      "step": 360
+    },
+    {
+      "epoch": 0.12512681772066284,
+      "grad_norm": 0.2972855865955353,
+      "learning_rate": 9.382298189202912e-05,
+      "loss": 0.3153,
+      "step": 370
+    },
+    {
+      "epoch": 0.12850862360500506,
+      "grad_norm": 0.2680118978023529,
+      "learning_rate": 9.3653748519208e-05,
+      "loss": 0.3332,
+      "step": 380
+    },
+    {
+      "epoch": 0.13189042948934732,
+      "grad_norm": 0.28004977107048035,
+      "learning_rate": 9.348451514638688e-05,
+      "loss": 0.342,
+      "step": 390
+    },
+    {
+      "epoch": 0.13527223537368954,
+      "grad_norm": 0.2519853115081787,
+      "learning_rate": 9.331528177356575e-05,
+      "loss": 0.3194,
+      "step": 400
+    },
+    {
+      "epoch": 0.1386540412580318,
+      "grad_norm": 0.2775399386882782,
+      "learning_rate": 9.314604840074464e-05,
+      "loss": 0.3113,
+      "step": 410
+    },
+    {
+      "epoch": 0.14203584714237402,
+      "grad_norm": 0.27479955554008484,
+      "learning_rate": 9.297681502792351e-05,
+      "loss": 0.3148,
+      "step": 420
+    },
+    {
+      "epoch": 0.14541765302671628,
+      "grad_norm": 0.3637036681175232,
+      "learning_rate": 9.28075816551024e-05,
+      "loss": 0.3065,
+      "step": 430
+    },
+    {
+      "epoch": 0.1487994589110585,
+      "grad_norm": 0.2898998260498047,
+      "learning_rate": 9.263834828228127e-05,
+      "loss": 0.2799,
+      "step": 440
+    },
+    {
+      "epoch": 0.15218126479540076,
+      "grad_norm": 0.295907199382782,
+      "learning_rate": 9.246911490946015e-05,
+      "loss": 0.2984,
+      "step": 450
+    },
+    {
+      "epoch": 0.15556307067974298,
+      "grad_norm": 0.3111323118209839,
+      "learning_rate": 9.229988153663903e-05,
+      "loss": 0.2854,
+      "step": 460
+    },
+    {
+      "epoch": 0.1589448765640852,
+      "grad_norm": 0.2768048346042633,
+      "learning_rate": 9.21306481638179e-05,
+      "loss": 0.2895,
+      "step": 470
+    },
+    {
+      "epoch": 0.16232668244842746,
+      "grad_norm": 0.34783947467803955,
+      "learning_rate": 9.19614147909968e-05,
+      "loss": 0.2686,
+      "step": 480
+    },
+    {
+      "epoch": 0.1657084883327697,
+      "grad_norm": 0.3182440400123596,
+      "learning_rate": 9.179218141817567e-05,
+      "loss": 0.2603,
+      "step": 490
+    },
+    {
+      "epoch": 0.16909029421711194,
+      "grad_norm": 0.3452625572681427,
+      "learning_rate": 9.162294804535455e-05,
+      "loss": 0.296,
+      "step": 500
+    },
+    {
+      "epoch": 0.17247210010145417,
+      "grad_norm": 0.329106867313385,
+      "learning_rate": 9.145371467253343e-05,
+      "loss": 0.2665,
+      "step": 510
+    },
+    {
+      "epoch": 0.17585390598579642,
+      "grad_norm": 0.3779897689819336,
+      "learning_rate": 9.128448129971231e-05,
+      "loss": 0.2545,
+      "step": 520
+    },
+    {
+      "epoch": 0.17923571187013865,
+      "grad_norm": 0.33585628867149353,
+      "learning_rate": 9.111524792689119e-05,
+      "loss": 0.2683,
+      "step": 530
+    },
+    {
+      "epoch": 0.1826175177544809,
+      "grad_norm": 0.27421489357948303,
+      "learning_rate": 9.094601455407007e-05,
+      "loss": 0.2585,
+      "step": 540
+    },
+    {
+      "epoch": 0.18599932363882313,
+      "grad_norm": 0.37735405564308167,
+      "learning_rate": 9.077678118124895e-05,
+      "loss": 0.2676,
+      "step": 550
+    },
+    {
+      "epoch": 0.18938112952316538,
+      "grad_norm": 0.30768734216690063,
+      "learning_rate": 9.060754780842783e-05,
+      "loss": 0.2543,
+      "step": 560
+    },
+    {
+      "epoch": 0.1927629354075076,
+      "grad_norm": 0.3587827682495117,
+      "learning_rate": 9.04383144356067e-05,
+      "loss": 0.2249,
+      "step": 570
+    },
+    {
+      "epoch": 0.19614474129184986,
+      "grad_norm": 0.3213902711868286,
+      "learning_rate": 9.026908106278559e-05,
+      "loss": 0.2443,
+      "step": 580
+    },
+    {
+      "epoch": 0.1995265471761921,
+      "grad_norm": 0.4303820729255676,
+      "learning_rate": 9.009984768996447e-05,
+      "loss": 0.2256,
+      "step": 590
+    },
+    {
+      "epoch": 0.2029083530605343,
+      "grad_norm": 0.2948843538761139,
+      "learning_rate": 8.993061431714335e-05,
+      "loss": 0.2295,
+      "step": 600
+    },
+    {
+      "epoch": 0.20629015894487657,
+      "grad_norm": 0.3263229727745056,
+      "learning_rate": 8.976138094432223e-05,
+      "loss": 0.239,
+      "step": 610
+    },
+    {
+      "epoch": 0.2096719648292188,
+      "grad_norm": 0.3471197783946991,
+      "learning_rate": 8.95921475715011e-05,
+      "loss": 0.2227,
+      "step": 620
+    },
+    {
+      "epoch": 0.21305377071356105,
+      "grad_norm": 0.3152846693992615,
+      "learning_rate": 8.942291419867999e-05,
+      "loss": 0.2359,
+      "step": 630
+    },
+    {
+      "epoch": 0.21643557659790327,
+      "grad_norm": 0.4284135401248932,
+      "learning_rate": 8.925368082585886e-05,
+      "loss": 0.2453,
+      "step": 640
+    },
+    {
+      "epoch": 0.21981738248224553,
+      "grad_norm": 0.37850794196128845,
+      "learning_rate": 8.908444745303775e-05,
+      "loss": 0.2245,
+      "step": 650
+    },
+    {
+      "epoch": 0.22319918836658775,
+      "grad_norm": 0.38802939653396606,
+      "learning_rate": 8.891521408021662e-05,
+      "loss": 0.2227,
+      "step": 660
+    },
+    {
+      "epoch": 0.22658099425093,
+      "grad_norm": 0.32752034068107605,
+      "learning_rate": 8.87459807073955e-05,
+      "loss": 0.2008,
+      "step": 670
+    },
+    {
+      "epoch": 0.22996280013527223,
+      "grad_norm": 0.3419714570045471,
+      "learning_rate": 8.857674733457438e-05,
+      "loss": 0.2049,
+      "step": 680
+    },
+    {
+      "epoch": 0.2333446060196145,
+      "grad_norm": 0.3725152313709259,
+      "learning_rate": 8.840751396175327e-05,
+      "loss": 0.1922,
+      "step": 690
+    },
+    {
+      "epoch": 0.2367264119039567,
+      "grad_norm": 0.3231143653392792,
+      "learning_rate": 8.823828058893214e-05,
+      "loss": 0.2094,
+      "step": 700
+    },
+    {
+      "epoch": 0.24010821778829894,
+      "grad_norm": 0.3786492347717285,
+      "learning_rate": 8.806904721611103e-05,
+      "loss": 0.1917,
+      "step": 710
+    },
+    {
+      "epoch": 0.2434900236726412,
+      "grad_norm": 0.3543280363082886,
+      "learning_rate": 8.78998138432899e-05,
+      "loss": 0.2005,
+      "step": 720
+    },
+    {
+      "epoch": 0.24687182955698342,
+      "grad_norm": 0.336761474609375,
+      "learning_rate": 8.773058047046879e-05,
+      "loss": 0.174,
+      "step": 730
+    },
+    {
+      "epoch": 0.2502536354413257,
+      "grad_norm": 0.3629204332828522,
+      "learning_rate": 8.756134709764766e-05,
+      "loss": 0.1851,
+      "step": 740
+    },
+    {
+      "epoch": 0.2536354413256679,
+      "grad_norm": 0.3555070459842682,
+      "learning_rate": 8.739211372482655e-05,
+      "loss": 0.1832,
+      "step": 750
+    },
+    {
+      "epoch": 0.2570172472100101,
+      "grad_norm": 0.41113823652267456,
+      "learning_rate": 8.722288035200542e-05,
+      "loss": 0.1772,
+      "step": 760
+    },
+    {
+      "epoch": 0.2603990530943524,
+      "grad_norm": 0.32061767578125,
+      "learning_rate": 8.70536469791843e-05,
+      "loss": 0.1707,
+      "step": 770
+    },
+    {
+      "epoch": 0.26378085897869463,
+      "grad_norm": 0.35991913080215454,
+      "learning_rate": 8.688441360636318e-05,
+      "loss": 0.1487,
+      "step": 780
+    },
+    {
+      "epoch": 0.26716266486303686,
+      "grad_norm": 0.3859257102012634,
+      "learning_rate": 8.671518023354207e-05,
+      "loss": 0.1688,
+      "step": 790
+    },
+    {
+      "epoch": 0.2705444707473791,
+      "grad_norm": 0.31017613410949707,
+      "learning_rate": 8.654594686072094e-05,
+      "loss": 0.1624,
+      "step": 800
+    },
+    {
+      "epoch": 0.27392627663172137,
+      "grad_norm": 0.35503777861595154,
+      "learning_rate": 8.637671348789981e-05,
+      "loss": 0.1642,
+      "step": 810
+    },
+    {
+      "epoch": 0.2773080825160636,
+      "grad_norm": 0.3433260917663574,
+      "learning_rate": 8.62074801150787e-05,
+      "loss": 0.1844,
+      "step": 820
+    },
+    {
+      "epoch": 0.2806898884004058,
+      "grad_norm": 0.3865962624549866,
+      "learning_rate": 8.603824674225757e-05,
+      "loss": 0.1506,
+      "step": 830
+    },
+    {
+      "epoch": 0.28407169428474804,
+      "grad_norm": 0.3324780762195587,
+      "learning_rate": 8.586901336943646e-05,
+      "loss": 0.1632,
+      "step": 840
+    },
+    {
+      "epoch": 0.28745350016909027,
+      "grad_norm": 0.4064556062221527,
+      "learning_rate": 8.569977999661533e-05,
+      "loss": 0.1519,
+      "step": 850
+    },
+    {
+      "epoch": 0.29083530605343255,
+      "grad_norm": 0.3946615159511566,
+      "learning_rate": 8.553054662379422e-05,
+      "loss": 0.157,
+      "step": 860
+    },
+    {
+      "epoch": 0.2942171119377748,
+      "grad_norm": 0.3700334429740906,
+      "learning_rate": 8.536131325097309e-05,
+      "loss": 0.141,
+      "step": 870
+    },
+    {
+      "epoch": 0.297598917822117,
+      "grad_norm": 0.420950710773468,
+      "learning_rate": 8.519207987815198e-05,
+      "loss": 0.1488,
+      "step": 880
+    },
+    {
+      "epoch": 0.30098072370645923,
+      "grad_norm": 0.39633265137672424,
+      "learning_rate": 8.502284650533085e-05,
+      "loss": 0.1545,
+      "step": 890
+    },
+    {
+      "epoch": 0.3043625295908015,
+      "grad_norm": 0.38442814350128174,
+      "learning_rate": 8.485361313250974e-05,
+      "loss": 0.141,
+      "step": 900
+    },
+    {
+      "epoch": 0.30774433547514374,
+      "grad_norm": 0.41429537534713745,
+      "learning_rate": 8.468437975968861e-05,
+      "loss": 0.144,
+      "step": 910
+    },
+    {
+      "epoch": 0.31112614135948596,
+      "grad_norm": 0.3561369478702545,
+      "learning_rate": 8.45151463868675e-05,
+      "loss": 0.134,
+      "step": 920
+    },
+    {
+      "epoch": 0.3145079472438282,
+      "grad_norm": 0.3710062801837921,
+      "learning_rate": 8.434591301404637e-05,
+      "loss": 0.1457,
+      "step": 930
+    },
+    {
+      "epoch": 0.3178897531281704,
+      "grad_norm": 0.3607446849346161,
+      "learning_rate": 8.417667964122526e-05,
+      "loss": 0.122,
+      "step": 940
+    },
+    {
+      "epoch": 0.3212715590125127,
+      "grad_norm": 0.34604352712631226,
+      "learning_rate": 8.400744626840413e-05,
+      "loss": 0.1414,
+      "step": 950
+    },
+    {
+      "epoch": 0.3246533648968549,
+      "grad_norm": 0.27822059392929077,
+      "learning_rate": 8.383821289558302e-05,
+      "loss": 0.1281,
+      "step": 960
+    },
+    {
+      "epoch": 0.32803517078119715,
+      "grad_norm": 0.38080519437789917,
+      "learning_rate": 8.366897952276189e-05,
+      "loss": 0.1392,
+      "step": 970
+    },
+    {
+      "epoch": 0.3314169766655394,
+      "grad_norm": 0.4061764180660248,
+      "learning_rate": 8.349974614994078e-05,
+      "loss": 0.1179,
+      "step": 980
+    },
+    {
+      "epoch": 0.33479878254988166,
+      "grad_norm": 0.28504157066345215,
+      "learning_rate": 8.333051277711965e-05,
+      "loss": 0.1284,
+      "step": 990
+    },
+    {
+      "epoch": 0.3381805884342239,
+      "grad_norm": 0.3242844343185425,
+      "learning_rate": 8.316127940429852e-05,
+      "loss": 0.1184,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3415623943185661,
+      "grad_norm": 0.3317604064941406,
+      "learning_rate": 8.299204603147741e-05,
+      "loss": 0.13,
+      "step": 1010
+    },
+    {
+      "epoch": 0.34494420020290834,
+      "grad_norm": 0.39583227038383484,
+      "learning_rate": 8.282281265865628e-05,
+      "loss": 0.1146,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3483260060872506,
+      "grad_norm": 0.33169683814048767,
+      "learning_rate": 8.265357928583517e-05,
+      "loss": 0.1162,
+      "step": 1030
+    },
+    {
+      "epoch": 0.35170781197159284,
+      "grad_norm": 0.38136041164398193,
+      "learning_rate": 8.248434591301404e-05,
+      "loss": 0.1268,
+      "step": 1040
+    },
+    {
+      "epoch": 0.35508961785593507,
+      "grad_norm": 0.28301218152046204,
+      "learning_rate": 8.231511254019293e-05,
+      "loss": 0.1117,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3584714237402773,
+      "grad_norm": 0.3950181007385254,
+      "learning_rate": 8.21458791673718e-05,
+      "loss": 0.1172,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3618532296246195,
+      "grad_norm": 0.40106379985809326,
+      "learning_rate": 8.197664579455069e-05,
+      "loss": 0.1134,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3652350355089618,
+      "grad_norm": 0.3036629259586334,
+      "learning_rate": 8.180741242172956e-05,
+      "loss": 0.0977,
+      "step": 1080
+    },
+    {
+      "epoch": 0.36861684139330403,
+      "grad_norm": 0.35834598541259766,
+      "learning_rate": 8.163817904890845e-05,
+      "loss": 0.1226,
+      "step": 1090
+    },
+    {
+      "epoch": 0.37199864727764625,
+      "grad_norm": 0.3107944130897522,
+      "learning_rate": 8.146894567608732e-05,
+      "loss": 0.0932,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3753804531619885,
+      "grad_norm": 0.2708600163459778,
+      "learning_rate": 8.129971230326621e-05,
+      "loss": 0.1071,
+      "step": 1110
+    },
+    {
+      "epoch": 0.37876225904633076,
+      "grad_norm": 0.3161390721797943,
+      "learning_rate": 8.113047893044508e-05,
+      "loss": 0.0951,
+      "step": 1120
+    },
+    {
+      "epoch": 0.382144064930673,
+      "grad_norm": 0.2967537045478821,
+      "learning_rate": 8.096124555762397e-05,
+      "loss": 0.0961,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3855258708150152,
+      "grad_norm": 0.3615525960922241,
+      "learning_rate": 8.079201218480284e-05,
+      "loss": 0.0955,
+      "step": 1140
+    },
+    {
+      "epoch": 0.38890767669935744,
+      "grad_norm": 0.33469459414482117,
+      "learning_rate": 8.062277881198173e-05,
+      "loss": 0.1028,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3922894825836997,
+      "grad_norm": 0.44299551844596863,
+      "learning_rate": 8.04535454391606e-05,
+      "loss": 0.1065,
+      "step": 1160
+    },
+    {
+      "epoch": 0.39567128846804195,
+      "grad_norm": 0.3324688971042633,
+      "learning_rate": 8.028431206633949e-05,
+      "loss": 0.1026,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3990530943523842,
+      "grad_norm": 0.3845025599002838,
+      "learning_rate": 8.011507869351836e-05,
+      "loss": 0.1166,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4024349002367264,
+      "grad_norm": 0.36462751030921936,
+      "learning_rate": 7.994584532069724e-05,
+      "loss": 0.0955,
+      "step": 1190
+    },
+    {
+      "epoch": 0.4058167061210686,
+      "grad_norm": 0.33478081226348877,
+      "learning_rate": 7.977661194787612e-05,
+      "loss": 0.0966,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4091985120054109,
+      "grad_norm": 0.3839697539806366,
+      "learning_rate": 7.9607378575055e-05,
+      "loss": 0.1011,
+      "step": 1210
+    },
+    {
+      "epoch": 0.41258031788975313,
+      "grad_norm": 0.33828845620155334,
+      "learning_rate": 7.943814520223388e-05,
+      "loss": 0.094,
+      "step": 1220
+    },
+    {
+      "epoch": 0.41596212377409536,
+      "grad_norm": 0.35532712936401367,
+      "learning_rate": 7.926891182941276e-05,
+      "loss": 0.0771,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4193439296584376,
+      "grad_norm": 0.32812246680259705,
+      "learning_rate": 7.909967845659164e-05,
+      "loss": 0.0931,
+      "step": 1240
+    },
+    {
+      "epoch": 0.42272573554277987,
+      "grad_norm": 0.4312772750854492,
+      "learning_rate": 7.893044508377052e-05,
+      "loss": 0.091,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4261075414271221,
+      "grad_norm": 0.38822853565216064,
+      "learning_rate": 7.87612117109494e-05,
+      "loss": 0.0936,
+      "step": 1260
+    },
+    {
+      "epoch": 0.4294893473114643,
+      "grad_norm": 0.30626174807548523,
+      "learning_rate": 7.859197833812828e-05,
+      "loss": 0.0773,
+      "step": 1270
+    },
+    {
+      "epoch": 0.43287115319580655,
+      "grad_norm": 0.36472904682159424,
+      "learning_rate": 7.842274496530716e-05,
+      "loss": 0.0776,
+      "step": 1280
+    },
+    {
+      "epoch": 0.43625295908014877,
+      "grad_norm": 0.312610000371933,
+      "learning_rate": 7.825351159248604e-05,
+      "loss": 0.0882,
+      "step": 1290
+    },
+    {
+      "epoch": 0.43963476496449105,
+      "grad_norm": 0.34325119853019714,
+      "learning_rate": 7.808427821966492e-05,
+      "loss": 0.0836,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4430165708488333,
+      "grad_norm": 0.39519986510276794,
+      "learning_rate": 7.79150448468438e-05,
+      "loss": 0.084,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4463983767331755,
+      "grad_norm": 0.35263752937316895,
+      "learning_rate": 7.774581147402268e-05,
+      "loss": 0.0713,
+      "step": 1320
+    },
+    {
+      "epoch": 0.44978018261751773,
+      "grad_norm": 0.36037376523017883,
+      "learning_rate": 7.757657810120156e-05,
+      "loss": 0.069,
+      "step": 1330
+    },
+    {
+      "epoch": 0.45316198850186,
+      "grad_norm": 0.38161391019821167,
+      "learning_rate": 7.740734472838044e-05,
+      "loss": 0.0693,
+      "step": 1340
+    },
+    {
+      "epoch": 0.45654379438620224,
+      "grad_norm": 0.3896481394767761,
+      "learning_rate": 7.723811135555932e-05,
+      "loss": 0.0793,
+      "step": 1350
+    },
+    {
+      "epoch": 0.45992560027054447,
+      "grad_norm": 0.3284771740436554,
+      "learning_rate": 7.70688779827382e-05,
+      "loss": 0.0874,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4633074061548867,
+      "grad_norm": 0.3116714358329773,
+      "learning_rate": 7.689964460991708e-05,
+      "loss": 0.0734,
+      "step": 1370
+    },
+    {
+      "epoch": 0.466689212039229,
+      "grad_norm": 0.29977601766586304,
+      "learning_rate": 7.673041123709595e-05,
+      "loss": 0.0696,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4700710179235712,
+      "grad_norm": 0.30298420786857605,
+      "learning_rate": 7.656117786427484e-05,
+      "loss": 0.0755,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4734528238079134,
+      "grad_norm": 0.3856925368309021,
+      "learning_rate": 7.639194449145371e-05,
+      "loss": 0.0791,
+      "step": 1400
+    },
+    {
+      "epoch": 0.47683462969225565,
+      "grad_norm": 0.3421929180622101,
+      "learning_rate": 7.62227111186326e-05,
+      "loss": 0.0674,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4802164355765979,
+      "grad_norm": 0.3304542899131775,
+      "learning_rate": 7.605347774581147e-05,
+      "loss": 0.076,
+      "step": 1420
+    },
+    {
+      "epoch": 0.48359824146094016,
+      "grad_norm": 0.29870203137397766,
+      "learning_rate": 7.588424437299036e-05,
+      "loss": 0.0703,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4869800473452824,
+      "grad_norm": 0.2472551167011261,
+      "learning_rate": 7.571501100016923e-05,
+      "loss": 0.069,
+      "step": 1440
+    },
+    {
+      "epoch": 0.4903618532296246,
+      "grad_norm": 0.2860611081123352,
+      "learning_rate": 7.554577762734812e-05,
+      "loss": 0.0667,
+      "step": 1450
+    },
+    {
+      "epoch": 0.49374365911396684,
+      "grad_norm": 0.3618302345275879,
+      "learning_rate": 7.537654425452699e-05,
+      "loss": 0.0597,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4971254649983091,
+      "grad_norm": 0.34842416644096375,
+      "learning_rate": 7.520731088170588e-05,
+      "loss": 0.0715,
+      "step": 1470
+    },
+    {
+      "epoch": 0.5005072708826513,
+      "grad_norm": 0.2862137258052826,
+      "learning_rate": 7.503807750888475e-05,
+      "loss": 0.0778,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5038890767669936,
+      "grad_norm": 0.3283160626888275,
+      "learning_rate": 7.486884413606364e-05,
+      "loss": 0.073,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5072708826513358,
+      "grad_norm": 0.2540145218372345,
+      "learning_rate": 7.469961076324251e-05,
+      "loss": 0.067,
+      "step": 1500
+    },
+    {
+      "epoch": 0.510652688535678,
+      "grad_norm": 0.32044336199760437,
+      "learning_rate": 7.45303773904214e-05,
+      "loss": 0.063,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5140344944200202,
+      "grad_norm": 0.3898891508579254,
+      "learning_rate": 7.436114401760027e-05,
+      "loss": 0.0695,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5174163003043625,
+      "grad_norm": 0.30922603607177734,
+      "learning_rate": 7.419191064477916e-05,
+      "loss": 0.071,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5207981061887048,
+      "grad_norm": 0.3513735234737396,
+      "learning_rate": 7.402267727195803e-05,
+      "loss": 0.0722,
+      "step": 1540
+    },
+    {
+      "epoch": 0.524179912073047,
+      "grad_norm": 0.2806611955165863,
+      "learning_rate": 7.38534438991369e-05,
+      "loss": 0.0662,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5275617179573893,
+      "grad_norm": 0.32641470432281494,
+      "learning_rate": 7.368421052631579e-05,
+      "loss": 0.0645,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5309435238417315,
+      "grad_norm": 0.2627265751361847,
+      "learning_rate": 7.351497715349466e-05,
+      "loss": 0.0594,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5343253297260737,
+      "grad_norm": 0.2586800158023834,
+      "learning_rate": 7.334574378067355e-05,
+      "loss": 0.0603,
+      "step": 1580
+    },
+    {
+      "epoch": 0.5377071356104159,
+      "grad_norm": 0.24359896779060364,
+      "learning_rate": 7.317651040785242e-05,
+      "loss": 0.0638,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5410889414947582,
+      "grad_norm": 0.3084506392478943,
+      "learning_rate": 7.300727703503131e-05,
+      "loss": 0.0658,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5444707473791004,
+      "grad_norm": 0.29731103777885437,
+      "learning_rate": 7.283804366221018e-05,
+      "loss": 0.0548,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5478525532634427,
+      "grad_norm": 0.25320005416870117,
+      "learning_rate": 7.266881028938907e-05,
+      "loss": 0.0533,
+      "step": 1620
+    },
+    {
+      "epoch": 0.551234359147785,
+      "grad_norm": 0.274166077375412,
+      "learning_rate": 7.249957691656794e-05,
+      "loss": 0.048,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5546161650321272,
+      "grad_norm": 0.29488497972488403,
+      "learning_rate": 7.233034354374683e-05,
+      "loss": 0.054,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5579979709164694,
+      "grad_norm": 0.2893562316894531,
+      "learning_rate": 7.21611101709257e-05,
+      "loss": 0.0589,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5613797768008116,
+      "grad_norm": 0.29615455865859985,
+      "learning_rate": 7.199187679810459e-05,
+      "loss": 0.0524,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5647615826851539,
+      "grad_norm": 0.3037727475166321,
+      "learning_rate": 7.182264342528346e-05,
+      "loss": 0.0486,
+      "step": 1670
+    },
+    {
+      "epoch": 0.5681433885694961,
+      "grad_norm": 0.2997746765613556,
+      "learning_rate": 7.165341005246235e-05,
+      "loss": 0.047,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5715251944538383,
+      "grad_norm": 0.31440579891204834,
+      "learning_rate": 7.148417667964122e-05,
+      "loss": 0.0467,
+      "step": 1690
+    },
+    {
+      "epoch": 0.5749070003381805,
+      "grad_norm": 0.345804363489151,
+      "learning_rate": 7.131494330682011e-05,
+      "loss": 0.0543,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5782888062225229,
+      "grad_norm": 0.3310382664203644,
+      "learning_rate": 7.114570993399898e-05,
+      "loss": 0.0561,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5816706121068651,
+      "grad_norm": 0.33962053060531616,
+      "learning_rate": 7.097647656117787e-05,
+      "loss": 0.0468,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5850524179912073,
+      "grad_norm": 0.303245484828949,
+      "learning_rate": 7.080724318835674e-05,
+      "loss": 0.0503,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5884342238755496,
+      "grad_norm": 0.3114894926548004,
+      "learning_rate": 7.063800981553562e-05,
+      "loss": 0.0482,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5918160297598918,
+      "grad_norm": 0.23478536307811737,
+      "learning_rate": 7.04687764427145e-05,
+      "loss": 0.0496,
+      "step": 1750
+    },
+    {
+      "epoch": 0.595197835644234,
+      "grad_norm": 0.3116193413734436,
+      "learning_rate": 7.029954306989338e-05,
+      "loss": 0.0498,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5985796415285762,
+      "grad_norm": 0.30709898471832275,
+      "learning_rate": 7.013030969707226e-05,
+      "loss": 0.0528,
+      "step": 1770
+    },
+    {
+      "epoch": 0.6019614474129185,
+      "grad_norm": 0.3018944561481476,
+      "learning_rate": 6.996107632425114e-05,
+      "loss": 0.0466,
+      "step": 1780
+    },
+    {
+      "epoch": 0.6053432532972607,
+      "grad_norm": 0.32787784934043884,
+      "learning_rate": 6.979184295143002e-05,
+      "loss": 0.0513,
+      "step": 1790
+    },
+    {
+      "epoch": 0.608725059181603,
+      "grad_norm": 0.30039939284324646,
+      "learning_rate": 6.962260957860891e-05,
+      "loss": 0.0504,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6121068650659452,
+      "grad_norm": 0.3414636552333832,
+      "learning_rate": 6.94533762057878e-05,
+      "loss": 0.0499,
+      "step": 1810
+    },
+    {
+      "epoch": 0.6154886709502875,
+      "grad_norm": 0.20751340687274933,
+      "learning_rate": 6.928414283296667e-05,
+      "loss": 0.0437,
+      "step": 1820
+    },
+    {
+      "epoch": 0.6188704768346297,
+      "grad_norm": 0.3419632315635681,
+      "learning_rate": 6.911490946014556e-05,
+      "loss": 0.049,
+      "step": 1830
+    },
+    {
+      "epoch": 0.6222522827189719,
+      "grad_norm": 0.27273499965667725,
+      "learning_rate": 6.894567608732443e-05,
+      "loss": 0.0444,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6256340886033142,
+      "grad_norm": 0.31872832775115967,
+      "learning_rate": 6.87764427145033e-05,
+      "loss": 0.0535,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6290158944876564,
+      "grad_norm": 0.24325081706047058,
+      "learning_rate": 6.860720934168219e-05,
+      "loss": 0.049,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6323977003719986,
+      "grad_norm": 0.3723507821559906,
+      "learning_rate": 6.843797596886106e-05,
+      "loss": 0.0434,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6357795062563408,
+      "grad_norm": 0.24727103114128113,
+      "learning_rate": 6.826874259603995e-05,
+      "loss": 0.0452,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6391613121406832,
+      "grad_norm": 0.3109544515609741,
+      "learning_rate": 6.809950922321882e-05,
+      "loss": 0.044,
+      "step": 1890
+    },
+    {
+      "epoch": 0.6425431180250254,
+      "grad_norm": 0.27794697880744934,
+      "learning_rate": 6.793027585039771e-05,
+      "loss": 0.0403,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6459249239093676,
+      "grad_norm": 0.3061220347881317,
+      "learning_rate": 6.776104247757658e-05,
+      "loss": 0.0425,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6493067297937098,
+      "grad_norm": 0.25127366185188293,
+      "learning_rate": 6.759180910475547e-05,
+      "loss": 0.0432,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6526885356780521,
+      "grad_norm": 0.26195022463798523,
+      "learning_rate": 6.742257573193434e-05,
+      "loss": 0.042,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6560703415623943,
+      "grad_norm": 0.23891383409500122,
+      "learning_rate": 6.725334235911323e-05,
+      "loss": 0.0404,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6594521474467365,
+      "grad_norm": 0.26952946186065674,
+      "learning_rate": 6.70841089862921e-05,
+      "loss": 0.0414,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6628339533310788,
+      "grad_norm": 0.2608715295791626,
+      "learning_rate": 6.691487561347099e-05,
+      "loss": 0.0418,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6662157592154211,
+      "grad_norm": 0.3117476999759674,
+      "learning_rate": 6.674564224064986e-05,
+      "loss": 0.0417,
+      "step": 1970
+    },
+    {
+      "epoch": 0.6695975650997633,
+      "grad_norm": 0.38972756266593933,
+      "learning_rate": 6.657640886782875e-05,
+      "loss": 0.0432,
+      "step": 1980
+    },
+    {
+      "epoch": 0.6729793709841055,
+      "grad_norm": 0.23405376076698303,
+      "learning_rate": 6.640717549500762e-05,
+      "loss": 0.0448,
+      "step": 1990
+    },
+    {
+      "epoch": 0.6763611768684478,
+      "grad_norm": 0.2317565679550171,
+      "learning_rate": 6.623794212218651e-05,
+      "loss": 0.0391,
+      "step": 2000
+    },
+    {
+      "epoch": 0.67974298275279,
+      "grad_norm": 0.2788334786891937,
+      "learning_rate": 6.606870874936538e-05,
+      "loss": 0.041,
+      "step": 2010
+    },
+    {
+      "epoch": 0.6831247886371322,
+      "grad_norm": 0.2472919374704361,
+      "learning_rate": 6.589947537654427e-05,
+      "loss": 0.0413,
+      "step": 2020
+    },
+    {
+      "epoch": 0.6865065945214744,
+      "grad_norm": 0.31981515884399414,
+      "learning_rate": 6.573024200372314e-05,
+      "loss": 0.0394,
+      "step": 2030
+    },
+    {
+      "epoch": 0.6898884004058167,
+      "grad_norm": 0.3317371606826782,
+      "learning_rate": 6.556100863090201e-05,
+      "loss": 0.0405,
+      "step": 2040
+    },
+    {
+      "epoch": 0.6932702062901589,
+      "grad_norm": 0.22539494931697845,
+      "learning_rate": 6.53917752580809e-05,
+      "loss": 0.0364,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6966520121745012,
+      "grad_norm": 0.2573080360889435,
+      "learning_rate": 6.522254188525977e-05,
+      "loss": 0.0404,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7000338180588435,
+      "grad_norm": 0.3429962992668152,
+      "learning_rate": 6.505330851243866e-05,
+      "loss": 0.0432,
+      "step": 2070
+    },
+    {
+      "epoch": 0.7034156239431857,
+      "grad_norm": 0.26425597071647644,
+      "learning_rate": 6.488407513961753e-05,
+      "loss": 0.0383,
+      "step": 2080
+    },
+    {
+      "epoch": 0.7067974298275279,
+      "grad_norm": 0.28535112738609314,
+      "learning_rate": 6.471484176679642e-05,
+      "loss": 0.042,
+      "step": 2090
+    },
+    {
+      "epoch": 0.7101792357118701,
+      "grad_norm": 0.27274271845817566,
+      "learning_rate": 6.45456083939753e-05,
+      "loss": 0.0358,
+      "step": 2100
+    },
+    {
+      "epoch": 0.7135610415962124,
+      "grad_norm": 0.24476689100265503,
+      "learning_rate": 6.437637502115418e-05,
+      "loss": 0.0358,
+      "step": 2110
+    },
+    {
+      "epoch": 0.7169428474805546,
+      "grad_norm": 0.295635461807251,
+      "learning_rate": 6.420714164833305e-05,
+      "loss": 0.0364,
+      "step": 2120
+    },
+    {
+      "epoch": 0.7203246533648968,
+      "grad_norm": 0.3371022343635559,
+      "learning_rate": 6.403790827551194e-05,
+      "loss": 0.0381,
+      "step": 2130
+    },
+    {
+      "epoch": 0.723706459249239,
+      "grad_norm": 0.24469931423664093,
+      "learning_rate": 6.386867490269081e-05,
+      "loss": 0.0384,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7270882651335814,
+      "grad_norm": 0.20693735778331757,
+      "learning_rate": 6.36994415298697e-05,
+      "loss": 0.0369,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7304700710179236,
+      "grad_norm": 0.24583208560943604,
+      "learning_rate": 6.353020815704857e-05,
+      "loss": 0.0367,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7338518769022658,
+      "grad_norm": 0.29009056091308594,
+      "learning_rate": 6.336097478422746e-05,
+      "loss": 0.038,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7372336827866081,
+      "grad_norm": 0.260128378868103,
+      "learning_rate": 6.319174141140633e-05,
+      "loss": 0.039,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7406154886709503,
+      "grad_norm": 0.2804548442363739,
+      "learning_rate": 6.302250803858522e-05,
+      "loss": 0.0399,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7439972945552925,
+      "grad_norm": 0.3354940116405487,
+      "learning_rate": 6.28532746657641e-05,
+      "loss": 0.0363,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7473791004396347,
+      "grad_norm": 0.23447370529174805,
+      "learning_rate": 6.268404129294298e-05,
+      "loss": 0.042,
+      "step": 2210
+    },
+    {
+      "epoch": 0.750760906323977,
+      "grad_norm": 0.26304641366004944,
+      "learning_rate": 6.251480792012185e-05,
+      "loss": 0.0364,
+      "step": 2220
+    },
+    {
+      "epoch": 0.7541427122083192,
+      "grad_norm": 0.26174917817115784,
+      "learning_rate": 6.234557454730073e-05,
+      "loss": 0.0333,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7575245180926615,
+      "grad_norm": 0.2674359977245331,
+      "learning_rate": 6.217634117447961e-05,
+      "loss": 0.0387,
+      "step": 2240
+    },
+    {
+      "epoch": 0.7609063239770038,
+      "grad_norm": 0.21239162981510162,
+      "learning_rate": 6.200710780165849e-05,
+      "loss": 0.0374,
+      "step": 2250
+    },
+    {
+      "epoch": 0.764288129861346,
+      "grad_norm": 0.2588539719581604,
+      "learning_rate": 6.183787442883737e-05,
+      "loss": 0.0354,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7676699357456882,
+      "grad_norm": 0.253009557723999,
+      "learning_rate": 6.166864105601625e-05,
+      "loss": 0.0346,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7710517416300304,
+      "grad_norm": 0.2577788531780243,
+      "learning_rate": 6.149940768319513e-05,
+      "loss": 0.032,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7744335475143727,
+      "grad_norm": 0.2186667025089264,
+      "learning_rate": 6.133017431037401e-05,
+      "loss": 0.0329,
+      "step": 2290
+    },
+    {
+      "epoch": 0.7778153533987149,
+      "grad_norm": 0.278125524520874,
+      "learning_rate": 6.11609409375529e-05,
+      "loss": 0.0338,
+      "step": 2300
+    },
+    {
+      "epoch": 0.7811971592830571,
+      "grad_norm": 0.3377946615219116,
+      "learning_rate": 6.0991707564731774e-05,
+      "loss": 0.037,
+      "step": 2310
+    },
+    {
+      "epoch": 0.7845789651673994,
+      "grad_norm": 0.29264363646507263,
+      "learning_rate": 6.082247419191065e-05,
+      "loss": 0.0361,
+      "step": 2320
+    },
+    {
+      "epoch": 0.7879607710517417,
+      "grad_norm": 0.23570716381072998,
+      "learning_rate": 6.065324081908953e-05,
+      "loss": 0.0368,
+      "step": 2330
+    },
+    {
+      "epoch": 0.7913425769360839,
+      "grad_norm": 0.27007460594177246,
+      "learning_rate": 6.048400744626841e-05,
+      "loss": 0.0334,
+      "step": 2340
+    },
+    {
+      "epoch": 0.7947243828204261,
+      "grad_norm": 0.2671580910682678,
+      "learning_rate": 6.031477407344729e-05,
+      "loss": 0.0342,
+      "step": 2350
+    },
+    {
+      "epoch": 0.7981061887047683,
+      "grad_norm": 0.2466389387845993,
+      "learning_rate": 6.014554070062617e-05,
+      "loss": 0.0403,
+      "step": 2360
+    },
+    {
+      "epoch": 0.8014879945891106,
+      "grad_norm": 0.252189964056015,
+      "learning_rate": 5.997630732780505e-05,
+      "loss": 0.0336,
+      "step": 2370
+    },
+    {
+      "epoch": 0.8048698004734528,
+      "grad_norm": 0.20659616589546204,
+      "learning_rate": 5.980707395498393e-05,
+      "loss": 0.0349,
+      "step": 2380
+    },
+    {
+      "epoch": 0.808251606357795,
+      "grad_norm": 0.32111021876335144,
+      "learning_rate": 5.963784058216281e-05,
+      "loss": 0.0334,
+      "step": 2390
+    },
+    {
+      "epoch": 0.8116334122421373,
+      "grad_norm": 0.21712100505828857,
+      "learning_rate": 5.946860720934169e-05,
+      "loss": 0.0325,
+      "step": 2400
+    },
+    {
+      "epoch": 0.8150152181264796,
+      "grad_norm": 0.2319028079509735,
+      "learning_rate": 5.929937383652057e-05,
+      "loss": 0.0316,
+      "step": 2410
+    },
+    {
+      "epoch": 0.8183970240108218,
+      "grad_norm": 0.2731321156024933,
+      "learning_rate": 5.9130140463699447e-05,
+      "loss": 0.0319,
+      "step": 2420
+    },
+    {
+      "epoch": 0.821778829895164,
+      "grad_norm": 0.25032860040664673,
+      "learning_rate": 5.8960907090878327e-05,
+      "loss": 0.0332,
+      "step": 2430
+    },
+    {
+      "epoch": 0.8251606357795063,
+      "grad_norm": 0.2227882593870163,
+      "learning_rate": 5.8791673718057207e-05,
+      "loss": 0.0329,
+      "step": 2440
+    },
+    {
+      "epoch": 0.8285424416638485,
+      "grad_norm": 0.2757587134838104,
+      "learning_rate": 5.8622440345236087e-05,
+      "loss": 0.0305,
+      "step": 2450
+    },
+    {
+      "epoch": 0.8319242475481907,
+      "grad_norm": 0.23499512672424316,
+      "learning_rate": 5.8453206972414966e-05,
+      "loss": 0.0342,
+      "step": 2460
+    },
+    {
+      "epoch": 0.835306053432533,
+      "grad_norm": 0.19104419648647308,
+      "learning_rate": 5.8283973599593846e-05,
+      "loss": 0.0341,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8386878593168752,
+      "grad_norm": 0.21760663390159607,
+      "learning_rate": 5.8114740226772726e-05,
+      "loss": 0.0287,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8420696652012174,
+      "grad_norm": 0.25679340958595276,
+      "learning_rate": 5.7945506853951606e-05,
+      "loss": 0.0356,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8454514710855597,
+      "grad_norm": 0.25008776783943176,
+      "learning_rate": 5.777627348113048e-05,
+      "loss": 0.0288,
+      "step": 2500
+    },
+    {
+      "epoch": 0.848833276969902,
+      "grad_norm": 0.25389808416366577,
+      "learning_rate": 5.760704010830936e-05,
+      "loss": 0.0316,
+      "step": 2510
+    },
+    {
+      "epoch": 0.8522150828542442,
+      "grad_norm": 0.2889176905155182,
+      "learning_rate": 5.743780673548824e-05,
+      "loss": 0.0333,
+      "step": 2520
+    },
+    {
+      "epoch": 0.8555968887385864,
+      "grad_norm": 0.2402850240468979,
+      "learning_rate": 5.726857336266712e-05,
+      "loss": 0.0313,
+      "step": 2530
+    },
+    {
+      "epoch": 0.8589786946229286,
+      "grad_norm": 0.217300683259964,
+      "learning_rate": 5.7099339989846e-05,
+      "loss": 0.0306,
+      "step": 2540
+    },
+    {
+      "epoch": 0.8623605005072709,
+      "grad_norm": 0.32617250084877014,
+      "learning_rate": 5.693010661702488e-05,
+      "loss": 0.0293,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8657423063916131,
+      "grad_norm": 0.2338167428970337,
+      "learning_rate": 5.676087324420376e-05,
+      "loss": 0.0275,
+      "step": 2560
+    },
+    {
+      "epoch": 0.8691241122759553,
+      "grad_norm": 0.2444329410791397,
+      "learning_rate": 5.659163987138264e-05,
+      "loss": 0.0281,
+      "step": 2570
+    },
+    {
+      "epoch": 0.8725059181602975,
+      "grad_norm": 0.2600383758544922,
+      "learning_rate": 5.642240649856152e-05,
+      "loss": 0.0293,
+      "step": 2580
+    },
+    {
+      "epoch": 0.8758877240446399,
+      "grad_norm": 0.19937843084335327,
+      "learning_rate": 5.62531731257404e-05,
+      "loss": 0.0327,
+      "step": 2590
+    },
+    {
+      "epoch": 0.8792695299289821,
+      "grad_norm": 0.25247564911842346,
+      "learning_rate": 5.608393975291928e-05,
+      "loss": 0.0353,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8826513358133243,
+      "grad_norm": 0.20916889607906342,
+      "learning_rate": 5.591470638009816e-05,
+      "loss": 0.0311,
+      "step": 2610
+    },
+    {
+      "epoch": 0.8860331416976666,
+      "grad_norm": 0.26659512519836426,
+      "learning_rate": 5.574547300727704e-05,
+      "loss": 0.0299,
+      "step": 2620
+    },
+    {
+      "epoch": 0.8894149475820088,
+      "grad_norm": 0.25376033782958984,
+      "learning_rate": 5.557623963445592e-05,
+      "loss": 0.03,
+      "step": 2630
+    },
+    {
+      "epoch": 0.892796753466351,
+      "grad_norm": 0.20384323596954346,
+      "learning_rate": 5.54070062616348e-05,
+      "loss": 0.0302,
+      "step": 2640
+    },
+    {
+      "epoch": 0.8961785593506932,
+      "grad_norm": 0.2751913070678711,
+      "learning_rate": 5.523777288881368e-05,
+      "loss": 0.0296,
+      "step": 2650
+    },
+    {
+      "epoch": 0.8995603652350355,
+      "grad_norm": 0.20069202780723572,
+      "learning_rate": 5.506853951599256e-05,
+      "loss": 0.0261,
+      "step": 2660
+    },
+    {
+      "epoch": 0.9029421711193778,
+      "grad_norm": 0.21684013307094574,
+      "learning_rate": 5.489930614317144e-05,
+      "loss": 0.0311,
+      "step": 2670
+    },
+    {
+      "epoch": 0.90632397700372,
+      "grad_norm": 0.23668934404850006,
+      "learning_rate": 5.473007277035032e-05,
+      "loss": 0.0303,
+      "step": 2680
+    },
+    {
+      "epoch": 0.9097057828880623,
+      "grad_norm": 0.24122962355613708,
+      "learning_rate": 5.456083939752919e-05,
+      "loss": 0.0294,
+      "step": 2690
+    },
+    {
+      "epoch": 0.9130875887724045,
+      "grad_norm": 0.19830548763275146,
+      "learning_rate": 5.439160602470807e-05,
+      "loss": 0.0264,
+      "step": 2700
+    },
+    {
+      "epoch": 0.9164693946567467,
+      "grad_norm": 0.20941804349422455,
+      "learning_rate": 5.422237265188695e-05,
+      "loss": 0.029,
+      "step": 2710
+    },
+    {
+      "epoch": 0.9198512005410889,
+      "grad_norm": 0.2681092619895935,
+      "learning_rate": 5.405313927906583e-05,
+      "loss": 0.0304,
+      "step": 2720
+    },
+    {
+      "epoch": 0.9232330064254312,
+      "grad_norm": 0.20790086686611176,
+      "learning_rate": 5.388390590624471e-05,
+      "loss": 0.027,
+      "step": 2730
+    },
+    {
+      "epoch": 0.9266148123097734,
+      "grad_norm": 0.227376326918602,
+      "learning_rate": 5.371467253342359e-05,
+      "loss": 0.0267,
+      "step": 2740
+    },
+    {
+      "epoch": 0.9299966181941156,
+      "grad_norm": 0.26523780822753906,
+      "learning_rate": 5.354543916060247e-05,
+      "loss": 0.0279,
+      "step": 2750
+    },
+    {
+      "epoch": 0.933378424078458,
+      "grad_norm": 0.21116450428962708,
+      "learning_rate": 5.337620578778135e-05,
+      "loss": 0.027,
+      "step": 2760
+    },
+    {
+      "epoch": 0.9367602299628002,
+      "grad_norm": 0.22488407790660858,
+      "learning_rate": 5.320697241496023e-05,
+      "loss": 0.0258,
+      "step": 2770
+    },
+    {
+      "epoch": 0.9401420358471424,
+      "grad_norm": 0.23635748028755188,
+      "learning_rate": 5.303773904213911e-05,
+      "loss": 0.0277,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9435238417314846,
+      "grad_norm": 0.1881859302520752,
+      "learning_rate": 5.286850566931799e-05,
+      "loss": 0.026,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9469056476158269,
+      "grad_norm": 0.28154221177101135,
+      "learning_rate": 5.269927229649687e-05,
+      "loss": 0.0269,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9502874535001691,
+      "grad_norm": 0.2318260818719864,
+      "learning_rate": 5.253003892367575e-05,
+      "loss": 0.0271,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9536692593845113,
+      "grad_norm": 0.2238345593214035,
+      "learning_rate": 5.236080555085463e-05,
+      "loss": 0.0253,
+      "step": 2820
+    },
+    {
+      "epoch": 0.9570510652688535,
+      "grad_norm": 0.2091434746980667,
+      "learning_rate": 5.219157217803351e-05,
+      "loss": 0.0251,
+      "step": 2830
+    },
+    {
+      "epoch": 0.9604328711531958,
+      "grad_norm": 0.2255164235830307,
+      "learning_rate": 5.202233880521239e-05,
+      "loss": 0.0281,
+      "step": 2840
+    },
+    {
+      "epoch": 0.9638146770375381,
+      "grad_norm": 0.245024174451828,
+      "learning_rate": 5.185310543239127e-05,
+      "loss": 0.0257,
+      "step": 2850
+    },
+    {
+      "epoch": 0.9671964829218803,
+      "grad_norm": 0.18670466542243958,
+      "learning_rate": 5.168387205957015e-05,
+      "loss": 0.027,
+      "step": 2860
+    },
+    {
+      "epoch": 0.9705782888062225,
+      "grad_norm": 0.22900165617465973,
+      "learning_rate": 5.1514638686749025e-05,
+      "loss": 0.0268,
+      "step": 2870
+    },
+    {
+      "epoch": 0.9739600946905648,
+      "grad_norm": 0.2123824506998062,
+      "learning_rate": 5.1345405313927905e-05,
+      "loss": 0.0255,
+      "step": 2880
+    },
+    {
+      "epoch": 0.977341900574907,
+      "grad_norm": 0.17296041548252106,
+      "learning_rate": 5.1176171941106785e-05,
+      "loss": 0.0256,
+      "step": 2890
+    },
+    {
+      "epoch": 0.9807237064592492,
+      "grad_norm": 0.21980832517147064,
+      "learning_rate": 5.1006938568285665e-05,
+      "loss": 0.027,
+      "step": 2900
+    },
+    {
+      "epoch": 0.9841055123435914,
+      "grad_norm": 0.21194496750831604,
+      "learning_rate": 5.0837705195464545e-05,
+      "loss": 0.0253,
+      "step": 2910
+    },
+    {
+      "epoch": 0.9874873182279337,
+      "grad_norm": 0.28237712383270264,
+      "learning_rate": 5.0668471822643425e-05,
+      "loss": 0.0261,
+      "step": 2920
+    },
+    {
+      "epoch": 0.9908691241122759,
+      "grad_norm": 0.22561867535114288,
+      "learning_rate": 5.0499238449822305e-05,
+      "loss": 0.0263,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9942509299966182,
+      "grad_norm": 0.19687789678573608,
+      "learning_rate": 5.0330005077001185e-05,
+      "loss": 0.0253,
+      "step": 2940
+    },
+    {
+      "epoch": 0.9976327358809605,
+      "grad_norm": 0.26815053820610046,
+      "learning_rate": 5.0160771704180065e-05,
+      "loss": 0.0315,
+      "step": 2950
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.025125615298748016,
+      "eval_runtime": 350.9983,
+      "eval_samples_per_second": 11.35,
+      "eval_steps_per_second": 0.356,
+      "step": 2957
+    },
+    {
+      "epoch": 1.0010145417653027,
+      "grad_norm": 0.17259906232357025,
+      "learning_rate": 4.9991538331358945e-05,
+      "loss": 0.0235,
+      "step": 2960
+    },
+    {
+      "epoch": 1.004396347649645,
+      "grad_norm": 0.23677606880664825,
+      "learning_rate": 4.9822304958537825e-05,
+      "loss": 0.0231,
+      "step": 2970
+    },
+    {
+      "epoch": 1.0077781535339871,
+      "grad_norm": 0.2753681540489197,
+      "learning_rate": 4.9653071585716705e-05,
+      "loss": 0.0225,
+      "step": 2980
+    },
+    {
+      "epoch": 1.0111599594183294,
+      "grad_norm": 0.1379474550485611,
+      "learning_rate": 4.9483838212895585e-05,
+      "loss": 0.0209,
+      "step": 2990
+    },
+    {
+      "epoch": 1.0145417653026716,
+      "grad_norm": 0.1725788563489914,
+      "learning_rate": 4.9314604840074465e-05,
+      "loss": 0.0231,
+      "step": 3000
+    },
+    {
+      "epoch": 1.0179235711870138,
+      "grad_norm": 0.16469170153141022,
+      "learning_rate": 4.9145371467253345e-05,
+      "loss": 0.0235,
+      "step": 3010
+    },
+    {
+      "epoch": 1.021305377071356,
+      "grad_norm": 0.1656704694032669,
+      "learning_rate": 4.8976138094432225e-05,
+      "loss": 0.0214,
+      "step": 3020
+    },
+    {
+      "epoch": 1.0246871829556983,
+      "grad_norm": 0.22718942165374756,
+      "learning_rate": 4.8806904721611104e-05,
+      "loss": 0.0235,
+      "step": 3030
+    },
+    {
+      "epoch": 1.0280689888400405,
+      "grad_norm": 0.18668334186077118,
+      "learning_rate": 4.8637671348789984e-05,
+      "loss": 0.0226,
+      "step": 3040
+    },
+    {
+      "epoch": 1.0314507947243827,
+      "grad_norm": 0.1626449078321457,
+      "learning_rate": 4.8468437975968864e-05,
+      "loss": 0.0211,
+      "step": 3050
+    },
+    {
+      "epoch": 1.0348326006087252,
+      "grad_norm": 0.14618806540966034,
+      "learning_rate": 4.829920460314774e-05,
+      "loss": 0.0224,
+      "step": 3060
+    },
+    {
+      "epoch": 1.0382144064930674,
+      "grad_norm": 0.20951628684997559,
+      "learning_rate": 4.812997123032662e-05,
+      "loss": 0.0225,
+      "step": 3070
+    },
+    {
+      "epoch": 1.0415962123774096,
+      "grad_norm": 0.15292339026927948,
+      "learning_rate": 4.79607378575055e-05,
+      "loss": 0.0236,
+      "step": 3080
+    },
+    {
+      "epoch": 1.0449780182617519,
+      "grad_norm": 0.23356632888317108,
+      "learning_rate": 4.779150448468438e-05,
+      "loss": 0.022,
+      "step": 3090
+    },
+    {
+      "epoch": 1.048359824146094,
+      "grad_norm": 0.17371255159378052,
+      "learning_rate": 4.762227111186326e-05,
+      "loss": 0.0221,
+      "step": 3100
+    },
+    {
+      "epoch": 1.0517416300304363,
+      "grad_norm": 0.20166213810443878,
+      "learning_rate": 4.745303773904214e-05,
+      "loss": 0.0227,
+      "step": 3110
+    },
+    {
+      "epoch": 1.0551234359147785,
+      "grad_norm": 0.18544991314411163,
+      "learning_rate": 4.728380436622102e-05,
+      "loss": 0.0227,
+      "step": 3120
+    },
+    {
+      "epoch": 1.0585052417991208,
+      "grad_norm": 0.23601798713207245,
+      "learning_rate": 4.71145709933999e-05,
+      "loss": 0.022,
+      "step": 3130
+    },
+    {
+      "epoch": 1.061887047683463,
+      "grad_norm": 0.20726293325424194,
+      "learning_rate": 4.6945337620578784e-05,
+      "loss": 0.021,
+      "step": 3140
+    },
+    {
+      "epoch": 1.0652688535678052,
+      "grad_norm": 0.20292699337005615,
+      "learning_rate": 4.6776104247757664e-05,
+      "loss": 0.0214,
+      "step": 3150
+    },
+    {
+      "epoch": 1.0686506594521474,
+      "grad_norm": 0.2641526162624359,
+      "learning_rate": 4.6606870874936544e-05,
+      "loss": 0.0222,
+      "step": 3160
+    },
+    {
+      "epoch": 1.0720324653364897,
+      "grad_norm": 0.19379054009914398,
+      "learning_rate": 4.6437637502115424e-05,
+      "loss": 0.0213,
+      "step": 3170
+    },
+    {
+      "epoch": 1.0754142712208319,
+      "grad_norm": 0.17844127118587494,
+      "learning_rate": 4.6268404129294304e-05,
+      "loss": 0.0247,
+      "step": 3180
+    },
+    {
+      "epoch": 1.0787960771051741,
+      "grad_norm": 0.19252558052539825,
+      "learning_rate": 4.6099170756473184e-05,
+      "loss": 0.0215,
+      "step": 3190
+    },
+    {
+      "epoch": 1.0821778829895163,
+      "grad_norm": 0.18375921249389648,
+      "learning_rate": 4.5929937383652064e-05,
+      "loss": 0.0225,
+      "step": 3200
+    },
+    {
+      "epoch": 1.0855596888738586,
+      "grad_norm": 0.20377720892429352,
+      "learning_rate": 4.576070401083094e-05,
+      "loss": 0.0218,
+      "step": 3210
+    },
+    {
+      "epoch": 1.0889414947582008,
+      "grad_norm": 0.17277295887470245,
+      "learning_rate": 4.559147063800982e-05,
+      "loss": 0.0203,
+      "step": 3220
+    },
+    {
+      "epoch": 1.092323300642543,
+      "grad_norm": 0.17021779716014862,
+      "learning_rate": 4.54222372651887e-05,
+      "loss": 0.0195,
+      "step": 3230
+    },
+    {
+      "epoch": 1.0957051065268852,
+      "grad_norm": 0.1649063676595688,
+      "learning_rate": 4.525300389236758e-05,
+      "loss": 0.0221,
+      "step": 3240
+    },
+    {
+      "epoch": 1.0990869124112277,
+      "grad_norm": 0.12819623947143555,
+      "learning_rate": 4.508377051954646e-05,
+      "loss": 0.0216,
+      "step": 3250
+    },
+    {
+      "epoch": 1.10246871829557,
+      "grad_norm": 0.20793533325195312,
+      "learning_rate": 4.491453714672534e-05,
+      "loss": 0.0205,
+      "step": 3260
+    },
+    {
+      "epoch": 1.1058505241799121,
+      "grad_norm": 0.23723694682121277,
+      "learning_rate": 4.474530377390422e-05,
+      "loss": 0.0208,
+      "step": 3270
+    },
+    {
+      "epoch": 1.1092323300642544,
+      "grad_norm": 0.20238593220710754,
+      "learning_rate": 4.45760704010831e-05,
+      "loss": 0.0212,
+      "step": 3280
+    },
+    {
+      "epoch": 1.1126141359485966,
+      "grad_norm": 0.1898476779460907,
+      "learning_rate": 4.440683702826198e-05,
+      "loss": 0.0215,
+      "step": 3290
+    },
+    {
+      "epoch": 1.1159959418329388,
+      "grad_norm": 0.1884363740682602,
+      "learning_rate": 4.423760365544086e-05,
+      "loss": 0.021,
+      "step": 3300
+    },
+    {
+      "epoch": 1.119377747717281,
+      "grad_norm": 0.1632470339536667,
+      "learning_rate": 4.406837028261974e-05,
+      "loss": 0.0213,
+      "step": 3310
+    },
+    {
+      "epoch": 1.1227595536016233,
+      "grad_norm": 0.1657821387052536,
+      "learning_rate": 4.389913690979862e-05,
+      "loss": 0.0203,
+      "step": 3320
+    },
+    {
+      "epoch": 1.1261413594859655,
+      "grad_norm": 0.12199438363313675,
+      "learning_rate": 4.37299035369775e-05,
+      "loss": 0.0202,
+      "step": 3330
+    },
+    {
+      "epoch": 1.1295231653703077,
+      "grad_norm": 0.16210108995437622,
+      "learning_rate": 4.356067016415638e-05,
+      "loss": 0.0205,
+      "step": 3340
+    },
+    {
+      "epoch": 1.13290497125465,
+      "grad_norm": 0.18802976608276367,
+      "learning_rate": 4.339143679133526e-05,
+      "loss": 0.0207,
+      "step": 3350
+    },
+    {
+      "epoch": 1.1362867771389922,
+      "grad_norm": 0.1435530036687851,
+      "learning_rate": 4.322220341851414e-05,
+      "loss": 0.0215,
+      "step": 3360
+    },
+    {
+      "epoch": 1.1396685830233344,
+      "grad_norm": 0.25706106424331665,
+      "learning_rate": 4.305297004569302e-05,
+      "loss": 0.022,
+      "step": 3370
+    },
+    {
+      "epoch": 1.1430503889076766,
+      "grad_norm": 0.19478261470794678,
+      "learning_rate": 4.28837366728719e-05,
+      "loss": 0.0207,
+      "step": 3380
+    },
+    {
+      "epoch": 1.1464321947920189,
+      "grad_norm": 0.14196892082691193,
+      "learning_rate": 4.271450330005077e-05,
+      "loss": 0.0203,
+      "step": 3390
+    },
+    {
+      "epoch": 1.149814000676361,
+      "grad_norm": 0.15680982172489166,
+      "learning_rate": 4.254526992722965e-05,
+      "loss": 0.0213,
+      "step": 3400
+    },
+    {
+      "epoch": 1.1531958065607033,
+      "grad_norm": 0.1810910403728485,
+      "learning_rate": 4.237603655440853e-05,
+      "loss": 0.0213,
+      "step": 3410
+    },
+    {
+      "epoch": 1.1565776124450458,
+      "grad_norm": 0.13093525171279907,
+      "learning_rate": 4.220680318158741e-05,
+      "loss": 0.0193,
+      "step": 3420
+    },
+    {
+      "epoch": 1.159959418329388,
+      "grad_norm": 0.1480521410703659,
+      "learning_rate": 4.203756980876629e-05,
+      "loss": 0.0192,
+      "step": 3430
+    },
+    {
+      "epoch": 1.1633412242137302,
+      "grad_norm": 0.20289231836795807,
+      "learning_rate": 4.186833643594517e-05,
+      "loss": 0.0189,
+      "step": 3440
+    },
+    {
+      "epoch": 1.1667230300980724,
+      "grad_norm": 0.15804171562194824,
+      "learning_rate": 4.169910306312405e-05,
+      "loss": 0.0204,
+      "step": 3450
+    },
+    {
+      "epoch": 1.1701048359824147,
+      "grad_norm": 0.12550042569637299,
+      "learning_rate": 4.152986969030293e-05,
+      "loss": 0.0193,
+      "step": 3460
+    },
+    {
+      "epoch": 1.1734866418667569,
+      "grad_norm": 0.153706893324852,
+      "learning_rate": 4.136063631748181e-05,
+      "loss": 0.0204,
+      "step": 3470
+    },
+    {
+      "epoch": 1.1768684477510991,
+      "grad_norm": 0.17369280755519867,
+      "learning_rate": 4.119140294466069e-05,
+      "loss": 0.0194,
+      "step": 3480
+    },
+    {
+      "epoch": 1.1802502536354413,
+      "grad_norm": 0.20219433307647705,
+      "learning_rate": 4.102216957183957e-05,
+      "loss": 0.0234,
+      "step": 3490
+    },
+    {
+      "epoch": 1.1836320595197836,
+      "grad_norm": 0.14274387061595917,
+      "learning_rate": 4.085293619901845e-05,
+      "loss": 0.0201,
+      "step": 3500
+    },
+    {
+      "epoch": 1.1870138654041258,
+      "grad_norm": 0.1277850866317749,
+      "learning_rate": 4.068370282619733e-05,
+      "loss": 0.0197,
+      "step": 3510
+    },
+    {
+      "epoch": 1.190395671288468,
+      "grad_norm": 0.12567327916622162,
+      "learning_rate": 4.051446945337621e-05,
+      "loss": 0.0193,
+      "step": 3520
+    },
+    {
+      "epoch": 1.1937774771728102,
+      "grad_norm": 0.18639731407165527,
+      "learning_rate": 4.034523608055509e-05,
+      "loss": 0.0208,
+      "step": 3530
+    },
+    {
+      "epoch": 1.1971592830571525,
+      "grad_norm": 0.17870990931987762,
+      "learning_rate": 4.017600270773397e-05,
+      "loss": 0.0207,
+      "step": 3540
+    },
+    {
+      "epoch": 1.2005410889414947,
+      "grad_norm": 0.1992855817079544,
+      "learning_rate": 4.000676933491285e-05,
+      "loss": 0.0197,
+      "step": 3550
+    },
+    {
+      "epoch": 1.203922894825837,
+      "grad_norm": 0.19009998440742493,
+      "learning_rate": 3.983753596209173e-05,
+      "loss": 0.0193,
+      "step": 3560
+    },
+    {
+      "epoch": 1.2073047007101791,
+      "grad_norm": 0.2152644246816635,
+      "learning_rate": 3.966830258927061e-05,
+      "loss": 0.0193,
+      "step": 3570
+    },
+    {
+      "epoch": 1.2106865065945214,
+      "grad_norm": 0.17644502222537994,
+      "learning_rate": 3.949906921644948e-05,
+      "loss": 0.0189,
+      "step": 3580
+    },
+    {
+      "epoch": 1.2140683124788638,
+      "grad_norm": 0.16940535604953766,
+      "learning_rate": 3.932983584362836e-05,
+      "loss": 0.0195,
+      "step": 3590
+    },
+    {
+      "epoch": 1.217450118363206,
+      "grad_norm": 0.1200951337814331,
+      "learning_rate": 3.916060247080724e-05,
+      "loss": 0.0205,
+      "step": 3600
+    },
+    {
+      "epoch": 1.2208319242475483,
+      "grad_norm": 0.12686198949813843,
+      "learning_rate": 3.899136909798612e-05,
+      "loss": 0.0208,
+      "step": 3610
+    },
+    {
+      "epoch": 1.2242137301318905,
+      "grad_norm": 0.16317637264728546,
+      "learning_rate": 3.8822135725165e-05,
+      "loss": 0.0198,
+      "step": 3620
+    },
+    {
+      "epoch": 1.2275955360162327,
+      "grad_norm": 0.16301840543746948,
+      "learning_rate": 3.865290235234388e-05,
+      "loss": 0.0196,
+      "step": 3630
+    },
+    {
+      "epoch": 1.230977341900575,
+      "grad_norm": 0.14271463453769684,
+      "learning_rate": 3.848366897952276e-05,
+      "loss": 0.0198,
+      "step": 3640
+    },
+    {
+      "epoch": 1.2343591477849172,
+      "grad_norm": 0.13855944573879242,
+      "learning_rate": 3.831443560670164e-05,
+      "loss": 0.0192,
+      "step": 3650
+    },
+    {
+      "epoch": 1.2377409536692594,
+      "grad_norm": 0.16393686830997467,
+      "learning_rate": 3.814520223388052e-05,
+      "loss": 0.0181,
+      "step": 3660
+    },
+    {
+      "epoch": 1.2411227595536016,
+      "grad_norm": 0.13381467759609222,
+      "learning_rate": 3.79759688610594e-05,
+      "loss": 0.0199,
+      "step": 3670
+    },
+    {
+      "epoch": 1.2445045654379439,
+      "grad_norm": 0.20619070529937744,
+      "learning_rate": 3.780673548823828e-05,
+      "loss": 0.02,
+      "step": 3680
+    },
+    {
+      "epoch": 1.247886371322286,
+      "grad_norm": 0.1646890640258789,
+      "learning_rate": 3.763750211541716e-05,
+      "loss": 0.0207,
+      "step": 3690
+    },
+    {
+      "epoch": 1.2512681772066283,
+      "grad_norm": 0.18202926218509674,
+      "learning_rate": 3.746826874259604e-05,
+      "loss": 0.0194,
+      "step": 3700
+    },
+    {
+      "epoch": 1.2546499830909705,
+      "grad_norm": 0.16690859198570251,
+      "learning_rate": 3.729903536977492e-05,
+      "loss": 0.0198,
+      "step": 3710
+    },
+    {
+      "epoch": 1.2580317889753128,
+      "grad_norm": 0.14021560549736023,
+      "learning_rate": 3.71298019969538e-05,
+      "loss": 0.019,
+      "step": 3720
+    },
+    {
+      "epoch": 1.261413594859655,
+      "grad_norm": 0.1812976896762848,
+      "learning_rate": 3.696056862413268e-05,
+      "loss": 0.0202,
+      "step": 3730
+    },
+    {
+      "epoch": 1.2647954007439974,
+      "grad_norm": 0.19519537687301636,
+      "learning_rate": 3.679133525131156e-05,
+      "loss": 0.0201,
+      "step": 3740
+    },
+    {
+      "epoch": 1.2681772066283394,
+      "grad_norm": 0.1934451311826706,
+      "learning_rate": 3.662210187849044e-05,
+      "loss": 0.0191,
+      "step": 3750
+    },
+    {
+      "epoch": 1.2715590125126819,
+      "grad_norm": 0.16278310120105743,
+      "learning_rate": 3.645286850566932e-05,
+      "loss": 0.0198,
+      "step": 3760
+    },
+    {
+      "epoch": 1.274940818397024,
+      "grad_norm": 0.13240766525268555,
+      "learning_rate": 3.6283635132848195e-05,
+      "loss": 0.0185,
+      "step": 3770
+    },
+    {
+      "epoch": 1.2783226242813663,
+      "grad_norm": 0.1842852085828781,
+      "learning_rate": 3.6114401760027075e-05,
+      "loss": 0.0192,
+      "step": 3780
+    },
+    {
+      "epoch": 1.2817044301657086,
+      "grad_norm": 0.15183594822883606,
+      "learning_rate": 3.5945168387205955e-05,
+      "loss": 0.0208,
+      "step": 3790
+    },
+    {
+      "epoch": 1.2850862360500508,
+      "grad_norm": 0.131156325340271,
+      "learning_rate": 3.5775935014384835e-05,
+      "loss": 0.0196,
+      "step": 3800
+    },
+    {
+      "epoch": 1.288468041934393,
+      "grad_norm": 0.12825001776218414,
+      "learning_rate": 3.5606701641563715e-05,
+      "loss": 0.0192,
+      "step": 3810
+    },
+    {
+      "epoch": 1.2918498478187352,
+      "grad_norm": 0.1289099007844925,
+      "learning_rate": 3.5437468268742595e-05,
+      "loss": 0.0193,
+      "step": 3820
+    },
+    {
+      "epoch": 1.2952316537030775,
+      "grad_norm": 0.18116918206214905,
+      "learning_rate": 3.5268234895921475e-05,
+      "loss": 0.0181,
+      "step": 3830
+    },
+    {
+      "epoch": 1.2986134595874197,
+      "grad_norm": 0.1824859082698822,
+      "learning_rate": 3.5099001523100355e-05,
+      "loss": 0.0189,
+      "step": 3840
+    },
+    {
+      "epoch": 1.301995265471762,
+      "grad_norm": 0.18777935206890106,
+      "learning_rate": 3.4929768150279235e-05,
+      "loss": 0.0184,
+      "step": 3850
+    },
+    {
+      "epoch": 1.3053770713561041,
+      "grad_norm": 0.11837860941886902,
+      "learning_rate": 3.4760534777458115e-05,
+      "loss": 0.0191,
+      "step": 3860
+    },
+    {
+      "epoch": 1.3087588772404464,
+      "grad_norm": 0.13773775100708008,
+      "learning_rate": 3.4591301404636995e-05,
+      "loss": 0.0189,
+      "step": 3870
+    },
+    {
+      "epoch": 1.3121406831247886,
+      "grad_norm": 0.148502916097641,
+      "learning_rate": 3.4422068031815875e-05,
+      "loss": 0.0184,
+      "step": 3880
+    },
+    {
+      "epoch": 1.3155224890091308,
+      "grad_norm": 0.12410616874694824,
+      "learning_rate": 3.4252834658994755e-05,
+      "loss": 0.018,
+      "step": 3890
+    },
+    {
+      "epoch": 1.318904294893473,
+      "grad_norm": 0.12317556142807007,
+      "learning_rate": 3.4083601286173635e-05,
+      "loss": 0.0177,
+      "step": 3900
+    },
+    {
+      "epoch": 1.3222861007778153,
+      "grad_norm": 0.133961483836174,
+      "learning_rate": 3.3914367913352515e-05,
+      "loss": 0.0184,
+      "step": 3910
+    },
+    {
+      "epoch": 1.3256679066621575,
+      "grad_norm": 0.1317528784275055,
+      "learning_rate": 3.3745134540531395e-05,
+      "loss": 0.02,
+      "step": 3920
+    },
+    {
+      "epoch": 1.3290497125465,
+      "grad_norm": 0.14328566193580627,
+      "learning_rate": 3.3575901167710275e-05,
+      "loss": 0.0185,
+      "step": 3930
+    },
+    {
+      "epoch": 1.332431518430842,
+      "grad_norm": 0.1426452249288559,
+      "learning_rate": 3.3406667794889155e-05,
+      "loss": 0.0187,
+      "step": 3940
+    },
+    {
+      "epoch": 1.3358133243151844,
+      "grad_norm": 0.13055744767189026,
+      "learning_rate": 3.323743442206803e-05,
+      "loss": 0.0178,
+      "step": 3950
+    },
+    {
+      "epoch": 1.3391951301995266,
+      "grad_norm": 0.14115405082702637,
+      "learning_rate": 3.306820104924691e-05,
+      "loss": 0.0201,
+      "step": 3960
+    },
+    {
+      "epoch": 1.3425769360838689,
+      "grad_norm": 0.11101550608873367,
+      "learning_rate": 3.289896767642579e-05,
+      "loss": 0.0177,
+      "step": 3970
+    },
+    {
+      "epoch": 1.345958741968211,
+      "grad_norm": 0.10724030435085297,
+      "learning_rate": 3.272973430360467e-05,
+      "loss": 0.0197,
+      "step": 3980
+    },
+    {
+      "epoch": 1.3493405478525533,
+      "grad_norm": 0.15012751519680023,
+      "learning_rate": 3.256050093078355e-05,
+      "loss": 0.0184,
+      "step": 3990
+    },
+    {
+      "epoch": 1.3527223537368955,
+      "grad_norm": 0.16174957156181335,
+      "learning_rate": 3.239126755796243e-05,
+      "loss": 0.0187,
+      "step": 4000
+    },
+    {
+      "epoch": 1.3561041596212378,
+      "grad_norm": 0.12000412493944168,
+      "learning_rate": 3.222203418514131e-05,
+      "loss": 0.0176,
+      "step": 4010
+    },
+    {
+      "epoch": 1.35948596550558,
+      "grad_norm": 0.14733338356018066,
+      "learning_rate": 3.205280081232019e-05,
+      "loss": 0.0177,
+      "step": 4020
+    },
+    {
+      "epoch": 1.3628677713899222,
+      "grad_norm": 0.12138606607913971,
+      "learning_rate": 3.188356743949907e-05,
+      "loss": 0.0176,
+      "step": 4030
+    },
+    {
+      "epoch": 1.3662495772742644,
+      "grad_norm": 0.1111988052725792,
+      "learning_rate": 3.1714334066677954e-05,
+      "loss": 0.0177,
+      "step": 4040
+    },
+    {
+      "epoch": 1.3696313831586067,
+      "grad_norm": 0.22300873696804047,
+      "learning_rate": 3.1545100693856834e-05,
+      "loss": 0.0182,
+      "step": 4050
+    },
+    {
+      "epoch": 1.373013189042949,
+      "grad_norm": 0.11778800189495087,
+      "learning_rate": 3.1375867321035714e-05,
+      "loss": 0.0176,
+      "step": 4060
+    },
+    {
+      "epoch": 1.3763949949272911,
+      "grad_norm": 0.14025694131851196,
+      "learning_rate": 3.1206633948214594e-05,
+      "loss": 0.018,
+      "step": 4070
+    },
+    {
+      "epoch": 1.3797768008116333,
+      "grad_norm": 0.22355903685092926,
+      "learning_rate": 3.1037400575393474e-05,
+      "loss": 0.02,
+      "step": 4080
+    },
+    {
+      "epoch": 1.3831586066959756,
+      "grad_norm": 0.15728473663330078,
+      "learning_rate": 3.0868167202572354e-05,
+      "loss": 0.0175,
+      "step": 4090
+    },
+    {
+      "epoch": 1.386540412580318,
+      "grad_norm": 0.13040538132190704,
+      "learning_rate": 3.069893382975123e-05,
+      "loss": 0.0177,
+      "step": 4100
+    },
+    {
+      "epoch": 1.38992221846466,
+      "grad_norm": 0.13674882054328918,
+      "learning_rate": 3.052970045693011e-05,
+      "loss": 0.0189,
+      "step": 4110
+    },
+    {
+      "epoch": 1.3933040243490025,
+      "grad_norm": 0.13062675297260284,
+      "learning_rate": 3.036046708410899e-05,
+      "loss": 0.0174,
+      "step": 4120
+    },
+    {
+      "epoch": 1.3966858302333445,
+      "grad_norm": 0.07586494088172913,
+      "learning_rate": 3.019123371128787e-05,
+      "loss": 0.0174,
+      "step": 4130
+    },
+    {
+      "epoch": 1.400067636117687,
+      "grad_norm": 0.12583595514297485,
+      "learning_rate": 3.002200033846675e-05,
+      "loss": 0.0174,
+      "step": 4140
+    },
+    {
+      "epoch": 1.4034494420020291,
+      "grad_norm": 0.11460530757904053,
+      "learning_rate": 2.9852766965645627e-05,
+      "loss": 0.017,
+      "step": 4150
+    },
+    {
+      "epoch": 1.4068312478863714,
+      "grad_norm": 0.14381511509418488,
+      "learning_rate": 2.9683533592824507e-05,
+      "loss": 0.0189,
+      "step": 4160
+    },
+    {
+      "epoch": 1.4102130537707136,
+      "grad_norm": 0.12002012878656387,
+      "learning_rate": 2.9514300220003387e-05,
+      "loss": 0.0172,
+      "step": 4170
+    },
+    {
+      "epoch": 1.4135948596550558,
+      "grad_norm": 0.1379200518131256,
+      "learning_rate": 2.9345066847182267e-05,
+      "loss": 0.018,
+      "step": 4180
+    },
+    {
+      "epoch": 1.416976665539398,
+      "grad_norm": 0.098530113697052,
+      "learning_rate": 2.9175833474361147e-05,
+      "loss": 0.0172,
+      "step": 4190
+    },
+    {
+      "epoch": 1.4203584714237403,
+      "grad_norm": 0.11576182395219803,
+      "learning_rate": 2.9006600101540027e-05,
+      "loss": 0.0187,
+      "step": 4200
+    },
+    {
+      "epoch": 1.4237402773080825,
+      "grad_norm": 0.15769566595554352,
+      "learning_rate": 2.8837366728718907e-05,
+      "loss": 0.0179,
+      "step": 4210
+    },
+    {
+      "epoch": 1.4271220831924247,
+      "grad_norm": 0.1257631480693817,
+      "learning_rate": 2.8668133355897787e-05,
+      "loss": 0.0174,
+      "step": 4220
+    },
+    {
+      "epoch": 1.430503889076767,
+      "grad_norm": 0.09543973207473755,
+      "learning_rate": 2.8498899983076667e-05,
+      "loss": 0.0174,
+      "step": 4230
+    },
+    {
+      "epoch": 1.4338856949611092,
+      "grad_norm": 0.12915478646755219,
+      "learning_rate": 2.8329666610255544e-05,
+      "loss": 0.0169,
+      "step": 4240
+    },
+    {
+      "epoch": 1.4372675008454514,
+      "grad_norm": 0.14382801949977875,
+      "learning_rate": 2.8160433237434424e-05,
+      "loss": 0.0169,
+      "step": 4250
+    },
+    {
+      "epoch": 1.4406493067297936,
+      "grad_norm": 0.1201377660036087,
+      "learning_rate": 2.7991199864613304e-05,
+      "loss": 0.0167,
+      "step": 4260
+    },
+    {
+      "epoch": 1.444031112614136,
+      "grad_norm": 0.11754725128412247,
+      "learning_rate": 2.7821966491792184e-05,
+      "loss": 0.0213,
+      "step": 4270
+    },
+    {
+      "epoch": 1.447412918498478,
+      "grad_norm": 0.0879068523645401,
+      "learning_rate": 2.7652733118971064e-05,
+      "loss": 0.0175,
+      "step": 4280
+    },
+    {
+      "epoch": 1.4507947243828205,
+      "grad_norm": 0.10662252455949783,
+      "learning_rate": 2.7483499746149944e-05,
+      "loss": 0.019,
+      "step": 4290
+    },
+    {
+      "epoch": 1.4541765302671625,
+      "grad_norm": 0.09592264890670776,
+      "learning_rate": 2.7314266373328824e-05,
+      "loss": 0.0182,
+      "step": 4300
+    },
+    {
+      "epoch": 1.457558336151505,
+      "grad_norm": 0.19000816345214844,
+      "learning_rate": 2.7145033000507703e-05,
+      "loss": 0.0181,
+      "step": 4310
+    },
+    {
+      "epoch": 1.4609401420358472,
+      "grad_norm": 0.18730321526527405,
+      "learning_rate": 2.6975799627686583e-05,
+      "loss": 0.017,
+      "step": 4320
+    },
+    {
+      "epoch": 1.4643219479201894,
+      "grad_norm": 0.11948239058256149,
+      "learning_rate": 2.680656625486546e-05,
+      "loss": 0.0163,
+      "step": 4330
+    },
+    {
+      "epoch": 1.4677037538045317,
+      "grad_norm": 0.11915738880634308,
+      "learning_rate": 2.663733288204434e-05,
+      "loss": 0.0176,
+      "step": 4340
+    },
+    {
+      "epoch": 1.471085559688874,
+      "grad_norm": 0.11469963937997818,
+      "learning_rate": 2.646809950922322e-05,
+      "loss": 0.0174,
+      "step": 4350
+    },
+    {
+      "epoch": 1.4744673655732161,
+      "grad_norm": 0.13674333691596985,
+      "learning_rate": 2.62988661364021e-05,
+      "loss": 0.0169,
+      "step": 4360
+    },
+    {
+      "epoch": 1.4778491714575583,
+      "grad_norm": 0.1072223111987114,
+      "learning_rate": 2.612963276358098e-05,
+      "loss": 0.017,
+      "step": 4370
+    },
+    {
+      "epoch": 1.4812309773419006,
+      "grad_norm": 0.12718436121940613,
+      "learning_rate": 2.596039939075986e-05,
+      "loss": 0.0185,
+      "step": 4380
+    },
+    {
+      "epoch": 1.4846127832262428,
+      "grad_norm": 0.09312810003757477,
+      "learning_rate": 2.579116601793874e-05,
+      "loss": 0.0167,
+      "step": 4390
+    },
+    {
+      "epoch": 1.487994589110585,
+      "grad_norm": 0.1793229579925537,
+      "learning_rate": 2.562193264511762e-05,
+      "loss": 0.0177,
+      "step": 4400
+    },
+    {
+      "epoch": 1.4913763949949272,
+      "grad_norm": 0.10192491114139557,
+      "learning_rate": 2.54526992722965e-05,
+      "loss": 0.0163,
+      "step": 4410
+    },
+    {
+      "epoch": 1.4947582008792695,
+      "grad_norm": 0.10123202204704285,
+      "learning_rate": 2.528346589947538e-05,
+      "loss": 0.017,
+      "step": 4420
+    },
+    {
+      "epoch": 1.4981400067636117,
+      "grad_norm": 0.13236495852470398,
+      "learning_rate": 2.5114232526654256e-05,
+      "loss": 0.0172,
+      "step": 4430
+    },
+    {
+      "epoch": 1.5015218126479541,
+      "grad_norm": 0.11559689790010452,
+      "learning_rate": 2.4944999153833136e-05,
+      "loss": 0.018,
+      "step": 4440
+    },
+    {
+      "epoch": 1.5049036185322962,
+      "grad_norm": 0.1355433166027069,
+      "learning_rate": 2.4775765781012016e-05,
+      "loss": 0.0177,
+      "step": 4450
+    },
+    {
+      "epoch": 1.5082854244166386,
+      "grad_norm": 0.09492843598127365,
+      "learning_rate": 2.4606532408190896e-05,
+      "loss": 0.0166,
+      "step": 4460
+    },
+    {
+      "epoch": 1.5116672303009806,
+      "grad_norm": 0.14215636253356934,
+      "learning_rate": 2.4437299035369776e-05,
+      "loss": 0.0169,
+      "step": 4470
+    },
+    {
+      "epoch": 1.515049036185323,
+      "grad_norm": 0.10210006684064865,
+      "learning_rate": 2.4268065662548656e-05,
+      "loss": 0.0168,
+      "step": 4480
+    },
+    {
+      "epoch": 1.518430842069665,
+      "grad_norm": 0.14845940470695496,
+      "learning_rate": 2.4098832289727536e-05,
+      "loss": 0.0179,
+      "step": 4490
+    },
+    {
+      "epoch": 1.5218126479540075,
+      "grad_norm": 0.18286077678203583,
+      "learning_rate": 2.3929598916906416e-05,
+      "loss": 0.0172,
+      "step": 4500
+    },
+    {
+      "epoch": 1.5251944538383497,
+      "grad_norm": 0.11884652823209763,
+      "learning_rate": 2.3760365544085296e-05,
+      "loss": 0.0161,
+      "step": 4510
+    },
+    {
+      "epoch": 1.528576259722692,
+      "grad_norm": 0.1169377937912941,
+      "learning_rate": 2.3591132171264173e-05,
+      "loss": 0.0168,
+      "step": 4520
+    },
+    {
+      "epoch": 1.5319580656070342,
+      "grad_norm": 0.10672768950462341,
+      "learning_rate": 2.3421898798443053e-05,
+      "loss": 0.017,
+      "step": 4530
+    },
+    {
+      "epoch": 1.5353398714913764,
+      "grad_norm": 0.06240490451455116,
+      "learning_rate": 2.3252665425621933e-05,
+      "loss": 0.0173,
+      "step": 4540
+    },
+    {
+      "epoch": 1.5387216773757186,
+      "grad_norm": 0.12996572256088257,
+      "learning_rate": 2.3083432052800813e-05,
+      "loss": 0.0164,
+      "step": 4550
+    },
+    {
+      "epoch": 1.5421034832600609,
+      "grad_norm": 0.10913591086864471,
+      "learning_rate": 2.2914198679979693e-05,
+      "loss": 0.0166,
+      "step": 4560
+    },
+    {
+      "epoch": 1.545485289144403,
+      "grad_norm": 0.09768196195363998,
+      "learning_rate": 2.2744965307158573e-05,
+      "loss": 0.0173,
+      "step": 4570
+    },
+    {
+      "epoch": 1.5488670950287453,
+      "grad_norm": 0.06998474895954132,
+      "learning_rate": 2.2575731934337453e-05,
+      "loss": 0.017,
+      "step": 4580
+    },
+    {
+      "epoch": 1.5522489009130875,
+      "grad_norm": 0.07349732518196106,
+      "learning_rate": 2.2406498561516333e-05,
+      "loss": 0.0212,
+      "step": 4590
+    },
+    {
+      "epoch": 1.5556307067974298,
+      "grad_norm": 0.13526155054569244,
+      "learning_rate": 2.2237265188695212e-05,
+      "loss": 0.016,
+      "step": 4600
+    },
+    {
+      "epoch": 1.5590125126817722,
+      "grad_norm": 0.12724880874156952,
+      "learning_rate": 2.206803181587409e-05,
+      "loss": 0.0171,
+      "step": 4610
+    },
+    {
+      "epoch": 1.5623943185661142,
+      "grad_norm": 0.08398682624101639,
+      "learning_rate": 2.189879844305297e-05,
+      "loss": 0.0168,
+      "step": 4620
+    },
+    {
+      "epoch": 1.5657761244504567,
+      "grad_norm": 0.0917474552989006,
+      "learning_rate": 2.172956507023185e-05,
+      "loss": 0.0179,
+      "step": 4630
+    },
+    {
+      "epoch": 1.5691579303347987,
+      "grad_norm": 0.12279586493968964,
+      "learning_rate": 2.156033169741073e-05,
+      "loss": 0.0166,
+      "step": 4640
+    },
+    {
+      "epoch": 1.5725397362191411,
+      "grad_norm": 0.14498373866081238,
+      "learning_rate": 2.139109832458961e-05,
+      "loss": 0.0174,
+      "step": 4650
+    },
+    {
+      "epoch": 1.5759215421034831,
+      "grad_norm": 0.1070031225681305,
+      "learning_rate": 2.122186495176849e-05,
+      "loss": 0.0164,
+      "step": 4660
+    },
+    {
+      "epoch": 1.5793033479878256,
+      "grad_norm": 0.0644792914390564,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.0164,
+      "step": 4670
+    },
+    {
+      "epoch": 1.5826851538721676,
+      "grad_norm": 0.13356371223926544,
+      "learning_rate": 2.088339820612625e-05,
+      "loss": 0.0171,
+      "step": 4680
+    },
+    {
+      "epoch": 1.58606695975651,
+      "grad_norm": 0.061863575130701065,
+      "learning_rate": 2.071416483330513e-05,
+      "loss": 0.0158,
+      "step": 4690
+    },
+    {
+      "epoch": 1.5894487656408522,
+      "grad_norm": 0.15974536538124084,
+      "learning_rate": 2.0544931460484005e-05,
+      "loss": 0.0168,
+      "step": 4700
+    },
+    {
+      "epoch": 1.5928305715251945,
+      "grad_norm": 0.10037930309772491,
+      "learning_rate": 2.037569808766289e-05,
+      "loss": 0.016,
+      "step": 4710
+    },
+    {
+      "epoch": 1.5962123774095367,
+      "grad_norm": 0.09223894774913788,
+      "learning_rate": 2.020646471484177e-05,
+      "loss": 0.0159,
+      "step": 4720
+    },
+    {
+      "epoch": 1.599594183293879,
+      "grad_norm": 0.08334498107433319,
+      "learning_rate": 2.003723134202065e-05,
+      "loss": 0.0169,
+      "step": 4730
+    },
+    {
+      "epoch": 1.6029759891782212,
+      "grad_norm": 0.127294659614563,
+      "learning_rate": 1.986799796919953e-05,
+      "loss": 0.0168,
+      "step": 4740
+    },
+    {
+      "epoch": 1.6063577950625634,
+      "grad_norm": 0.0788227841258049,
+      "learning_rate": 1.969876459637841e-05,
+      "loss": 0.0159,
+      "step": 4750
+    },
+    {
+      "epoch": 1.6097396009469056,
+      "grad_norm": 0.11868683248758316,
+      "learning_rate": 1.952953122355729e-05,
+      "loss": 0.0156,
+      "step": 4760
+    },
+    {
+      "epoch": 1.6131214068312478,
+      "grad_norm": 0.09021388739347458,
+      "learning_rate": 1.936029785073617e-05,
+      "loss": 0.0169,
+      "step": 4770
+    },
+    {
+      "epoch": 1.6165032127155903,
+      "grad_norm": 0.10817821323871613,
+      "learning_rate": 1.9191064477915045e-05,
+      "loss": 0.0168,
+      "step": 4780
+    },
+    {
+      "epoch": 1.6198850185999323,
+      "grad_norm": 0.09980209171772003,
+      "learning_rate": 1.9021831105093925e-05,
+      "loss": 0.0163,
+      "step": 4790
+    },
+    {
+      "epoch": 1.6232668244842747,
+      "grad_norm": 0.09977928549051285,
+      "learning_rate": 1.8852597732272805e-05,
+      "loss": 0.0165,
+      "step": 4800
+    },
+    {
+      "epoch": 1.6266486303686167,
+      "grad_norm": 0.07304428517818451,
+      "learning_rate": 1.8683364359451685e-05,
+      "loss": 0.0159,
+      "step": 4810
+    },
+    {
+      "epoch": 1.6300304362529592,
+      "grad_norm": 0.10551423579454422,
+      "learning_rate": 1.8514130986630565e-05,
+      "loss": 0.0167,
+      "step": 4820
+    },
+    {
+      "epoch": 1.6334122421373012,
+      "grad_norm": 0.06610710173845291,
+      "learning_rate": 1.8344897613809445e-05,
+      "loss": 0.0164,
+      "step": 4830
+    },
+    {
+      "epoch": 1.6367940480216436,
+      "grad_norm": 0.19381451606750488,
+      "learning_rate": 1.8175664240988325e-05,
+      "loss": 0.0166,
+      "step": 4840
+    },
+    {
+      "epoch": 1.6401758539059856,
+      "grad_norm": 0.09147250652313232,
+      "learning_rate": 1.8006430868167205e-05,
+      "loss": 0.0163,
+      "step": 4850
+    },
+    {
+      "epoch": 1.643557659790328,
+      "grad_norm": 0.1002357229590416,
+      "learning_rate": 1.7837197495346085e-05,
+      "loss": 0.0164,
+      "step": 4860
+    },
+    {
+      "epoch": 1.6469394656746703,
+      "grad_norm": 0.08374287933111191,
+      "learning_rate": 1.766796412252496e-05,
+      "loss": 0.0168,
+      "step": 4870
+    },
+    {
+      "epoch": 1.6503212715590125,
+      "grad_norm": 0.11019756644964218,
+      "learning_rate": 1.749873074970384e-05,
+      "loss": 0.0175,
+      "step": 4880
+    },
+    {
+      "epoch": 1.6537030774433548,
+      "grad_norm": 0.06535135954618454,
+      "learning_rate": 1.732949737688272e-05,
+      "loss": 0.0157,
+      "step": 4890
+    },
+    {
+      "epoch": 1.657084883327697,
+      "grad_norm": 0.0966026782989502,
+      "learning_rate": 1.71602640040616e-05,
+      "loss": 0.0159,
+      "step": 4900
+    },
+    {
+      "epoch": 1.6604666892120392,
+      "grad_norm": 0.07409293204545975,
+      "learning_rate": 1.699103063124048e-05,
+      "loss": 0.0168,
+      "step": 4910
+    },
+    {
+      "epoch": 1.6638484950963814,
+      "grad_norm": 0.13448843359947205,
+      "learning_rate": 1.682179725841936e-05,
+      "loss": 0.017,
+      "step": 4920
+    },
+    {
+      "epoch": 1.6672303009807237,
+      "grad_norm": 0.11097908765077591,
+      "learning_rate": 1.665256388559824e-05,
+      "loss": 0.0164,
+      "step": 4930
+    },
+    {
+      "epoch": 1.670612106865066,
+      "grad_norm": 0.08099315315485,
+      "learning_rate": 1.648333051277712e-05,
+      "loss": 0.0169,
+      "step": 4940
+    },
+    {
+      "epoch": 1.6739939127494083,
+      "grad_norm": 0.10239287465810776,
+      "learning_rate": 1.6314097139956e-05,
+      "loss": 0.0168,
+      "step": 4950
+    },
+    {
+      "epoch": 1.6773757186337503,
+      "grad_norm": 0.07663831859827042,
+      "learning_rate": 1.614486376713488e-05,
+      "loss": 0.0152,
+      "step": 4960
+    },
+    {
+      "epoch": 1.6807575245180928,
+      "grad_norm": 0.07033158838748932,
+      "learning_rate": 1.5975630394313758e-05,
+      "loss": 0.0162,
+      "step": 4970
+    },
+    {
+      "epoch": 1.6841393304024348,
+      "grad_norm": 0.10326120257377625,
+      "learning_rate": 1.5806397021492638e-05,
+      "loss": 0.0157,
+      "step": 4980
+    },
+    {
+      "epoch": 1.6875211362867772,
+      "grad_norm": 0.12648841738700867,
+      "learning_rate": 1.5637163648671518e-05,
+      "loss": 0.0165,
+      "step": 4990
+    },
+    {
+      "epoch": 1.6909029421711192,
+      "grad_norm": 0.07291960716247559,
+      "learning_rate": 1.5467930275850398e-05,
+      "loss": 0.0163,
+      "step": 5000
+    },
+    {
+      "epoch": 1.6942847480554617,
+      "grad_norm": 0.062065865844488144,
+      "learning_rate": 1.5298696903029278e-05,
+      "loss": 0.0156,
+      "step": 5010
+    },
+    {
+      "epoch": 1.6976665539398037,
+      "grad_norm": 0.08348127454519272,
+      "learning_rate": 1.5129463530208158e-05,
+      "loss": 0.0161,
+      "step": 5020
+    },
+    {
+      "epoch": 1.7010483598241462,
+      "grad_norm": 0.05632692202925682,
+      "learning_rate": 1.4960230157387036e-05,
+      "loss": 0.0162,
+      "step": 5030
+    },
+    {
+      "epoch": 1.7044301657084884,
+      "grad_norm": 0.08220499753952026,
+      "learning_rate": 1.4790996784565916e-05,
+      "loss": 0.0156,
+      "step": 5040
+    },
+    {
+      "epoch": 1.7078119715928306,
+      "grad_norm": 0.09087113291025162,
+      "learning_rate": 1.4621763411744796e-05,
+      "loss": 0.0163,
+      "step": 5050
+    },
+    {
+      "epoch": 1.7111937774771728,
+      "grad_norm": 0.07073336839675903,
+      "learning_rate": 1.4452530038923676e-05,
+      "loss": 0.0157,
+      "step": 5060
+    },
+    {
+      "epoch": 1.714575583361515,
+      "grad_norm": 0.0676584392786026,
+      "learning_rate": 1.4283296666102556e-05,
+      "loss": 0.0156,
+      "step": 5070
+    },
+    {
+      "epoch": 1.7179573892458573,
+      "grad_norm": 0.08832427114248276,
+      "learning_rate": 1.4114063293281434e-05,
+      "loss": 0.016,
+      "step": 5080
+    },
+    {
+      "epoch": 1.7213391951301995,
+      "grad_norm": 0.14378510415554047,
+      "learning_rate": 1.3944829920460314e-05,
+      "loss": 0.0193,
+      "step": 5090
+    },
+    {
+      "epoch": 1.7247210010145417,
+      "grad_norm": 0.0655590146780014,
+      "learning_rate": 1.3775596547639194e-05,
+      "loss": 0.0158,
+      "step": 5100
+    },
+    {
+      "epoch": 1.728102806898884,
+      "grad_norm": 0.0975392535328865,
+      "learning_rate": 1.3606363174818074e-05,
+      "loss": 0.0163,
+      "step": 5110
+    },
+    {
+      "epoch": 1.7314846127832264,
+      "grad_norm": 0.060749635100364685,
+      "learning_rate": 1.3437129801996954e-05,
+      "loss": 0.0154,
+      "step": 5120
+    },
+    {
+      "epoch": 1.7348664186675684,
+      "grad_norm": 0.07566985487937927,
+      "learning_rate": 1.3267896429175832e-05,
+      "loss": 0.0165,
+      "step": 5130
+    },
+    {
+      "epoch": 1.7382482245519109,
+      "grad_norm": 0.09878894686698914,
+      "learning_rate": 1.3098663056354712e-05,
+      "loss": 0.0159,
+      "step": 5140
+    },
+    {
+      "epoch": 1.7416300304362529,
+      "grad_norm": 0.07339813560247421,
+      "learning_rate": 1.2929429683533592e-05,
+      "loss": 0.0166,
+      "step": 5150
+    },
+    {
+      "epoch": 1.7450118363205953,
+      "grad_norm": 0.11021166294813156,
+      "learning_rate": 1.2760196310712474e-05,
+      "loss": 0.0153,
+      "step": 5160
+    },
+    {
+      "epoch": 1.7483936422049373,
+      "grad_norm": 0.15555544197559357,
+      "learning_rate": 1.2590962937891354e-05,
+      "loss": 0.0156,
+      "step": 5170
+    },
+    {
+      "epoch": 1.7517754480892798,
+      "grad_norm": 0.07689628005027771,
+      "learning_rate": 1.2421729565070232e-05,
+      "loss": 0.0156,
+      "step": 5180
+    },
+    {
+      "epoch": 1.7551572539736218,
+      "grad_norm": 0.06506321579217911,
+      "learning_rate": 1.2252496192249112e-05,
+      "loss": 0.0151,
+      "step": 5190
+    },
+    {
+      "epoch": 1.7585390598579642,
+      "grad_norm": 0.07474467903375626,
+      "learning_rate": 1.2083262819427992e-05,
+      "loss": 0.0152,
+      "step": 5200
+    },
+    {
+      "epoch": 1.7619208657423064,
+      "grad_norm": 0.0740489810705185,
+      "learning_rate": 1.191402944660687e-05,
+      "loss": 0.0156,
+      "step": 5210
+    },
+    {
+      "epoch": 1.7653026716266487,
+      "grad_norm": 0.05762564763426781,
+      "learning_rate": 1.174479607378575e-05,
+      "loss": 0.0148,
+      "step": 5220
+    },
+    {
+      "epoch": 1.768684477510991,
+      "grad_norm": 0.06422838568687439,
+      "learning_rate": 1.157556270096463e-05,
+      "loss": 0.016,
+      "step": 5230
+    },
+    {
+      "epoch": 1.7720662833953331,
+      "grad_norm": 0.06867840886116028,
+      "learning_rate": 1.140632932814351e-05,
+      "loss": 0.0148,
+      "step": 5240
+    },
+    {
+      "epoch": 1.7754480892796753,
+      "grad_norm": 0.09431049972772598,
+      "learning_rate": 1.123709595532239e-05,
+      "loss": 0.0162,
+      "step": 5250
+    },
+    {
+      "epoch": 1.7788298951640176,
+      "grad_norm": 0.13230328261852264,
+      "learning_rate": 1.1067862582501269e-05,
+      "loss": 0.016,
+      "step": 5260
+    },
+    {
+      "epoch": 1.7822117010483598,
+      "grad_norm": 0.08878026902675629,
+      "learning_rate": 1.089862920968015e-05,
+      "loss": 0.0156,
+      "step": 5270
+    },
+    {
+      "epoch": 1.785593506932702,
+      "grad_norm": 0.11089925467967987,
+      "learning_rate": 1.072939583685903e-05,
+      "loss": 0.0157,
+      "step": 5280
+    },
+    {
+      "epoch": 1.7889753128170442,
+      "grad_norm": 0.08735585957765579,
+      "learning_rate": 1.056016246403791e-05,
+      "loss": 0.0161,
+      "step": 5290
+    },
+    {
+      "epoch": 1.7923571187013865,
+      "grad_norm": 0.08069667965173721,
+      "learning_rate": 1.0390929091216788e-05,
+      "loss": 0.0154,
+      "step": 5300
+    },
+    {
+      "epoch": 1.795738924585729,
+      "grad_norm": 0.09315579384565353,
+      "learning_rate": 1.0221695718395668e-05,
+      "loss": 0.0152,
+      "step": 5310
+    },
+    {
+      "epoch": 1.799120730470071,
+      "grad_norm": 0.10615842044353485,
+      "learning_rate": 1.0052462345574548e-05,
+      "loss": 0.0158,
+      "step": 5320
+    },
+    {
+      "epoch": 1.8025025363544134,
+      "grad_norm": 0.08738986402750015,
+      "learning_rate": 9.883228972753428e-06,
+      "loss": 0.0159,
+      "step": 5330
+    },
+    {
+      "epoch": 1.8058843422387554,
+      "grad_norm": 0.08308973908424377,
+      "learning_rate": 9.713995599932307e-06,
+      "loss": 0.0154,
+      "step": 5340
+    },
+    {
+      "epoch": 1.8092661481230978,
+      "grad_norm": 0.056414760649204254,
+      "learning_rate": 9.544762227111187e-06,
+      "loss": 0.0159,
+      "step": 5350
+    },
+    {
+      "epoch": 1.8126479540074398,
+      "grad_norm": 0.11174483597278595,
+      "learning_rate": 9.375528854290067e-06,
+      "loss": 0.0157,
+      "step": 5360
+    },
+    {
+      "epoch": 1.8160297598917823,
+      "grad_norm": 0.12031450867652893,
+      "learning_rate": 9.206295481468947e-06,
+      "loss": 0.016,
+      "step": 5370
+    },
+    {
+      "epoch": 1.8194115657761243,
+      "grad_norm": 0.07165198028087616,
+      "learning_rate": 9.037062108647826e-06,
+      "loss": 0.0154,
+      "step": 5380
+    },
+    {
+      "epoch": 1.8227933716604667,
+      "grad_norm": 0.09684468805789948,
+      "learning_rate": 8.867828735826705e-06,
+      "loss": 0.0161,
+      "step": 5390
+    },
+    {
+      "epoch": 1.826175177544809,
+      "grad_norm": 0.06861387938261032,
+      "learning_rate": 8.698595363005585e-06,
+      "loss": 0.0157,
+      "step": 5400
+    },
+    {
+      "epoch": 1.8295569834291512,
+      "grad_norm": 0.06754298508167267,
+      "learning_rate": 8.529361990184465e-06,
+      "loss": 0.0156,
+      "step": 5410
+    },
+    {
+      "epoch": 1.8329387893134934,
+      "grad_norm": 0.07989432662725449,
+      "learning_rate": 8.360128617363345e-06,
+      "loss": 0.0166,
+      "step": 5420
+    },
+    {
+      "epoch": 1.8363205951978356,
+      "grad_norm": 0.07799744606018066,
+      "learning_rate": 8.190895244542223e-06,
+      "loss": 0.0153,
+      "step": 5430
+    },
+    {
+      "epoch": 1.8397024010821779,
+      "grad_norm": 0.07402709126472473,
+      "learning_rate": 8.021661871721103e-06,
+      "loss": 0.0158,
+      "step": 5440
+    },
+    {
+      "epoch": 1.84308420696652,
+      "grad_norm": 0.05629491060972214,
+      "learning_rate": 7.852428498899983e-06,
+      "loss": 0.0158,
+      "step": 5450
+    },
+    {
+      "epoch": 1.8464660128508623,
+      "grad_norm": 0.05271398648619652,
+      "learning_rate": 7.683195126078863e-06,
+      "loss": 0.0161,
+      "step": 5460
+    },
+    {
+      "epoch": 1.8498478187352045,
+      "grad_norm": 0.06697215884923935,
+      "learning_rate": 7.513961753257742e-06,
+      "loss": 0.0151,
+      "step": 5470
+    },
+    {
+      "epoch": 1.853229624619547,
+      "grad_norm": 0.06437050551176071,
+      "learning_rate": 7.344728380436622e-06,
+      "loss": 0.0162,
+      "step": 5480
+    },
+    {
+      "epoch": 1.856611430503889,
+      "grad_norm": 0.0728522315621376,
+      "learning_rate": 7.175495007615503e-06,
+      "loss": 0.0157,
+      "step": 5490
+    },
+    {
+      "epoch": 1.8599932363882314,
+      "grad_norm": 0.08476530015468597,
+      "learning_rate": 7.006261634794383e-06,
+      "loss": 0.0153,
+      "step": 5500
+    },
+    {
+      "epoch": 1.8633750422725734,
+      "grad_norm": 0.06566642969846725,
+      "learning_rate": 6.837028261973262e-06,
+      "loss": 0.0148,
+      "step": 5510
+    },
+    {
+      "epoch": 1.866756848156916,
+      "grad_norm": 0.09338746964931488,
+      "learning_rate": 6.667794889152142e-06,
+      "loss": 0.0157,
+      "step": 5520
+    },
+    {
+      "epoch": 1.870138654041258,
+      "grad_norm": 0.05955313891172409,
+      "learning_rate": 6.498561516331021e-06,
+      "loss": 0.0157,
+      "step": 5530
+    },
+    {
+      "epoch": 1.8735204599256003,
+      "grad_norm": 0.11766281723976135,
+      "learning_rate": 6.329328143509901e-06,
+      "loss": 0.0157,
+      "step": 5540
+    },
+    {
+      "epoch": 1.8769022658099423,
+      "grad_norm": 0.06120260804891586,
+      "learning_rate": 6.16009477068878e-06,
+      "loss": 0.0154,
+      "step": 5550
+    },
+    {
+      "epoch": 1.8802840716942848,
+      "grad_norm": 0.09879166632890701,
+      "learning_rate": 5.99086139786766e-06,
+      "loss": 0.015,
+      "step": 5560
+    },
+    {
+      "epoch": 1.883665877578627,
+      "grad_norm": 0.0957876592874527,
+      "learning_rate": 5.82162802504654e-06,
+      "loss": 0.0161,
+      "step": 5570
+    },
+    {
+      "epoch": 1.8870476834629692,
+      "grad_norm": 0.07821828126907349,
+      "learning_rate": 5.652394652225419e-06,
+      "loss": 0.0162,
+      "step": 5580
+    },
+    {
+      "epoch": 1.8904294893473115,
+      "grad_norm": 0.06772307306528091,
+      "learning_rate": 5.483161279404299e-06,
+      "loss": 0.0153,
+      "step": 5590
+    },
+    {
+      "epoch": 1.8938112952316537,
+      "grad_norm": 0.09948345273733139,
+      "learning_rate": 5.313927906583178e-06,
+      "loss": 0.0156,
+      "step": 5600
+    },
+    {
+      "epoch": 1.897193101115996,
+      "grad_norm": 0.07948364317417145,
+      "learning_rate": 5.144694533762058e-06,
+      "loss": 0.017,
+      "step": 5610
+    },
+    {
+      "epoch": 1.9005749070003382,
+      "grad_norm": 0.10279282182455063,
+      "learning_rate": 4.975461160940937e-06,
+      "loss": 0.0151,
+      "step": 5620
+    },
+    {
+      "epoch": 1.9039567128846804,
+      "grad_norm": 0.08826450258493423,
+      "learning_rate": 4.806227788119817e-06,
+      "loss": 0.0154,
+      "step": 5630
+    },
+    {
+      "epoch": 1.9073385187690226,
+      "grad_norm": 0.09827665984630585,
+      "learning_rate": 4.636994415298697e-06,
+      "loss": 0.0156,
+      "step": 5640
+    },
+    {
+      "epoch": 1.910720324653365,
+      "grad_norm": 0.06779050081968307,
+      "learning_rate": 4.467761042477576e-06,
+      "loss": 0.0154,
+      "step": 5650
+    },
+    {
+      "epoch": 1.914102130537707,
+      "grad_norm": 0.05314110219478607,
+      "learning_rate": 4.298527669656456e-06,
+      "loss": 0.0156,
+      "step": 5660
+    },
+    {
+      "epoch": 1.9174839364220495,
+      "grad_norm": 0.0623813197016716,
+      "learning_rate": 4.129294296835336e-06,
+      "loss": 0.0163,
+      "step": 5670
+    },
+    {
+      "epoch": 1.9208657423063915,
+      "grad_norm": 0.06000444293022156,
+      "learning_rate": 3.960060924014216e-06,
+      "loss": 0.0152,
+      "step": 5680
+    },
+    {
+      "epoch": 1.924247548190734,
+      "grad_norm": 0.059466857463121414,
+      "learning_rate": 3.7908275511930954e-06,
+      "loss": 0.0152,
+      "step": 5690
+    },
+    {
+      "epoch": 1.927629354075076,
+      "grad_norm": 0.04620284587144852,
+      "learning_rate": 3.6215941783719754e-06,
+      "loss": 0.0151,
+      "step": 5700
+    },
+    {
+      "epoch": 1.9310111599594184,
+      "grad_norm": 0.129261776804924,
+      "learning_rate": 3.452360805550855e-06,
+      "loss": 0.0152,
+      "step": 5710
+    },
+    {
+      "epoch": 1.9343929658437604,
+      "grad_norm": 0.07549330592155457,
+      "learning_rate": 3.2831274327297345e-06,
+      "loss": 0.0157,
+      "step": 5720
+    },
+    {
+      "epoch": 1.9377747717281029,
+      "grad_norm": 0.0915280282497406,
+      "learning_rate": 3.113894059908614e-06,
+      "loss": 0.0148,
+      "step": 5730
+    },
+    {
+      "epoch": 1.941156577612445,
+      "grad_norm": 0.09341620653867722,
+      "learning_rate": 2.9446606870874936e-06,
+      "loss": 0.0151,
+      "step": 5740
+    },
+    {
+      "epoch": 1.9445383834967873,
+      "grad_norm": 0.07050041854381561,
+      "learning_rate": 2.7754273142663736e-06,
+      "loss": 0.0153,
+      "step": 5750
+    },
+    {
+      "epoch": 1.9479201893811295,
+      "grad_norm": 0.05427863821387291,
+      "learning_rate": 2.606193941445253e-06,
+      "loss": 0.0149,
+      "step": 5760
+    },
+    {
+      "epoch": 1.9513019952654718,
+      "grad_norm": 0.08598488569259644,
+      "learning_rate": 2.4369605686241327e-06,
+      "loss": 0.0149,
+      "step": 5770
+    },
+    {
+      "epoch": 1.954683801149814,
+      "grad_norm": 0.08857986330986023,
+      "learning_rate": 2.267727195803012e-06,
+      "loss": 0.0147,
+      "step": 5780
+    },
+    {
+      "epoch": 1.9580656070341562,
+      "grad_norm": 0.09314049780368805,
+      "learning_rate": 2.098493822981892e-06,
+      "loss": 0.0158,
+      "step": 5790
+    },
+    {
+      "epoch": 1.9614474129184984,
+      "grad_norm": 0.057003553956747055,
+      "learning_rate": 1.929260450160772e-06,
+      "loss": 0.0155,
+      "step": 5800
+    },
+    {
+      "epoch": 1.9648292188028407,
+      "grad_norm": 0.0754491537809372,
+      "learning_rate": 1.7600270773396515e-06,
+      "loss": 0.0152,
+      "step": 5810
+    },
+    {
+      "epoch": 1.9682110246871831,
+      "grad_norm": 0.07122204452753067,
+      "learning_rate": 1.5907937045185312e-06,
+      "loss": 0.0152,
+      "step": 5820
+    },
+    {
+      "epoch": 1.9715928305715251,
+      "grad_norm": 0.06150359660387039,
+      "learning_rate": 1.4215603316974108e-06,
+      "loss": 0.0144,
+      "step": 5830
+    },
+    {
+      "epoch": 1.9749746364558676,
+      "grad_norm": 0.04940520599484444,
+      "learning_rate": 1.2523269588762905e-06,
+      "loss": 0.0149,
+      "step": 5840
+    },
+    {
+      "epoch": 1.9783564423402096,
+      "grad_norm": 0.06114586442708969,
+      "learning_rate": 1.08309358605517e-06,
+      "loss": 0.0157,
+      "step": 5850
+    },
+    {
+      "epoch": 1.981738248224552,
+      "grad_norm": 0.06917215138673782,
+      "learning_rate": 9.138602132340497e-07,
+      "loss": 0.0155,
+      "step": 5860
+    },
+    {
+      "epoch": 1.985120054108894,
+      "grad_norm": 0.04913664981722832,
+      "learning_rate": 7.446268404129294e-07,
+      "loss": 0.0155,
+      "step": 5870
+    },
+    {
+      "epoch": 1.9885018599932365,
+      "grad_norm": 0.07977385073900223,
+      "learning_rate": 5.753934675918092e-07,
+      "loss": 0.0144,
+      "step": 5880
+    },
+    {
+      "epoch": 1.9918836658775785,
+      "grad_norm": 0.07336314022541046,
+      "learning_rate": 4.061600947706888e-07,
+      "loss": 0.015,
+      "step": 5890
+    },
+    {
+      "epoch": 1.995265471761921,
+      "grad_norm": 0.05290009453892708,
+      "learning_rate": 2.3692672194956846e-07,
+      "loss": 0.0154,
+      "step": 5900
+    },
+    {
+      "epoch": 1.9986472776462632,
+      "grad_norm": 0.06475949287414551,
+      "learning_rate": 6.769334912844813e-08,
+      "loss": 0.0151,
+      "step": 5910
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.016348782926797867,
+      "eval_runtime": 342.3753,
+      "eval_samples_per_second": 11.636,
+      "eval_steps_per_second": 0.365,
+      "step": 5914
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5914,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "total_flos": 8.364323669652013e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}