diff --git "a/checkpoint-5000/trainer_state.json" "b/checkpoint-5000/trainer_state.json"
deleted file mode 100644--- "a/checkpoint-5000/trainer_state.json"
+++ /dev/null
@@ -1,35483 +0,0 @@
-{
-  "best_metric": 0.1314370483160019,
-  "best_model_checkpoint": "/scratch/csg337/pmf-grn-3dc/train_prior_network/hp_sweep/finetune_nt_lr0.00001_ga32_cw0.4_1.0_dr0.6/checkpoint-5000",
-  "epoch": 0.852492207688414,
-  "eval_steps": 100,
-  "global_step": 5000,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.00017049844153768282,
-      "grad_norm": 8.869589805603027,
-      "learning_rate": 9.998294970161979e-06,
-      "loss": 0.7814,
-      "step": 1
-    },
-    {
-      "epoch": 0.00034099688307536564,
-      "grad_norm": 7.021997451782227,
-      "learning_rate": 9.996589940323957e-06,
-      "loss": 0.6035,
-      "step": 2
-    },
-    {
-      "epoch": 0.0005114953246130485,
-      "grad_norm": 5.577943801879883,
-      "learning_rate": 9.994884910485935e-06,
-      "loss": 0.4712,
-      "step": 3
-    },
-    {
-      "epoch": 0.0006819937661507313,
-      "grad_norm": 4.31350564956665,
-      "learning_rate": 9.993179880647913e-06,
-      "loss": 0.3819,
-      "step": 4
-    },
-    {
-      "epoch": 0.0008524922076884142,
-      "grad_norm": 3.362149477005005,
-      "learning_rate": 9.99147485080989e-06,
-      "loss": 0.3102,
-      "step": 5
-    },
-    {
-      "epoch": 0.001022990649226097,
-      "grad_norm": 1.3885729312896729,
-      "learning_rate": 9.989769820971867e-06,
-      "loss": 0.3531,
-      "step": 6
-    },
-    {
-      "epoch": 0.0011934890907637798,
-      "grad_norm": 1.580288052558899,
-      "learning_rate": 9.988064791133846e-06,
-      "loss": 0.2337,
-      "step": 7
-    },
-    {
-      "epoch": 0.0013639875323014626,
-      "grad_norm": 0.9532984495162964,
-      "learning_rate": 9.986359761295823e-06,
-      "loss": 0.2231,
-      "step": 8
-    },
-    {
-      "epoch": 0.0015344859738391453,
-      "grad_norm": 0.7379500865936279,
-      "learning_rate": 9.9846547314578e-06,
-      "loss": 0.1995,
-      "step": 9
-    },
-    {
-      "epoch": 0.0017049844153768283,
-      "grad_norm": 0.45577266812324524,
-      "learning_rate": 9.98294970161978e-06,
-      "loss": 0.2418,
-      "step": 10
-    },
-    {
-      "epoch": 0.001875482856914511,
-      "grad_norm": 0.64174884557724,
-      "learning_rate": 9.981244671781757e-06,
-      "loss": 0.122,
-      "step": 11
-    },
-    {
-      "epoch": 0.002045981298452194,
-      "grad_norm": 1.0696319341659546,
-      "learning_rate": 9.979539641943734e-06,
-      "loss": 0.3283,
-      "step": 12
-    },
-    {
-      "epoch": 0.0022164797399898766,
-      "grad_norm": 0.6181367635726929,
-      "learning_rate": 9.977834612105712e-06,
-      "loss": 0.1975,
-      "step": 13
-    },
-    {
-      "epoch": 0.0023869781815275596,
-      "grad_norm": 0.6715260148048401,
-      "learning_rate": 9.97612958226769e-06,
-      "loss": 0.096,
-      "step": 14
-    },
-    {
-      "epoch": 0.002557476623065242,
-      "grad_norm": 0.6611419320106506,
-      "learning_rate": 9.974424552429668e-06,
-      "loss": 0.1216,
-      "step": 15
-    },
-    {
-      "epoch": 0.002727975064602925,
-      "grad_norm": 0.5713757276535034,
-      "learning_rate": 9.972719522591646e-06,
-      "loss": 0.199,
-      "step": 16
-    },
-    {
-      "epoch": 0.002898473506140608,
-      "grad_norm": 0.9805899262428284,
-      "learning_rate": 9.971014492753624e-06,
-      "loss": 0.2685,
-      "step": 17
-    },
-    {
-      "epoch": 0.0030689719476782906,
-      "grad_norm": 0.5129013061523438,
-      "learning_rate": 9.969309462915602e-06,
-      "loss": 0.2221,
-      "step": 18
-    },
-    {
-      "epoch": 0.0032394703892159736,
-      "grad_norm": 0.503084659576416,
-      "learning_rate": 9.96760443307758e-06,
-      "loss": 0.1479,
-      "step": 19
-    },
-    {
-      "epoch": 0.0034099688307536566,
-      "grad_norm": 0.70774906873703,
-      "learning_rate": 9.965899403239558e-06,
-      "loss": 0.1203,
-      "step": 20
-    },
-    {
-      "epoch": 0.003580467272291339,
-      "grad_norm": 0.5927731394767761,
-      "learning_rate": 9.964194373401536e-06,
-      "loss": 0.1222,
-      "step": 21
-    },
-    {
-      "epoch": 0.003750965713829022,
-      "grad_norm": 0.5986411571502686,
-      "learning_rate": 9.962489343563512e-06,
-      "loss": 0.1092,
-      "step": 22
-    },
-    {
-      "epoch": 0.003921464155366705,
-      "grad_norm": 0.3205660879611969,
-      "learning_rate": 9.960784313725492e-06,
-      "loss": 0.1677,
-      "step": 23
-    },
-    {
-      "epoch": 0.004091962596904388,
-      "grad_norm": 0.4281774163246155,
-      "learning_rate": 9.959079283887468e-06,
-      "loss": 0.1171,
-      "step": 24
-    },
-    {
-      "epoch": 0.00426246103844207,
-      "grad_norm": 0.5072917938232422,
-      "learning_rate": 9.957374254049446e-06,
-      "loss": 0.195,
-      "step": 25
-    },
-    {
-      "epoch": 0.004432959479979753,
-      "grad_norm": 0.45099368691444397,
-      "learning_rate": 9.955669224211426e-06,
-      "loss": 0.1907,
-      "step": 26
-    },
-    {
-      "epoch": 0.004603457921517436,
-      "grad_norm": 0.2903910279273987,
-      "learning_rate": 9.953964194373402e-06,
-      "loss": 0.1712,
-      "step": 27
-    },
-    {
-      "epoch": 0.004773956363055119,
-      "grad_norm": 0.39423874020576477,
-      "learning_rate": 9.95225916453538e-06,
-      "loss": 0.0873,
-      "step": 28
-    },
-    {
-      "epoch": 0.004944454804592802,
-      "grad_norm": 1.3082679510116577,
-      "learning_rate": 9.950554134697358e-06,
-      "loss": 0.3284,
-      "step": 29
-    },
-    {
-      "epoch": 0.005114953246130484,
-      "grad_norm": 0.22596949338912964,
-      "learning_rate": 9.948849104859336e-06,
-      "loss": 0.136,
-      "step": 30
-    },
-    {
-      "epoch": 0.005285451687668167,
-      "grad_norm": 0.32791393995285034,
-      "learning_rate": 9.947144075021314e-06,
-      "loss": 0.1208,
-      "step": 31
-    },
-    {
-      "epoch": 0.00545595012920585,
-      "grad_norm": 0.9837888479232788,
-      "learning_rate": 9.945439045183292e-06,
-      "loss": 0.2933,
-      "step": 32
-    },
-    {
-      "epoch": 0.005626448570743533,
-      "grad_norm": 0.26505422592163086,
-      "learning_rate": 9.943734015345268e-06,
-      "loss": 0.1744,
-      "step": 33
-    },
-    {
-      "epoch": 0.005796947012281216,
-      "grad_norm": 0.3285515308380127,
-      "learning_rate": 9.942028985507248e-06,
-      "loss": 0.1913,
-      "step": 34
-    },
-    {
-      "epoch": 0.005967445453818898,
-      "grad_norm": 0.4692244231700897,
-      "learning_rate": 9.940323955669226e-06,
-      "loss": 0.1721,
-      "step": 35
-    },
-    {
-      "epoch": 0.006137943895356581,
-      "grad_norm": 0.7220498323440552,
-      "learning_rate": 9.938618925831202e-06,
-      "loss": 0.1271,
-      "step": 36
-    },
-    {
-      "epoch": 0.006308442336894264,
-      "grad_norm": 0.5188308358192444,
-      "learning_rate": 9.936913895993181e-06,
-      "loss": 0.1699,
-      "step": 37
-    },
-    {
-      "epoch": 0.006478940778431947,
-      "grad_norm": 0.29906973242759705,
-      "learning_rate": 9.935208866155158e-06,
-      "loss": 0.1882,
-      "step": 38
-    },
-    {
-      "epoch": 0.00664943921996963,
-      "grad_norm": 0.858116626739502,
-      "learning_rate": 9.933503836317137e-06,
-      "loss": 0.2837,
-      "step": 39
-    },
-    {
-      "epoch": 0.006819937661507313,
-      "grad_norm": 0.38193100690841675,
-      "learning_rate": 9.931798806479114e-06,
-      "loss": 0.221,
-      "step": 40
-    },
-    {
-      "epoch": 0.006990436103044995,
-      "grad_norm": 0.3163479268550873,
-      "learning_rate": 9.930093776641092e-06,
-      "loss": 0.1911,
-      "step": 41
-    },
-    {
-      "epoch": 0.007160934544582678,
-      "grad_norm": 0.6785820722579956,
-      "learning_rate": 9.928388746803071e-06,
-      "loss": 0.1457,
-      "step": 42
-    },
-    {
-      "epoch": 0.007331432986120361,
-      "grad_norm": 0.4343841075897217,
-      "learning_rate": 9.926683716965047e-06,
-      "loss": 0.2055,
-      "step": 43
-    },
-    {
-      "epoch": 0.007501931427658044,
-      "grad_norm": 0.5253484845161438,
-      "learning_rate": 9.924978687127025e-06,
-      "loss": 0.2558,
-      "step": 44
-    },
-    {
-      "epoch": 0.007672429869195727,
-      "grad_norm": 0.4470311105251312,
-      "learning_rate": 9.923273657289003e-06,
-      "loss": 0.1753,
-      "step": 45
-    },
-    {
-      "epoch": 0.00784292831073341,
-      "grad_norm": 0.7119751572608948,
-      "learning_rate": 9.921568627450981e-06,
-      "loss": 0.2984,
-      "step": 46
-    },
-    {
-      "epoch": 0.008013426752271093,
-      "grad_norm": 0.634812593460083,
-      "learning_rate": 9.91986359761296e-06,
-      "loss": 0.1811,
-      "step": 47
-    },
-    {
-      "epoch": 0.008183925193808776,
-      "grad_norm": 0.6000546216964722,
-      "learning_rate": 9.918158567774937e-06,
-      "loss": 0.2744,
-      "step": 48
-    },
-    {
-      "epoch": 0.008354423635346457,
-      "grad_norm": 0.7269101142883301,
-      "learning_rate": 9.916453537936913e-06,
-      "loss": 0.1516,
-      "step": 49
-    },
-    {
-      "epoch": 0.00852492207688414,
-      "grad_norm": 0.5989276766777039,
-      "learning_rate": 9.914748508098893e-06,
-      "loss": 0.189,
-      "step": 50
-    },
-    {
-      "epoch": 0.008695420518421823,
-      "grad_norm": 0.5950725078582764,
-      "learning_rate": 9.913043478260871e-06,
-      "loss": 0.1791,
-      "step": 51
-    },
-    {
-      "epoch": 0.008865918959959506,
-      "grad_norm": 0.4925420582294464,
-      "learning_rate": 9.911338448422847e-06,
-      "loss": 0.1503,
-      "step": 52
-    },
-    {
-      "epoch": 0.00903641740149719,
-      "grad_norm": 0.8844504356384277,
-      "learning_rate": 9.909633418584827e-06,
-      "loss": 0.2852,
-      "step": 53
-    },
-    {
-      "epoch": 0.009206915843034872,
-      "grad_norm": 0.32095837593078613,
-      "learning_rate": 9.907928388746803e-06,
-      "loss": 0.167,
-      "step": 54
-    },
-    {
-      "epoch": 0.009377414284572555,
-      "grad_norm": 0.42962777614593506,
-      "learning_rate": 9.906223358908781e-06,
-      "loss": 0.1481,
-      "step": 55
-    },
-    {
-      "epoch": 0.009547912726110238,
-      "grad_norm": 1.1052173376083374,
-      "learning_rate": 9.904518329070759e-06,
-      "loss": 0.306,
-      "step": 56
-    },
-    {
-      "epoch": 0.009718411167647921,
-      "grad_norm": 0.5791013836860657,
-      "learning_rate": 9.902813299232737e-06,
-      "loss": 0.2187,
-      "step": 57
-    },
-    {
-      "epoch": 0.009888909609185604,
-      "grad_norm": 0.5379682183265686,
-      "learning_rate": 9.901108269394715e-06,
-      "loss": 0.1645,
-      "step": 58
-    },
-    {
-      "epoch": 0.010059408050723286,
-      "grad_norm": 0.6275455355644226,
-      "learning_rate": 9.899403239556693e-06,
-      "loss": 0.1489,
-      "step": 59
-    },
-    {
-      "epoch": 0.010229906492260969,
-      "grad_norm": 0.6161112785339355,
-      "learning_rate": 9.897698209718671e-06,
-      "loss": 0.1487,
-      "step": 60
-    },
-    {
-      "epoch": 0.010400404933798651,
-      "grad_norm": 0.5155672430992126,
-      "learning_rate": 9.895993179880649e-06,
-      "loss": 0.1816,
-      "step": 61
-    },
-    {
-      "epoch": 0.010570903375336334,
-      "grad_norm": 1.4397295713424683,
-      "learning_rate": 9.894288150042627e-06,
-      "loss": 0.3635,
-      "step": 62
-    },
-    {
-      "epoch": 0.010741401816874017,
-      "grad_norm": 0.4438450038433075,
-      "learning_rate": 9.892583120204605e-06,
-      "loss": 0.1475,
-      "step": 63
-    },
-    {
-      "epoch": 0.0109119002584117,
-      "grad_norm": 0.8751521110534668,
-      "learning_rate": 9.890878090366583e-06,
-      "loss": 0.2665,
-      "step": 64
-    },
-    {
-      "epoch": 0.011082398699949383,
-      "grad_norm": 0.4153095483779907,
-      "learning_rate": 9.889173060528559e-06,
-      "loss": 0.1866,
-      "step": 65
-    },
-    {
-      "epoch": 0.011252897141487066,
-      "grad_norm": 0.35557761788368225,
-      "learning_rate": 9.887468030690539e-06,
-      "loss": 0.1678,
-      "step": 66
-    },
-    {
-      "epoch": 0.01142339558302475,
-      "grad_norm": 0.7395035624504089,
-      "learning_rate": 9.885763000852517e-06,
-      "loss": 0.1002,
-      "step": 67
-    },
-    {
-      "epoch": 0.011593894024562432,
-      "grad_norm": 0.3700765073299408,
-      "learning_rate": 9.884057971014493e-06,
-      "loss": 0.2083,
-      "step": 68
-    },
-    {
-      "epoch": 0.011764392466100115,
-      "grad_norm": 0.8187807202339172,
-      "learning_rate": 9.882352941176472e-06,
-      "loss": 0.072,
-      "step": 69
-    },
-    {
-      "epoch": 0.011934890907637797,
-      "grad_norm": 0.35573920607566833,
-      "learning_rate": 9.880647911338449e-06,
-      "loss": 0.2068,
-      "step": 70
-    },
-    {
-      "epoch": 0.01210538934917548,
-      "grad_norm": 0.4666770100593567,
-      "learning_rate": 9.878942881500427e-06,
-      "loss": 0.2112,
-      "step": 71
-    },
-    {
-      "epoch": 0.012275887790713163,
-      "grad_norm": 0.23817773163318634,
-      "learning_rate": 9.877237851662405e-06,
-      "loss": 0.1685,
-      "step": 72
-    },
-    {
-      "epoch": 0.012446386232250846,
-      "grad_norm": 0.4650826156139374,
-      "learning_rate": 9.875532821824383e-06,
-      "loss": 0.2222,
-      "step": 73
-    },
-    {
-      "epoch": 0.012616884673788529,
-      "grad_norm": 0.2902311384677887,
-      "learning_rate": 9.87382779198636e-06,
-      "loss": 0.1436,
-      "step": 74
-    },
-    {
-      "epoch": 0.012787383115326212,
-      "grad_norm": 0.38183656334877014,
-      "learning_rate": 9.872122762148338e-06,
-      "loss": 0.1163,
-      "step": 75
-    },
-    {
-      "epoch": 0.012957881556863894,
-      "grad_norm": 0.3778226375579834,
-      "learning_rate": 9.870417732310316e-06,
-      "loss": 0.1371,
-      "step": 76
-    },
-    {
-      "epoch": 0.013128379998401577,
-      "grad_norm": 1.2414989471435547,
-      "learning_rate": 9.868712702472294e-06,
-      "loss": 0.3363,
-      "step": 77
-    },
-    {
-      "epoch": 0.01329887843993926,
-      "grad_norm": 0.1740960031747818,
-      "learning_rate": 9.867007672634272e-06,
-      "loss": 0.1663,
-      "step": 78
-    },
-    {
-      "epoch": 0.013469376881476943,
-      "grad_norm": 0.8906394839286804,
-      "learning_rate": 9.86530264279625e-06,
-      "loss": 0.2848,
-      "step": 79
-    },
-    {
-      "epoch": 0.013639875323014626,
-      "grad_norm": 0.41351833939552307,
-      "learning_rate": 9.863597612958228e-06,
-      "loss": 0.2146,
-      "step": 80
-    },
-    {
-      "epoch": 0.013810373764552308,
-      "grad_norm": 0.4237087070941925,
-      "learning_rate": 9.861892583120204e-06,
-      "loss": 0.2408,
-      "step": 81
-    },
-    {
-      "epoch": 0.01398087220608999,
-      "grad_norm": 0.2752889394760132,
-      "learning_rate": 9.860187553282184e-06,
-      "loss": 0.2159,
-      "step": 82
-    },
-    {
-      "epoch": 0.014151370647627674,
-      "grad_norm": 0.3237917423248291,
-      "learning_rate": 9.85848252344416e-06,
-      "loss": 0.2191,
-      "step": 83
-    },
-    {
-      "epoch": 0.014321869089165357,
-      "grad_norm": 0.41194331645965576,
-      "learning_rate": 9.856777493606138e-06,
-      "loss": 0.2122,
-      "step": 84
-    },
-    {
-      "epoch": 0.01449236753070304,
-      "grad_norm": 0.7924314141273499,
-      "learning_rate": 9.855072463768118e-06,
-      "loss": 0.1722,
-      "step": 85
-    },
-    {
-      "epoch": 0.014662865972240723,
-      "grad_norm": 1.0692120790481567,
-      "learning_rate": 9.853367433930094e-06,
-      "loss": 0.137,
-      "step": 86
-    },
-    {
-      "epoch": 0.014833364413778406,
-      "grad_norm": 0.615675687789917,
-      "learning_rate": 9.851662404092072e-06,
-      "loss": 0.1637,
-      "step": 87
-    },
-    {
-      "epoch": 0.015003862855316089,
-      "grad_norm": 0.23946630954742432,
-      "learning_rate": 9.84995737425405e-06,
-      "loss": 0.2113,
-      "step": 88
-    },
-    {
-      "epoch": 0.015174361296853772,
-      "grad_norm": 0.40212199091911316,
-      "learning_rate": 9.848252344416028e-06,
-      "loss": 0.2339,
-      "step": 89
-    },
-    {
-      "epoch": 0.015344859738391455,
-      "grad_norm": 0.3420916497707367,
-      "learning_rate": 9.846547314578006e-06,
-      "loss": 0.145,
-      "step": 90
-    },
-    {
-      "epoch": 0.015515358179929136,
-      "grad_norm": 0.19429825246334076,
-      "learning_rate": 9.844842284739984e-06,
-      "loss": 0.1709,
-      "step": 91
-    },
-    {
-      "epoch": 0.01568585662146682,
-      "grad_norm": 0.3276514708995819,
-      "learning_rate": 9.843137254901962e-06,
-      "loss": 0.1178,
-      "step": 92
-    },
-    {
-      "epoch": 0.015856355063004503,
-      "grad_norm": 0.7621896862983704,
-      "learning_rate": 9.84143222506394e-06,
-      "loss": 0.2462,
-      "step": 93
-    },
-    {
-      "epoch": 0.016026853504542186,
-      "grad_norm": 0.18447129428386688,
-      "learning_rate": 9.839727195225918e-06,
-      "loss": 0.1408,
-      "step": 94
-    },
-    {
-      "epoch": 0.01619735194607987,
-      "grad_norm": 0.42043575644493103,
-      "learning_rate": 9.838022165387896e-06,
-      "loss": 0.1872,
-      "step": 95
-    },
-    {
-      "epoch": 0.016367850387617552,
-      "grad_norm": 0.23067864775657654,
-      "learning_rate": 9.836317135549874e-06,
-      "loss": 0.1475,
-      "step": 96
-    },
-    {
-      "epoch": 0.016538348829155232,
-      "grad_norm": 0.22350552678108215,
-      "learning_rate": 9.83461210571185e-06,
-      "loss": 0.1387,
-      "step": 97
-    },
-    {
-      "epoch": 0.016708847270692915,
-      "grad_norm": 0.18144337832927704,
-      "learning_rate": 9.83290707587383e-06,
-      "loss": 0.1316,
-      "step": 98
-    },
-    {
-      "epoch": 0.016879345712230598,
-      "grad_norm": 0.4881460666656494,
-      "learning_rate": 9.831202046035806e-06,
-      "loss": 0.1929,
-      "step": 99
-    },
-    {
-      "epoch": 0.01704984415376828,
-      "grad_norm": 0.6317399740219116,
-      "learning_rate": 9.829497016197784e-06,
-      "loss": 0.2182,
-      "step": 100
-    },
-    {
-      "epoch": 0.01704984415376828,
-      "eval_f1_score": 0.0,
-      "eval_loss": 0.22505781054496765,
-      "eval_runtime": 183.3482,
-      "eval_samples_per_second": 54.541,
-      "eval_steps_per_second": 3.409,
-      "step": 100
-    },
-    {
-      "epoch": 0.017220342595305964,
-      "grad_norm": 0.2174391895532608,
-      "learning_rate": 9.827791986359763e-06,
-      "loss": 0.1484,
-      "step": 101
-    },
-    {
-      "epoch": 0.017390841036843647,
-      "grad_norm": 0.21559378504753113,
-      "learning_rate": 9.82608695652174e-06,
-      "loss": 0.1462,
-      "step": 102
-    },
-    {
-      "epoch": 0.01756133947838133,
-      "grad_norm": 0.38197803497314453,
-      "learning_rate": 9.824381926683718e-06,
-      "loss": 0.1137,
-      "step": 103
-    },
-    {
-      "epoch": 0.017731837919919013,
-      "grad_norm": 0.609444260597229,
-      "learning_rate": 9.822676896845696e-06,
-      "loss": 0.2368,
-      "step": 104
-    },
-    {
-      "epoch": 0.017902336361456696,
-      "grad_norm": 0.2939724922180176,
-      "learning_rate": 9.820971867007673e-06,
-      "loss": 0.1617,
-      "step": 105
-    },
-    {
-      "epoch": 0.01807283480299438,
-      "grad_norm": 0.5116530656814575,
-      "learning_rate": 9.819266837169651e-06,
-      "loss": 0.1289,
-      "step": 106
-    },
-    {
-      "epoch": 0.01824333324453206,
-      "grad_norm": 0.6672798991203308,
-      "learning_rate": 9.81756180733163e-06,
-      "loss": 0.2506,
-      "step": 107
-    },
-    {
-      "epoch": 0.018413831686069745,
-      "grad_norm": 0.6284186840057373,
-      "learning_rate": 9.815856777493606e-06,
-      "loss": 0.0931,
-      "step": 108
-    },
-    {
-      "epoch": 0.018584330127607428,
-      "grad_norm": 0.38140490651130676,
-      "learning_rate": 9.814151747655585e-06,
-      "loss": 0.1199,
-      "step": 109
-    },
-    {
-      "epoch": 0.01875482856914511,
-      "grad_norm": 0.2449476420879364,
-      "learning_rate": 9.812446717817563e-06,
-      "loss": 0.1697,
-      "step": 110
-    },
-    {
-      "epoch": 0.018925327010682794,
-      "grad_norm": 0.4863952100276947,
-      "learning_rate": 9.81074168797954e-06,
-      "loss": 0.2203,
-      "step": 111
-    },
-    {
-      "epoch": 0.019095825452220477,
-      "grad_norm": 1.2018239498138428,
-      "learning_rate": 9.809036658141519e-06,
-      "loss": 0.3048,
-      "step": 112
-    },
-    {
-      "epoch": 0.01926632389375816,
-      "grad_norm": 0.3844000995159149,
-      "learning_rate": 9.807331628303495e-06,
-      "loss": 0.1784,
-      "step": 113
-    },
-    {
-      "epoch": 0.019436822335295843,
-      "grad_norm": 0.2755303680896759,
-      "learning_rate": 9.805626598465475e-06,
-      "loss": 0.1875,
-      "step": 114
-    },
-    {
-      "epoch": 0.019607320776833526,
-      "grad_norm": 0.3493352234363556,
-      "learning_rate": 9.803921568627451e-06,
-      "loss": 0.1688,
-      "step": 115
-    },
-    {
-      "epoch": 0.01977781921837121,
-      "grad_norm": 0.6989247798919678,
-      "learning_rate": 9.80221653878943e-06,
-      "loss": 0.1311,
-      "step": 116
-    },
-    {
-      "epoch": 0.01994831765990889,
-      "grad_norm": 0.3620999753475189,
-      "learning_rate": 9.800511508951407e-06,
-      "loss": 0.1595,
-      "step": 117
-    },
-    {
-      "epoch": 0.02011881610144657,
-      "grad_norm": 0.36481228470802307,
-      "learning_rate": 9.798806479113385e-06,
-      "loss": 0.1641,
-      "step": 118
-    },
-    {
-      "epoch": 0.020289314542984254,
-      "grad_norm": 0.7398555278778076,
-      "learning_rate": 9.797101449275363e-06,
-      "loss": 0.1037,
-      "step": 119
-    },
-    {
-      "epoch": 0.020459812984521937,
-      "grad_norm": 0.5512336492538452,
-      "learning_rate": 9.795396419437341e-06,
-      "loss": 0.2342,
-      "step": 120
-    },
-    {
-      "epoch": 0.02063031142605962,
-      "grad_norm": 0.25768429040908813,
-      "learning_rate": 9.793691389599319e-06,
-      "loss": 0.1686,
-      "step": 121
-    },
-    {
-      "epoch": 0.020800809867597303,
-      "grad_norm": 0.36216527223587036,
-      "learning_rate": 9.791986359761297e-06,
-      "loss": 0.1923,
-      "step": 122
-    },
-    {
-      "epoch": 0.020971308309134986,
-      "grad_norm": 0.26241618394851685,
-      "learning_rate": 9.790281329923275e-06,
-      "loss": 0.1424,
-      "step": 123
-    },
-    {
-      "epoch": 0.02114180675067267,
-      "grad_norm": 0.25239989161491394,
-      "learning_rate": 9.788576300085251e-06,
-      "loss": 0.1437,
-      "step": 124
-    },
-    {
-      "epoch": 0.021312305192210352,
-      "grad_norm": 0.3067416548728943,
-      "learning_rate": 9.78687127024723e-06,
-      "loss": 0.1787,
-      "step": 125
-    },
-    {
-      "epoch": 0.021482803633748035,
-      "grad_norm": 0.7511736750602722,
-      "learning_rate": 9.785166240409209e-06,
-      "loss": 0.2259,
-      "step": 126
-    },
-    {
-      "epoch": 0.021653302075285718,
-      "grad_norm": 0.5204250812530518,
-      "learning_rate": 9.783461210571185e-06,
-      "loss": 0.1876,
-      "step": 127
-    },
-    {
-      "epoch": 0.0218238005168234,
-      "grad_norm": 0.5185781121253967,
-      "learning_rate": 9.781756180733165e-06,
-      "loss": 0.0949,
-      "step": 128
-    },
-    {
-      "epoch": 0.021994298958361084,
-      "grad_norm": 0.5859097838401794,
-      "learning_rate": 9.78005115089514e-06,
-      "loss": 0.1002,
-      "step": 129
-    },
-    {
-      "epoch": 0.022164797399898767,
-      "grad_norm": 0.3053465187549591,
-      "learning_rate": 9.778346121057119e-06,
-      "loss": 0.168,
-      "step": 130
-    },
-    {
-      "epoch": 0.02233529584143645,
-      "grad_norm": 0.2791380286216736,
-      "learning_rate": 9.776641091219097e-06,
-      "loss": 0.1916,
-      "step": 131
-    },
-    {
-      "epoch": 0.022505794282974133,
-      "grad_norm": 0.5178912878036499,
-      "learning_rate": 9.774936061381075e-06,
-      "loss": 0.2206,
-      "step": 132
-    },
-    {
-      "epoch": 0.022676292724511816,
-      "grad_norm": 0.594906210899353,
-      "learning_rate": 9.773231031543053e-06,
-      "loss": 0.08,
-      "step": 133
-    },
-    {
-      "epoch": 0.0228467911660495,
-      "grad_norm": 0.789562463760376,
-      "learning_rate": 9.77152600170503e-06,
-      "loss": 0.2564,
-      "step": 134
-    },
-    {
-      "epoch": 0.023017289607587182,
-      "grad_norm": 0.8185597062110901,
-      "learning_rate": 9.769820971867009e-06,
-      "loss": 0.2694,
-      "step": 135
-    },
-    {
-      "epoch": 0.023187788049124865,
-      "grad_norm": 0.5645036101341248,
-      "learning_rate": 9.768115942028986e-06,
-      "loss": 0.2195,
-      "step": 136
-    },
-    {
-      "epoch": 0.023358286490662548,
-      "grad_norm": 0.3041114807128906,
-      "learning_rate": 9.766410912190964e-06,
-      "loss": 0.1951,
-      "step": 137
-    },
-    {
-      "epoch": 0.02352878493220023,
-      "grad_norm": 0.6470907926559448,
-      "learning_rate": 9.764705882352942e-06,
-      "loss": 0.2616,
-      "step": 138
-    },
-    {
-      "epoch": 0.02369928337373791,
-      "grad_norm": 0.5063560009002686,
-      "learning_rate": 9.76300085251492e-06,
-      "loss": 0.257,
-      "step": 139
-    },
-    {
-      "epoch": 0.023869781815275593,
-      "grad_norm": 0.5629501342773438,
-      "learning_rate": 9.761295822676897e-06,
-      "loss": 0.201,
-      "step": 140
-    },
-    {
-      "epoch": 0.024040280256813276,
-      "grad_norm": 1.2421547174453735,
-      "learning_rate": 9.759590792838876e-06,
-      "loss": 0.124,
-      "step": 141
-    },
-    {
-      "epoch": 0.02421077869835096,
-      "grad_norm": 0.9891913533210754,
-      "learning_rate": 9.757885763000852e-06,
-      "loss": 0.1581,
-      "step": 142
-    },
-    {
-      "epoch": 0.024381277139888642,
-      "grad_norm": 0.8758956789970398,
-      "learning_rate": 9.75618073316283e-06,
-      "loss": 0.1515,
-      "step": 143
-    },
-    {
-      "epoch": 0.024551775581426325,
-      "grad_norm": 0.6748665571212769,
-      "learning_rate": 9.75447570332481e-06,
-      "loss": 0.1611,
-      "step": 144
-    },
-    {
-      "epoch": 0.024722274022964008,
-      "grad_norm": 0.4170930087566376,
-      "learning_rate": 9.752770673486786e-06,
-      "loss": 0.2401,
-      "step": 145
-    },
-    {
-      "epoch": 0.02489277246450169,
-      "grad_norm": 0.30556097626686096,
-      "learning_rate": 9.751065643648764e-06,
-      "loss": 0.1555,
-      "step": 146
-    },
-    {
-      "epoch": 0.025063270906039374,
-      "grad_norm": 0.3912142813205719,
-      "learning_rate": 9.749360613810742e-06,
-      "loss": 0.2025,
-      "step": 147
-    },
-    {
-      "epoch": 0.025233769347577057,
-      "grad_norm": 0.5749161839485168,
-      "learning_rate": 9.74765558397272e-06,
-      "loss": 0.2275,
-      "step": 148
-    },
-    {
-      "epoch": 0.02540426778911474,
-      "grad_norm": 0.6735514402389526,
-      "learning_rate": 9.745950554134698e-06,
-      "loss": 0.2427,
-      "step": 149
-    },
-    {
-      "epoch": 0.025574766230652423,
-      "grad_norm": 0.33073481917381287,
-      "learning_rate": 9.744245524296676e-06,
-      "loss": 0.1338,
-      "step": 150
-    },
-    {
-      "epoch": 0.025745264672190106,
-      "grad_norm": 0.2953212261199951,
-      "learning_rate": 9.742540494458654e-06,
-      "loss": 0.1467,
-      "step": 151
-    },
-    {
-      "epoch": 0.02591576311372779,
-      "grad_norm": 0.5662497282028198,
-      "learning_rate": 9.740835464620632e-06,
-      "loss": 0.2264,
-      "step": 152
-    },
-    {
-      "epoch": 0.026086261555265472,
-      "grad_norm": 0.5139706134796143,
-      "learning_rate": 9.73913043478261e-06,
-      "loss": 0.2136,
-      "step": 153
-    },
-    {
-      "epoch": 0.026256759996803155,
-      "grad_norm": 0.35498949885368347,
-      "learning_rate": 9.737425404944588e-06,
-      "loss": 0.1618,
-      "step": 154
-    },
-    {
-      "epoch": 0.026427258438340838,
-      "grad_norm": 0.5710799694061279,
-      "learning_rate": 9.735720375106566e-06,
-      "loss": 0.2465,
-      "step": 155
-    },
-    {
-      "epoch": 0.02659775687987852,
-      "grad_norm": 0.4561382830142975,
-      "learning_rate": 9.734015345268542e-06,
-      "loss": 0.1765,
-      "step": 156
-    },
-    {
-      "epoch": 0.026768255321416204,
-      "grad_norm": 0.7372007369995117,
-      "learning_rate": 9.732310315430522e-06,
-      "loss": 0.0938,
-      "step": 157
-    },
-    {
-      "epoch": 0.026938753762953887,
-      "grad_norm": 0.45218053460121155,
-      "learning_rate": 9.730605285592498e-06,
-      "loss": 0.2387,
-      "step": 158
-    },
-    {
-      "epoch": 0.02710925220449157,
-      "grad_norm": 0.6522502899169922,
-      "learning_rate": 9.728900255754476e-06,
-      "loss": 0.1163,
-      "step": 159
-    },
-    {
-      "epoch": 0.027279750646029253,
-      "grad_norm": 0.4411844313144684,
-      "learning_rate": 9.727195225916456e-06,
-      "loss": 0.138,
-      "step": 160
-    },
-    {
-      "epoch": 0.027450249087566932,
-      "grad_norm": 0.3477385938167572,
-      "learning_rate": 9.725490196078432e-06,
-      "loss": 0.1822,
-      "step": 161
-    },
-    {
-      "epoch": 0.027620747529104615,
-      "grad_norm": 0.300945907831192,
-      "learning_rate": 9.72378516624041e-06,
-      "loss": 0.1916,
-      "step": 162
-    },
-    {
-      "epoch": 0.0277912459706423,
-      "grad_norm": 0.37067076563835144,
-      "learning_rate": 9.722080136402388e-06,
-      "loss": 0.1904,
-      "step": 163
-    },
-    {
-      "epoch": 0.02796174441217998,
-      "grad_norm": 0.5103992223739624,
-      "learning_rate": 9.720375106564366e-06,
-      "loss": 0.2164,
-      "step": 164
-    },
-    {
-      "epoch": 0.028132242853717664,
-      "grad_norm": 0.7241944074630737,
-      "learning_rate": 9.718670076726344e-06,
-      "loss": 0.2236,
-      "step": 165
-    },
-    {
-      "epoch": 0.028302741295255347,
-      "grad_norm": 0.6501290202140808,
-      "learning_rate": 9.716965046888322e-06,
-      "loss": 0.0946,
-      "step": 166
-    },
-    {
-      "epoch": 0.02847323973679303,
-      "grad_norm": 0.3597775101661682,
-      "learning_rate": 9.7152600170503e-06,
-      "loss": 0.1611,
-      "step": 167
-    },
-    {
-      "epoch": 0.028643738178330713,
-      "grad_norm": 0.563456654548645,
-      "learning_rate": 9.713554987212277e-06,
-      "loss": 0.2554,
-      "step": 168
-    },
-    {
-      "epoch": 0.028814236619868396,
-      "grad_norm": 0.3869915008544922,
-      "learning_rate": 9.711849957374255e-06,
-      "loss": 0.2054,
-      "step": 169
-    },
-    {
-      "epoch": 0.02898473506140608,
-      "grad_norm": 0.45588812232017517,
-      "learning_rate": 9.710144927536233e-06,
-      "loss": 0.2157,
-      "step": 170
-    },
-    {
-      "epoch": 0.029155233502943762,
-      "grad_norm": 0.32176175713539124,
-      "learning_rate": 9.708439897698211e-06,
-      "loss": 0.1984,
-      "step": 171
-    },
-    {
-      "epoch": 0.029325731944481445,
-      "grad_norm": 0.5632541179656982,
-      "learning_rate": 9.706734867860187e-06,
-      "loss": 0.1409,
-      "step": 172
-    },
-    {
-      "epoch": 0.029496230386019128,
-      "grad_norm": 0.3443940281867981,
-      "learning_rate": 9.705029838022167e-06,
-      "loss": 0.1957,
-      "step": 173
-    },
-    {
-      "epoch": 0.02966672882755681,
-      "grad_norm": 0.5562809109687805,
-      "learning_rate": 9.703324808184143e-06,
-      "loss": 0.2574,
-      "step": 174
-    },
-    {
-      "epoch": 0.029837227269094494,
-      "grad_norm": 0.35625970363616943,
-      "learning_rate": 9.701619778346121e-06,
-      "loss": 0.2414,
-      "step": 175
-    },
-    {
-      "epoch": 0.030007725710632177,
-      "grad_norm": 0.32852473855018616,
-      "learning_rate": 9.699914748508101e-06,
-      "loss": 0.2019,
-      "step": 176
-    },
-    {
-      "epoch": 0.03017822415216986,
-      "grad_norm": 0.6662753224372864,
-      "learning_rate": 9.698209718670077e-06,
-      "loss": 0.2822,
-      "step": 177
-    },
-    {
-      "epoch": 0.030348722593707543,
-      "grad_norm": 0.5043244361877441,
-      "learning_rate": 9.696504688832055e-06,
-      "loss": 0.208,
-      "step": 178
-    },
-    {
-      "epoch": 0.030519221035245226,
-      "grad_norm": 0.542964518070221,
-      "learning_rate": 9.694799658994033e-06,
-      "loss": 0.2004,
-      "step": 179
-    },
-    {
-      "epoch": 0.03068971947678291,
-      "grad_norm": 0.36917969584465027,
-      "learning_rate": 9.693094629156011e-06,
-      "loss": 0.22,
-      "step": 180
-    },
-    {
-      "epoch": 0.030860217918320592,
-      "grad_norm": 0.5634118914604187,
-      "learning_rate": 9.691389599317989e-06,
-      "loss": 0.2737,
-      "step": 181
-    },
-    {
-      "epoch": 0.03103071635985827,
-      "grad_norm": 0.3653755187988281,
-      "learning_rate": 9.689684569479967e-06,
-      "loss": 0.2528,
-      "step": 182
-    },
-    {
-      "epoch": 0.031201214801395954,
-      "grad_norm": 0.5282630920410156,
-      "learning_rate": 9.687979539641943e-06,
-      "loss": 0.1684,
-      "step": 183
-    },
-    {
-      "epoch": 0.03137171324293364,
-      "grad_norm": 0.5614996552467346,
-      "learning_rate": 9.686274509803923e-06,
-      "loss": 0.1646,
-      "step": 184
-    },
-    {
-      "epoch": 0.03154221168447132,
-      "grad_norm": 0.6981064677238464,
-      "learning_rate": 9.6845694799659e-06,
-      "loss": 0.15,
-      "step": 185
-    },
-    {
-      "epoch": 0.03171271012600901,
-      "grad_norm": 0.5235413312911987,
-      "learning_rate": 9.682864450127877e-06,
-      "loss": 0.2261,
-      "step": 186
-    },
-    {
-      "epoch": 0.031883208567546686,
-      "grad_norm": 0.4519073963165283,
-      "learning_rate": 9.681159420289857e-06,
-      "loss": 0.215,
-      "step": 187
-    },
-    {
-      "epoch": 0.03205370700908437,
-      "grad_norm": 0.34490248560905457,
-      "learning_rate": 9.679454390451833e-06,
-      "loss": 0.1827,
-      "step": 188
-    },
-    {
-      "epoch": 0.03222420545062205,
-      "grad_norm": 0.5453905463218689,
-      "learning_rate": 9.677749360613813e-06,
-      "loss": 0.1213,
-      "step": 189
-    },
-    {
-      "epoch": 0.03239470389215974,
-      "grad_norm": 0.42571887373924255,
-      "learning_rate": 9.676044330775789e-06,
-      "loss": 0.1913,
-      "step": 190
-    },
-    {
-      "epoch": 0.03256520233369742,
-      "grad_norm": 0.6918254494667053,
-      "learning_rate": 9.674339300937767e-06,
-      "loss": 0.2399,
-      "step": 191
-    },
-    {
-      "epoch": 0.032735700775235105,
-      "grad_norm": 0.3514208495616913,
-      "learning_rate": 9.672634271099745e-06,
-      "loss": 0.171,
-      "step": 192
-    },
-    {
-      "epoch": 0.032906199216772784,
-      "grad_norm": 0.3752657473087311,
-      "learning_rate": 9.670929241261723e-06,
-      "loss": 0.1189,
-      "step": 193
-    },
-    {
-      "epoch": 0.033076697658310464,
-      "grad_norm": 0.7846695780754089,
-      "learning_rate": 9.6692242114237e-06,
-      "loss": 0.2253,
-      "step": 194
-    },
-    {
-      "epoch": 0.03324719609984815,
-      "grad_norm": 0.5222557187080383,
-      "learning_rate": 9.667519181585679e-06,
-      "loss": 0.1143,
-      "step": 195
-    },
-    {
-      "epoch": 0.03341769454138583,
-      "grad_norm": 0.5478008389472961,
-      "learning_rate": 9.665814151747657e-06,
-      "loss": 0.1919,
-      "step": 196
-    },
-    {
-      "epoch": 0.033588192982923516,
-      "grad_norm": 0.8064622282981873,
-      "learning_rate": 9.664109121909634e-06,
-      "loss": 0.2858,
-      "step": 197
-    },
-    {
-      "epoch": 0.033758691424461196,
-      "grad_norm": 0.33207300305366516,
-      "learning_rate": 9.662404092071612e-06,
-      "loss": 0.182,
-      "step": 198
-    },
-    {
-      "epoch": 0.03392918986599888,
-      "grad_norm": 0.5906654596328735,
-      "learning_rate": 9.660699062233589e-06,
-      "loss": 0.1758,
-      "step": 199
-    },
-    {
-      "epoch": 0.03409968830753656,
-      "grad_norm": 0.48277583718299866,
-      "learning_rate": 9.658994032395568e-06,
-      "loss": 0.2225,
-      "step": 200
-    },
-    {
-      "epoch": 0.03409968830753656,
-      "eval_f1_score": 0.0,
-      "eval_loss": 0.21619169414043427,
-      "eval_runtime": 183.3012,
-      "eval_samples_per_second": 54.555,
-      "eval_steps_per_second": 3.41,
-      "step": 200
-    },
-    {
-      "epoch": 0.03427018674907425,
-      "grad_norm": 0.425251305103302,
-      "learning_rate": 9.657289002557546e-06,
-      "loss": 0.2334,
-      "step": 201
-    },
-    {
-      "epoch": 0.03444068519061193,
-      "grad_norm": 0.6187649965286255,
-      "learning_rate": 9.655583972719523e-06,
-      "loss": 0.2057,
-      "step": 202
-    },
-    {
-      "epoch": 0.034611183632149614,
-      "grad_norm": 0.582363486289978,
-      "learning_rate": 9.653878942881502e-06,
-      "loss": 0.1884,
-      "step": 203
-    },
-    {
-      "epoch": 0.034781682073687294,
-      "grad_norm": 0.5536232590675354,
-      "learning_rate": 9.652173913043478e-06,
-      "loss": 0.1723,
-      "step": 204
-    },
-    {
-      "epoch": 0.03495218051522498,
-      "grad_norm": 0.43162399530410767,
-      "learning_rate": 9.650468883205456e-06,
-      "loss": 0.2154,
-      "step": 205
-    },
-    {
-      "epoch": 0.03512267895676266,
-      "grad_norm": 0.887392520904541,
-      "learning_rate": 9.648763853367434e-06,
-      "loss": 0.2948,
-      "step": 206
-    },
-    {
-      "epoch": 0.035293177398300346,
-      "grad_norm": 0.3913290500640869,
-      "learning_rate": 9.647058823529412e-06,
-      "loss": 0.1446,
-      "step": 207
-    },
-    {
-      "epoch": 0.035463675839838026,
-      "grad_norm": 0.32068392634391785,
-      "learning_rate": 9.64535379369139e-06,
-      "loss": 0.1863,
-      "step": 208
-    },
-    {
-      "epoch": 0.03563417428137571,
-      "grad_norm": 0.4926113784313202,
-      "learning_rate": 9.643648763853368e-06,
-      "loss": 0.2029,
-      "step": 209
-    },
-    {
-      "epoch": 0.03580467272291339,
-      "grad_norm": 0.6515610218048096,
-      "learning_rate": 9.641943734015346e-06,
-      "loss": 0.2609,
-      "step": 210
-    },
-    {
-      "epoch": 0.03597517116445108,
-      "grad_norm": 0.3915189802646637,
-      "learning_rate": 9.640238704177324e-06,
-      "loss": 0.1486,
-      "step": 211
-    },
-    {
-      "epoch": 0.03614566960598876,
-      "grad_norm": 0.8154141306877136,
-      "learning_rate": 9.638533674339302e-06,
-      "loss": 0.2535,
-      "step": 212
-    },
-    {
-      "epoch": 0.036316168047526444,
-      "grad_norm": 0.4246624708175659,
-      "learning_rate": 9.63682864450128e-06,
-      "loss": 0.1773,
-      "step": 213
-    },
-    {
-      "epoch": 0.03648666648906412,
-      "grad_norm": 0.37893757224082947,
-      "learning_rate": 9.635123614663258e-06,
-      "loss": 0.2304,
-      "step": 214
-    },
-    {
-      "epoch": 0.0366571649306018,
-      "grad_norm": 0.4338793158531189,
-      "learning_rate": 9.633418584825234e-06,
-      "loss": 0.2138,
-      "step": 215
-    },
-    {
-      "epoch": 0.03682766337213949,
-      "grad_norm": 1.151077151298523,
-      "learning_rate": 9.631713554987214e-06,
-      "loss": 0.1226,
-      "step": 216
-    },
-    {
-      "epoch": 0.03699816181367717,
-      "grad_norm": 0.7071660161018372,
-      "learning_rate": 9.63000852514919e-06,
-      "loss": 0.2902,
-      "step": 217
-    },
-    {
-      "epoch": 0.037168660255214855,
-      "grad_norm": 0.8551703095436096,
-      "learning_rate": 9.628303495311168e-06,
-      "loss": 0.1265,
-      "step": 218
-    },
-    {
-      "epoch": 0.037339158696752535,
-      "grad_norm": 0.6096398234367371,
-      "learning_rate": 9.626598465473148e-06,
-      "loss": 0.1407,
-      "step": 219
-    },
-    {
-      "epoch": 0.03750965713829022,
-      "grad_norm": 0.6498945355415344,
-      "learning_rate": 9.624893435635124e-06,
-      "loss": 0.2396,
-      "step": 220
-    },
-    {
-      "epoch": 0.0376801555798279,
-      "grad_norm": 0.8286121487617493,
-      "learning_rate": 9.623188405797102e-06,
-      "loss": 0.2918,
-      "step": 221
-    },
-    {
-      "epoch": 0.03785065402136559,
-      "grad_norm": 1.0903311967849731,
-      "learning_rate": 9.62148337595908e-06,
-      "loss": 0.2785,
-      "step": 222
-    },
-    {
-      "epoch": 0.03802115246290327,
-      "grad_norm": 0.5139489769935608,
-      "learning_rate": 9.619778346121058e-06,
-      "loss": 0.1452,
-      "step": 223
-    },
-    {
-      "epoch": 0.03819165090444095,
-      "grad_norm": 0.5093590617179871,
-      "learning_rate": 9.618073316283036e-06,
-      "loss": 0.1903,
-      "step": 224
-    },
-    {
-      "epoch": 0.03836214934597863,
-      "grad_norm": 0.548340916633606,
-      "learning_rate": 9.616368286445014e-06,
-      "loss": 0.1516,
-      "step": 225
-    },
-    {
-      "epoch": 0.03853264778751632,
-      "grad_norm": 0.46318405866622925,
-      "learning_rate": 9.614663256606992e-06,
-      "loss": 0.202,
-      "step": 226
-    },
-    {
-      "epoch": 0.038703146229054,
-      "grad_norm": 0.5825653076171875,
-      "learning_rate": 9.61295822676897e-06,
-      "loss": 0.2477,
-      "step": 227
-    },
-    {
-      "epoch": 0.038873644670591685,
-      "grad_norm": 0.9491086602210999,
-      "learning_rate": 9.611253196930947e-06,
-      "loss": 0.1244,
-      "step": 228
-    },
-    {
-      "epoch": 0.039044143112129365,
-      "grad_norm": 0.6441144943237305,
-      "learning_rate": 9.609548167092925e-06,
-      "loss": 0.1457,
-      "step": 229
-    },
-    {
-      "epoch": 0.03921464155366705,
-      "grad_norm": 0.8043190836906433,
-      "learning_rate": 9.607843137254903e-06,
-      "loss": 0.1355,
-      "step": 230
-    },
-    {
-      "epoch": 0.03938513999520473,
-      "grad_norm": 0.5751570463180542,
-      "learning_rate": 9.60613810741688e-06,
-      "loss": 0.2567,
-      "step": 231
-    },
-    {
-      "epoch": 0.03955563843674242,
-      "grad_norm": 0.4224618375301361,
-      "learning_rate": 9.60443307757886e-06,
-      "loss": 0.187,
-      "step": 232
-    },
-    {
-      "epoch": 0.0397261368782801,
-      "grad_norm": 0.3785933256149292,
-      "learning_rate": 9.602728047740836e-06,
-      "loss": 0.2013,
-      "step": 233
-    },
-    {
-      "epoch": 0.03989663531981778,
-      "grad_norm": 0.4648973345756531,
-      "learning_rate": 9.601023017902813e-06,
-      "loss": 0.1626,
-      "step": 234
-    },
-    {
-      "epoch": 0.04006713376135546,
-      "grad_norm": 0.5512021780014038,
-      "learning_rate": 9.599317988064793e-06,
-      "loss": 0.214,
-      "step": 235
-    },
-    {
-      "epoch": 0.04023763220289314,
-      "grad_norm": 0.7625359296798706,
-      "learning_rate": 9.59761295822677e-06,
-      "loss": 0.1994,
-      "step": 236
-    },
-    {
-      "epoch": 0.04040813064443083,
-      "grad_norm": 0.34150898456573486,
-      "learning_rate": 9.595907928388747e-06,
-      "loss": 0.1718,
-      "step": 237
-    },
-    {
-      "epoch": 0.04057862908596851,
-      "grad_norm": 0.3384284973144531,
-      "learning_rate": 9.594202898550725e-06,
-      "loss": 0.144,
-      "step": 238
-    },
-    {
-      "epoch": 0.040749127527506195,
-      "grad_norm": 0.29428425431251526,
-      "learning_rate": 9.592497868712703e-06,
-      "loss": 0.167,
-      "step": 239
-    },
-    {
-      "epoch": 0.040919625969043874,
-      "grad_norm": 1.1551722288131714,
-      "learning_rate": 9.590792838874681e-06,
-      "loss": 0.338,
-      "step": 240
-    },
-    {
-      "epoch": 0.04109012441058156,
-      "grad_norm": 0.37451595067977905,
-      "learning_rate": 9.589087809036659e-06,
-      "loss": 0.2053,
-      "step": 241
-    },
-    {
-      "epoch": 0.04126062285211924,
-      "grad_norm": 0.3323061168193817,
-      "learning_rate": 9.587382779198637e-06,
-      "loss": 0.2096,
-      "step": 242
-    },
-    {
-      "epoch": 0.041431121293656926,
-      "grad_norm": 0.36669957637786865,
-      "learning_rate": 9.585677749360615e-06,
-      "loss": 0.1777,
-      "step": 243
-    },
-    {
-      "epoch": 0.041601619735194606,
-      "grad_norm": 0.5516036152839661,
-      "learning_rate": 9.583972719522593e-06,
-      "loss": 0.1608,
-      "step": 244
-    },
-    {
-      "epoch": 0.04177211817673229,
-      "grad_norm": 0.47251853346824646,
-      "learning_rate": 9.582267689684571e-06,
-      "loss": 0.1808,
-      "step": 245
-    },
-    {
-      "epoch": 0.04194261661826997,
-      "grad_norm": 0.49306562542915344,
-      "learning_rate": 9.580562659846549e-06,
-      "loss": 0.2067,
-      "step": 246
-    },
-    {
-      "epoch": 0.04211311505980766,
-      "grad_norm": 0.5047063231468201,
-      "learning_rate": 9.578857630008525e-06,
-      "loss": 0.2037,
-      "step": 247
-    },
-    {
-      "epoch": 0.04228361350134534,
-      "grad_norm": 0.4270981550216675,
-      "learning_rate": 9.577152600170505e-06,
-      "loss": 0.1763,
-      "step": 248
-    },
-    {
-      "epoch": 0.042454111942883024,
-      "grad_norm": 0.7507210969924927,
-      "learning_rate": 9.575447570332481e-06,
-      "loss": 0.3042,
-      "step": 249
-    },
-    {
-      "epoch": 0.042624610384420704,
-      "grad_norm": 0.39905881881713867,
-      "learning_rate": 9.573742540494459e-06,
-      "loss": 0.226,
-      "step": 250
-    },
-    {
-      "epoch": 0.04279510882595839,
-      "grad_norm": 0.29342296719551086,
-      "learning_rate": 9.572037510656439e-06,
-      "loss": 0.2141,
-      "step": 251
-    },
-    {
-      "epoch": 0.04296560726749607,
-      "grad_norm": 0.31684985756874084,
-      "learning_rate": 9.570332480818415e-06,
-      "loss": 0.2133,
-      "step": 252
-    },
-    {
-      "epoch": 0.043136105709033756,
-      "grad_norm": 0.4727517366409302,
-      "learning_rate": 9.568627450980393e-06,
-      "loss": 0.1621,
-      "step": 253
-    },
-    {
-      "epoch": 0.043306604150571436,
-      "grad_norm": 0.35550886392593384,
-      "learning_rate": 9.56692242114237e-06,
-      "loss": 0.196,
-      "step": 254
-    },
-    {
-      "epoch": 0.04347710259210912,
-      "grad_norm": 0.31078484654426575,
-      "learning_rate": 9.565217391304349e-06,
-      "loss": 0.1974,
-      "step": 255
-    },
-    {
-      "epoch": 0.0436476010336468,
-      "grad_norm": 0.31537219882011414,
-      "learning_rate": 9.563512361466327e-06,
-      "loss": 0.1756,
-      "step": 256
-    },
-    {
-      "epoch": 0.04381809947518448,
-      "grad_norm": 0.44731876254081726,
-      "learning_rate": 9.561807331628305e-06,
-      "loss": 0.1415,
-      "step": 257
-    },
-    {
-      "epoch": 0.04398859791672217,
-      "grad_norm": 0.2439146786928177,
-      "learning_rate": 9.56010230179028e-06,
-      "loss": 0.1922,
-      "step": 258
-    },
-    {
-      "epoch": 0.04415909635825985,
-      "grad_norm": 0.3228803277015686,
-      "learning_rate": 9.55839727195226e-06,
-      "loss": 0.167,
-      "step": 259
-    },
-    {
-      "epoch": 0.044329594799797534,
-      "grad_norm": 0.34834024310112,
-      "learning_rate": 9.556692242114238e-06,
-      "loss": 0.1194,
-      "step": 260
-    },
-    {
-      "epoch": 0.04450009324133521,
-      "grad_norm": 0.49005141854286194,
-      "learning_rate": 9.554987212276215e-06,
-      "loss": 0.0857,
-      "step": 261
-    },
-    {
-      "epoch": 0.0446705916828729,
-      "grad_norm": 0.5683344602584839,
-      "learning_rate": 9.553282182438194e-06,
-      "loss": 0.2118,
-      "step": 262
-    },
-    {
-      "epoch": 0.04484109012441058,
-      "grad_norm": 0.3251049220561981,
-      "learning_rate": 9.55157715260017e-06,
-      "loss": 0.1616,
-      "step": 263
-    },
-    {
-      "epoch": 0.045011588565948266,
-      "grad_norm": 0.220848947763443,
-      "learning_rate": 9.54987212276215e-06,
-      "loss": 0.149,
-      "step": 264
-    },
-    {
-      "epoch": 0.045182087007485945,
-      "grad_norm": 0.4781118929386139,
-      "learning_rate": 9.548167092924126e-06,
-      "loss": 0.175,
-      "step": 265
-    },
-    {
-      "epoch": 0.04535258544902363,
-      "grad_norm": 0.8802744150161743,
-      "learning_rate": 9.546462063086104e-06,
-      "loss": 0.2548,
-      "step": 266
-    },
-    {
-      "epoch": 0.04552308389056131,
-      "grad_norm": 1.0413671731948853,
-      "learning_rate": 9.544757033248082e-06,
-      "loss": 0.2834,
-      "step": 267
-    },
-    {
-      "epoch": 0.045693582332099,
-      "grad_norm": 0.687307596206665,
-      "learning_rate": 9.54305200341006e-06,
-      "loss": 0.2089,
-      "step": 268
-    },
-    {
-      "epoch": 0.04586408077363668,
-      "grad_norm": 0.4433498978614807,
-      "learning_rate": 9.541346973572038e-06,
-      "loss": 0.195,
-      "step": 269
-    },
-    {
-      "epoch": 0.046034579215174364,
-      "grad_norm": 0.470577210187912,
-      "learning_rate": 9.539641943734016e-06,
-      "loss": 0.1615,
-      "step": 270
-    },
-    {
-      "epoch": 0.04620507765671204,
-      "grad_norm": 0.3836967945098877,
-      "learning_rate": 9.537936913895994e-06,
-      "loss": 0.1634,
-      "step": 271
-    },
-    {
-      "epoch": 0.04637557609824973,
-      "grad_norm": 0.5679737329483032,
-      "learning_rate": 9.536231884057972e-06,
-      "loss": 0.2454,
-      "step": 272
-    },
-    {
-      "epoch": 0.04654607453978741,
-      "grad_norm": 0.4014706015586853,
-      "learning_rate": 9.53452685421995e-06,
-      "loss": 0.2199,
-      "step": 273
-    },
-    {
-      "epoch": 0.046716572981325095,
-      "grad_norm": 0.8329876065254211,
-      "learning_rate": 9.532821824381926e-06,
-      "loss": 0.1728,
-      "step": 274
-    },
-    {
-      "epoch": 0.046887071422862775,
-      "grad_norm": 1.1290709972381592,
-      "learning_rate": 9.531116794543906e-06,
-      "loss": 0.1316,
-      "step": 275
-    },
-    {
-      "epoch": 0.04705756986440046,
-      "grad_norm": 0.7160772085189819,
-      "learning_rate": 9.529411764705882e-06,
-      "loss": 0.1479,
-      "step": 276
-    },
-    {
-      "epoch": 0.04722806830593814,
-      "grad_norm": 0.5050806403160095,
-      "learning_rate": 9.52770673486786e-06,
-      "loss": 0.1707,
-      "step": 277
-    },
-    {
-      "epoch": 0.04739856674747582,
-      "grad_norm": 0.4004954993724823,
-      "learning_rate": 9.52600170502984e-06,
-      "loss": 0.1797,
-      "step": 278
-    },
-    {
-      "epoch": 0.04756906518901351,
-      "grad_norm": 0.5790531039237976,
-      "learning_rate": 9.524296675191816e-06,
-      "loss": 0.2352,
-      "step": 279
-    },
-    {
-      "epoch": 0.047739563630551186,
-      "grad_norm": 1.0683724880218506,
-      "learning_rate": 9.522591645353794e-06,
-      "loss": 0.2977,
-      "step": 280
-    },
-    {
-      "epoch": 0.04791006207208887,
-      "grad_norm": 0.44837862253189087,
-      "learning_rate": 9.520886615515772e-06,
-      "loss": 0.1652,
-      "step": 281
-    },
-    {
-      "epoch": 0.04808056051362655,
-      "grad_norm": 0.3902139961719513,
-      "learning_rate": 9.51918158567775e-06,
-      "loss": 0.1839,
-      "step": 282
-    },
-    {
-      "epoch": 0.04825105895516424,
-      "grad_norm": 0.8578200936317444,
-      "learning_rate": 9.517476555839728e-06,
-      "loss": 0.2493,
-      "step": 283
-    },
-    {
-      "epoch": 0.04842155739670192,
-      "grad_norm": 0.8971724510192871,
-      "learning_rate": 9.515771526001706e-06,
-      "loss": 0.2395,
-      "step": 284
-    },
-    {
-      "epoch": 0.048592055838239605,
-      "grad_norm": 0.3035443127155304,
-      "learning_rate": 9.514066496163684e-06,
-      "loss": 0.1992,
-      "step": 285
-    },
-    {
-      "epoch": 0.048762554279777284,
-      "grad_norm": 1.0432074069976807,
-      "learning_rate": 9.512361466325662e-06,
-      "loss": 0.2962,
-      "step": 286
-    },
-    {
-      "epoch": 0.04893305272131497,
-      "grad_norm": 0.7007878422737122,
-      "learning_rate": 9.51065643648764e-06,
-      "loss": 0.2185,
-      "step": 287
-    },
-    {
-      "epoch": 0.04910355116285265,
-      "grad_norm": 0.8017603754997253,
-      "learning_rate": 9.508951406649618e-06,
-      "loss": 0.2113,
-      "step": 288
-    },
-    {
-      "epoch": 0.04927404960439034,
-      "grad_norm": 1.4467297792434692,
-      "learning_rate": 9.507246376811596e-06,
-      "loss": 0.1355,
-      "step": 289
-    },
-    {
-      "epoch": 0.049444548045928016,
-      "grad_norm": 1.106168270111084,
-      "learning_rate": 9.505541346973572e-06,
-      "loss": 0.1692,
-      "step": 290
-    },
-    {
-      "epoch": 0.0496150464874657,
-      "grad_norm": 0.5016661882400513,
-      "learning_rate": 9.503836317135551e-06,
-      "loss": 0.2279,
-      "step": 291
-    },
-    {
-      "epoch": 0.04978554492900338,
-      "grad_norm": 0.644110381603241,
-      "learning_rate": 9.502131287297528e-06,
-      "loss": 0.1934,
-      "step": 292
-    },
-    {
-      "epoch": 0.04995604337054107,
-      "grad_norm": 0.49578043818473816,
-      "learning_rate": 9.500426257459506e-06,
-      "loss": 0.1911,
-      "step": 293
-    },
-    {
-      "epoch": 0.05012654181207875,
-      "grad_norm": 0.42051151394844055,
-      "learning_rate": 9.498721227621485e-06,
-      "loss": 0.1954,
-      "step": 294
-    },
-    {
-      "epoch": 0.050297040253616435,
-      "grad_norm": 0.42945224046707153,
-      "learning_rate": 9.497016197783462e-06,
-      "loss": 0.1173,
-      "step": 295
-    },
-    {
-      "epoch": 0.050467538695154114,
-      "grad_norm": 0.6755464673042297,
-      "learning_rate": 9.49531116794544e-06,
-      "loss": 0.2033,
-      "step": 296
-    },
-    {
-      "epoch": 0.0506380371366918,
-      "grad_norm": 0.294484943151474,
-      "learning_rate": 9.493606138107417e-06,
-      "loss": 0.1475,
-      "step": 297
-    },
-    {
-      "epoch": 0.05080853557822948,
-      "grad_norm": 1.1652215719223022,
-      "learning_rate": 9.491901108269395e-06,
-      "loss": 0.2815,
-      "step": 298
-    },
-    {
-      "epoch": 0.050979034019767167,
-      "grad_norm": 0.4456086754798889,
-      "learning_rate": 9.490196078431373e-06,
-      "loss": 0.1949,
-      "step": 299
-    },
-    {
-      "epoch": 0.051149532461304846,
-      "grad_norm": 0.8742743730545044,
-      "learning_rate": 9.488491048593351e-06,
-      "loss": 0.2193,
-      "step": 300
-    },
-    {
-      "epoch": 0.051149532461304846,
-      "eval_f1_score": 0.0,
-      "eval_loss": 0.21666215360164642,
-      "eval_runtime": 183.2895,
-      "eval_samples_per_second": 54.559,
-      "eval_steps_per_second": 3.41,
-      "step": 300
-    },
-    {
-      "epoch": 0.051320030902842526,
-      "grad_norm": 0.7776352763175964,
-      "learning_rate": 9.48678601875533e-06,
-      "loss": 0.1994,
-      "step": 301
-    },
-    {
-      "epoch": 0.05149052934438021,
-      "grad_norm": 0.6161231398582458,
-      "learning_rate": 9.485080988917307e-06,
-      "loss": 0.1971,
-      "step": 302
-    },
-    {
-      "epoch": 0.05166102778591789,
-      "grad_norm": 0.4451903700828552,
-      "learning_rate": 9.483375959079285e-06,
-      "loss": 0.1922,
-      "step": 303
-    },
-    {
-      "epoch": 0.05183152622745558,
-      "grad_norm": 0.5707091093063354,
-      "learning_rate": 9.481670929241263e-06,
-      "loss": 0.203,
-      "step": 304
-    },
-    {
-      "epoch": 0.05200202466899326,
-      "grad_norm": 0.5663946866989136,
-      "learning_rate": 9.479965899403241e-06,
-      "loss": 0.1883,
-      "step": 305
-    },
-    {
-      "epoch": 0.052172523110530944,
-      "grad_norm": 0.7128487229347229,
-      "learning_rate": 9.478260869565217e-06,
-      "loss": 0.2129,
-      "step": 306
-    },
-    {
-      "epoch": 0.05234302155206862,
-      "grad_norm": 0.6801848411560059,
-      "learning_rate": 9.476555839727197e-06,
-      "loss": 0.226,
-      "step": 307
-    },
-    {
-      "epoch": 0.05251351999360631,
-      "grad_norm": 0.8383636474609375,
-      "learning_rate": 9.474850809889173e-06,
-      "loss": 0.1514,
-      "step": 308
-    },
-    {
-      "epoch": 0.05268401843514399,
-      "grad_norm": 0.9758368730545044,
-      "learning_rate": 9.473145780051151e-06,
-      "loss": 0.0985,
-      "step": 309
-    },
-    {
-      "epoch": 0.052854516876681676,
-      "grad_norm": 0.9041672945022583,
-      "learning_rate": 9.47144075021313e-06,
-      "loss": 0.2796,
-      "step": 310
-    },
-    {
-      "epoch": 0.053025015318219355,
-      "grad_norm": 0.721714198589325,
-      "learning_rate": 9.469735720375107e-06,
-      "loss": 0.2241,
-      "step": 311
-    },
-    {
-      "epoch": 0.05319551375975704,
-      "grad_norm": 0.6851015686988831,
-      "learning_rate": 9.468030690537085e-06,
-      "loss": 0.1626,
-      "step": 312
-    },
-    {
-      "epoch": 0.05336601220129472,
-      "grad_norm": 0.914460301399231,
-      "learning_rate": 9.466325660699063e-06,
-      "loss": 0.2942,
-      "step": 313
-    },
-    {
-      "epoch": 0.05353651064283241,
-      "grad_norm": 0.42635348439216614,
-      "learning_rate": 9.46462063086104e-06,
-      "loss": 0.1516,
-      "step": 314
-    },
-    {
-      "epoch": 0.05370700908437009,
-      "grad_norm": 0.4017740786075592,
-      "learning_rate": 9.462915601023019e-06,
-      "loss": 0.1743,
-      "step": 315
-    },
-    {
-      "epoch": 0.053877507525907774,
-      "grad_norm": 0.3991027772426605,
-      "learning_rate": 9.461210571184997e-06,
-      "loss": 0.1942,
-      "step": 316
-    },
-    {
-      "epoch": 0.05404800596744545,
-      "grad_norm": 0.3740632236003876,
-      "learning_rate": 9.459505541346975e-06,
-      "loss": 0.1794,
-      "step": 317
-    },
-    {
-      "epoch": 0.05421850440898314,
-      "grad_norm": 0.8277780413627625,
-      "learning_rate": 9.457800511508953e-06,
-      "loss": 0.207,
-      "step": 318
-    },
-    {
-      "epoch": 0.05438900285052082,
-      "grad_norm": 0.4441271722316742,
-      "learning_rate": 9.45609548167093e-06,
-      "loss": 0.1879,
-      "step": 319
-    },
-    {
-      "epoch": 0.054559501292058506,
-      "grad_norm": 0.4899628460407257,
-      "learning_rate": 9.454390451832909e-06,
-      "loss": 0.195,
-      "step": 320
-    },
-    {
-      "epoch": 0.054729999733596185,
-      "grad_norm": 0.6418739557266235,
-      "learning_rate": 9.452685421994886e-06,
-      "loss": 0.1267,
-      "step": 321
-    },
-    {
-      "epoch": 0.054900498175133865,
-      "grad_norm": 0.9745388031005859,
-      "learning_rate": 9.450980392156863e-06,
-      "loss": 0.2859,
-      "step": 322
-    },
-    {
-      "epoch": 0.05507099661667155,
-      "grad_norm": 0.8999006748199463,
-      "learning_rate": 9.449275362318842e-06,
-      "loss": 0.0801,
-      "step": 323
-    },
-    {
-      "epoch": 0.05524149505820923,
-      "grad_norm": 0.5319007635116577,
-      "learning_rate": 9.447570332480819e-06,
-      "loss": 0.2203,
-      "step": 324
-    },
-    {
-      "epoch": 0.05541199349974692,
-      "grad_norm": 0.7109606266021729,
-      "learning_rate": 9.445865302642797e-06,
-      "loss": 0.2016,
-      "step": 325
-    },
-    {
-      "epoch": 0.0555824919412846,
-      "grad_norm": 0.3947546184062958,
-      "learning_rate": 9.444160272804774e-06,
-      "loss": 0.1562,
-      "step": 326
-    },
-    {
-      "epoch": 0.05575299038282228,
-      "grad_norm": 0.5612919330596924,
-      "learning_rate": 9.442455242966752e-06,
-      "loss": 0.1078,
-      "step": 327
-    },
-    {
-      "epoch": 0.05592348882435996,
-      "grad_norm": 0.4919300079345703,
-      "learning_rate": 9.44075021312873e-06,
-      "loss": 0.1453,
-      "step": 328
-    },
-    {
-      "epoch": 0.05609398726589765,
-      "grad_norm": 1.1095222234725952,
-      "learning_rate": 9.439045183290708e-06,
-      "loss": 0.2045,
-      "step": 329
-    },
-    {
-      "epoch": 0.05626448570743533,
-      "grad_norm": 0.6746119856834412,
-      "learning_rate": 9.437340153452686e-06,
-      "loss": 0.2171,
-      "step": 330
-    },
-    {
-      "epoch": 0.056434984148973015,
-      "grad_norm": 0.6914483308792114,
-      "learning_rate": 9.435635123614664e-06,
-      "loss": 0.208,
-      "step": 331
-    },
-    {
-      "epoch": 0.056605482590510695,
-      "grad_norm": 0.9006068110466003,
-      "learning_rate": 9.433930093776642e-06,
-      "loss": 0.2473,
-      "step": 332
-    },
-    {
-      "epoch": 0.05677598103204838,
-      "grad_norm": 0.5845337510108948,
-      "learning_rate": 9.432225063938618e-06,
-      "loss": 0.153,
-      "step": 333
-    },
-    {
-      "epoch": 0.05694647947358606,
-      "grad_norm": 1.3863515853881836,
-      "learning_rate": 9.430520034100598e-06,
-      "loss": 0.1166,
-      "step": 334
-    },
-    {
-      "epoch": 0.05711697791512375,
-      "grad_norm": 0.9808059334754944,
-      "learning_rate": 9.428815004262576e-06,
-      "loss": 0.1342,
-      "step": 335
-    },
-    {
-      "epoch": 0.057287476356661426,
-      "grad_norm": 0.6818749904632568,
-      "learning_rate": 9.427109974424552e-06,
-      "loss": 0.2624,
-      "step": 336
-    },
-    {
-      "epoch": 0.05745797479819911,
-      "grad_norm": 1.034446120262146,
-      "learning_rate": 9.425404944586532e-06,
-      "loss": 0.2494,
-      "step": 337
-    },
-    {
-      "epoch": 0.05762847323973679,
-      "grad_norm": 0.5185638666152954,
-      "learning_rate": 9.423699914748508e-06,
-      "loss": 0.2122,
-      "step": 338
-    },
-    {
-      "epoch": 0.05779897168127448,
-      "grad_norm": 0.5386376976966858,
-      "learning_rate": 9.421994884910488e-06,
-      "loss": 0.1788,
-      "step": 339
-    },
-    {
-      "epoch": 0.05796947012281216,
-      "grad_norm": 0.6817571520805359,
-      "learning_rate": 9.420289855072464e-06,
-      "loss": 0.2466,
-      "step": 340
-    },
-    {
-      "epoch": 0.058139968564349845,
-      "grad_norm": 0.5243096947669983,
-      "learning_rate": 9.418584825234442e-06,
-      "loss": 0.1567,
-      "step": 341
-    },
-    {
-      "epoch": 0.058310467005887524,
-      "grad_norm": 0.9190734624862671,
-      "learning_rate": 9.41687979539642e-06,
-      "loss": 0.2006,
-      "step": 342
-    },
-    {
-      "epoch": 0.058480965447425204,
-      "grad_norm": 0.659359872341156,
-      "learning_rate": 9.415174765558398e-06,
-      "loss": 0.1342,
-      "step": 343
-    },
-    {
-      "epoch": 0.05865146388896289,
-      "grad_norm": 0.637925922870636,
-      "learning_rate": 9.413469735720376e-06,
-      "loss": 0.1694,
-      "step": 344
-    },
-    {
-      "epoch": 0.05882196233050057,
-      "grad_norm": 1.2218626737594604,
-      "learning_rate": 9.411764705882354e-06,
-      "loss": 0.2691,
-      "step": 345
-    },
-    {
-      "epoch": 0.058992460772038256,
-      "grad_norm": 0.6028181314468384,
-      "learning_rate": 9.410059676044332e-06,
-      "loss": 0.1151,
-      "step": 346
-    },
-    {
-      "epoch": 0.059162959213575936,
-      "grad_norm": 1.4712268114089966,
-      "learning_rate": 9.40835464620631e-06,
-      "loss": 0.2712,
-      "step": 347
-    },
-    {
-      "epoch": 0.05933345765511362,
-      "grad_norm": 0.5645212531089783,
-      "learning_rate": 9.406649616368288e-06,
-      "loss": 0.1584,
-      "step": 348
-    },
-    {
-      "epoch": 0.0595039560966513,
-      "grad_norm": 0.7886714339256287,
-      "learning_rate": 9.404944586530264e-06,
-      "loss": 0.1549,
-      "step": 349
-    },
-    {
-      "epoch": 0.05967445453818899,
-      "grad_norm": 0.8502355813980103,
-      "learning_rate": 9.403239556692244e-06,
-      "loss": 0.095,
-      "step": 350
-    },
-    {
-      "epoch": 0.05984495297972667,
-      "grad_norm": 0.5809105634689331,
-      "learning_rate": 9.40153452685422e-06,
-      "loss": 0.1412,
-      "step": 351
-    },
-    {
-      "epoch": 0.060015451421264354,
-      "grad_norm": 0.7457136511802673,
-      "learning_rate": 9.399829497016198e-06,
-      "loss": 0.2125,
-      "step": 352
-    },
-    {
-      "epoch": 0.060185949862802034,
-      "grad_norm": 0.8455162048339844,
-      "learning_rate": 9.398124467178177e-06,
-      "loss": 0.1107,
-      "step": 353
-    },
-    {
-      "epoch": 0.06035644830433972,
-      "grad_norm": 0.5305351614952087,
-      "learning_rate": 9.396419437340154e-06,
-      "loss": 0.1568,
-      "step": 354
-    },
-    {
-      "epoch": 0.0605269467458774,
-      "grad_norm": 0.6263880729675293,
-      "learning_rate": 9.394714407502132e-06,
-      "loss": 0.1768,
-      "step": 355
-    },
-    {
-      "epoch": 0.060697445187415086,
-      "grad_norm": 0.6323095560073853,
-      "learning_rate": 9.39300937766411e-06,
-      "loss": 0.1563,
-      "step": 356
-    },
-    {
-      "epoch": 0.060867943628952766,
-      "grad_norm": 0.9921282529830933,
-      "learning_rate": 9.391304347826087e-06,
-      "loss": 0.2141,
-      "step": 357
-    },
-    {
-      "epoch": 0.06103844207049045,
-      "grad_norm": 1.3965635299682617,
-      "learning_rate": 9.389599317988065e-06,
-      "loss": 0.1306,
-      "step": 358
-    },
-    {
-      "epoch": 0.06120894051202813,
-      "grad_norm": 0.6225385069847107,
-      "learning_rate": 9.387894288150043e-06,
-      "loss": 0.1945,
-      "step": 359
-    },
-    {
-      "epoch": 0.06137943895356582,
-      "grad_norm": 1.5763331651687622,
-      "learning_rate": 9.386189258312021e-06,
-      "loss": 0.3171,
-      "step": 360
-    },
-    {
-      "epoch": 0.0615499373951035,
-      "grad_norm": 0.6186936497688293,
-      "learning_rate": 9.384484228474e-06,
-      "loss": 0.1883,
-      "step": 361
-    },
-    {
-      "epoch": 0.061720435836641184,
-      "grad_norm": 0.9674474596977234,
-      "learning_rate": 9.382779198635977e-06,
-      "loss": 0.2046,
-      "step": 362
-    },
-    {
-      "epoch": 0.061890934278178864,
-      "grad_norm": 1.3343751430511475,
-      "learning_rate": 9.381074168797955e-06,
-      "loss": 0.1711,
-      "step": 363
-    },
-    {
-      "epoch": 0.06206143271971654,
-      "grad_norm": 0.9126800298690796,
-      "learning_rate": 9.379369138959933e-06,
-      "loss": 0.2874,
-      "step": 364
-    },
-    {
-      "epoch": 0.06223193116125423,
-      "grad_norm": 0.7982877492904663,
-      "learning_rate": 9.37766410912191e-06,
-      "loss": 0.2453,
-      "step": 365
-    },
-    {
-      "epoch": 0.06240242960279191,
-      "grad_norm": 0.8203127384185791,
-      "learning_rate": 9.375959079283889e-06,
-      "loss": 0.1784,
-      "step": 366
-    },
-    {
-      "epoch": 0.0625729280443296,
-      "grad_norm": 0.7186422944068909,
-      "learning_rate": 9.374254049445865e-06,
-      "loss": 0.1632,
-      "step": 367
-    },
-    {
-      "epoch": 0.06274342648586728,
-      "grad_norm": 0.7624729871749878,
-      "learning_rate": 9.372549019607843e-06,
-      "loss": 0.1037,
-      "step": 368
-    },
-    {
-      "epoch": 0.06291392492740495,
-      "grad_norm": 0.6697052121162415,
-      "learning_rate": 9.370843989769823e-06,
-      "loss": 0.16,
-      "step": 369
-    },
-    {
-      "epoch": 0.06308442336894264,
-      "grad_norm": 1.1724293231964111,
-      "learning_rate": 9.369138959931799e-06,
-      "loss": 0.2248,
-      "step": 370
-    },
-    {
-      "epoch": 0.06325492181048033,
-      "grad_norm": 1.5148735046386719,
-      "learning_rate": 9.367433930093777e-06,
-      "loss": 0.2683,
-      "step": 371
-    },
-    {
-      "epoch": 0.06342542025201801,
-      "grad_norm": 1.0104899406433105,
-      "learning_rate": 9.365728900255755e-06,
-      "loss": 0.2333,
-      "step": 372
-    },
-    {
-      "epoch": 0.06359591869355569,
-      "grad_norm": 0.5941121578216553,
-      "learning_rate": 9.364023870417733e-06,
-      "loss": 0.1206,
-      "step": 373
-    },
-    {
-      "epoch": 0.06376641713509337,
-      "grad_norm": 0.39642173051834106,
-      "learning_rate": 9.362318840579711e-06,
-      "loss": 0.1378,
-      "step": 374
-    },
-    {
-      "epoch": 0.06393691557663106,
-      "grad_norm": 0.6411215662956238,
-      "learning_rate": 9.360613810741689e-06,
-      "loss": 0.0772,
-      "step": 375
-    },
-    {
-      "epoch": 0.06410741401816875,
-      "grad_norm": 1.834947109222412,
-      "learning_rate": 9.358908780903667e-06,
-      "loss": 0.3382,
-      "step": 376
-    },
-    {
-      "epoch": 0.06427791245970642,
-      "grad_norm": 1.322509765625,
-      "learning_rate": 9.357203751065645e-06,
-      "loss": 0.2573,
-      "step": 377
-    },
-    {
-      "epoch": 0.0644484109012441,
-      "grad_norm": 1.019263505935669,
-      "learning_rate": 9.355498721227623e-06,
-      "loss": 0.1991,
-      "step": 378
-    },
-    {
-      "epoch": 0.06461890934278179,
-      "grad_norm": 1.1220760345458984,
-      "learning_rate": 9.3537936913896e-06,
-      "loss": 0.2133,
-      "step": 379
-    },
-    {
-      "epoch": 0.06478940778431948,
-      "grad_norm": 1.7320735454559326,
-      "learning_rate": 9.352088661551579e-06,
-      "loss": 0.1466,
-      "step": 380
-    },
-    {
-      "epoch": 0.06495990622585715,
-      "grad_norm": 1.1254284381866455,
-      "learning_rate": 9.350383631713555e-06,
-      "loss": 0.2786,
-      "step": 381
-    },
-    {
-      "epoch": 0.06513040466739484,
-      "grad_norm": 0.7988210320472717,
-      "learning_rate": 9.348678601875534e-06,
-      "loss": 0.2236,
-      "step": 382
-    },
-    {
-      "epoch": 0.06530090310893252,
-      "grad_norm": 1.4079415798187256,
-      "learning_rate": 9.34697357203751e-06,
-      "loss": 0.1681,
-      "step": 383
-    },
-    {
-      "epoch": 0.06547140155047021,
-      "grad_norm": 0.8565404415130615,
-      "learning_rate": 9.345268542199489e-06,
-      "loss": 0.2203,
-      "step": 384
-    },
-    {
-      "epoch": 0.06564189999200788,
-      "grad_norm": 0.8782585859298706,
-      "learning_rate": 9.343563512361468e-06,
-      "loss": 0.1947,
-      "step": 385
-    },
-    {
-      "epoch": 0.06581239843354557,
-      "grad_norm": 0.6504030227661133,
-      "learning_rate": 9.341858482523445e-06,
-      "loss": 0.1231,
-      "step": 386
-    },
-    {
-      "epoch": 0.06598289687508326,
-      "grad_norm": 0.7962436079978943,
-      "learning_rate": 9.340153452685423e-06,
-      "loss": 0.1644,
-      "step": 387
-    },
-    {
-      "epoch": 0.06615339531662093,
-      "grad_norm": 1.63428795337677,
-      "learning_rate": 9.3384484228474e-06,
-      "loss": 0.2766,
-      "step": 388
-    },
-    {
-      "epoch": 0.06632389375815861,
-      "grad_norm": 0.98063725233078,
-      "learning_rate": 9.336743393009378e-06,
-      "loss": 0.1638,
-      "step": 389
-    },
-    {
-      "epoch": 0.0664943921996963,
-      "grad_norm": 1.217576503753662,
-      "learning_rate": 9.335038363171356e-06,
-      "loss": 0.2317,
-      "step": 390
-    },
-    {
-      "epoch": 0.06666489064123399,
-      "grad_norm": 0.7941992878913879,
-      "learning_rate": 9.333333333333334e-06,
-      "loss": 0.1589,
-      "step": 391
-    },
-    {
-      "epoch": 0.06683538908277166,
-      "grad_norm": 0.8984582424163818,
-      "learning_rate": 9.331628303495312e-06,
-      "loss": 0.2159,
-      "step": 392
-    },
-    {
-      "epoch": 0.06700588752430935,
-      "grad_norm": 0.7985866665840149,
-      "learning_rate": 9.32992327365729e-06,
-      "loss": 0.1914,
-      "step": 393
-    },
-    {
-      "epoch": 0.06717638596584703,
-      "grad_norm": 0.7738232016563416,
-      "learning_rate": 9.328218243819268e-06,
-      "loss": 0.2207,
-      "step": 394
-    },
-    {
-      "epoch": 0.06734688440738472,
-      "grad_norm": 1.0320310592651367,
-      "learning_rate": 9.326513213981246e-06,
-      "loss": 0.2657,
-      "step": 395
-    },
-    {
-      "epoch": 0.06751738284892239,
-      "grad_norm": 0.8637803196907043,
-      "learning_rate": 9.324808184143224e-06,
-      "loss": 0.2897,
-      "step": 396
-    },
-    {
-      "epoch": 0.06768788129046008,
-      "grad_norm": 1.9716813564300537,
-      "learning_rate": 9.3231031543052e-06,
-      "loss": 0.1362,
-      "step": 397
-    },
-    {
-      "epoch": 0.06785837973199776,
-      "grad_norm": 1.3121943473815918,
-      "learning_rate": 9.32139812446718e-06,
-      "loss": 0.1508,
-      "step": 398
-    },
-    {
-      "epoch": 0.06802887817353545,
-      "grad_norm": 1.190568447113037,
-      "learning_rate": 9.319693094629156e-06,
-      "loss": 0.1805,
-      "step": 399
-    },
-    {
-      "epoch": 0.06819937661507312,
-      "grad_norm": 0.6798188090324402,
-      "learning_rate": 9.317988064791134e-06,
-      "loss": 0.238,
-      "step": 400
-    },
-    {
-      "epoch": 0.06819937661507312,
-      "eval_f1_score": 0.007434944237918215,
-      "eval_loss": 0.19798505306243896,
-      "eval_runtime": 183.2576,
-      "eval_samples_per_second": 54.568,
-      "eval_steps_per_second": 3.411,
-      "step": 400
-    },
-    {
-      "epoch": 0.06836987505661081,
-      "grad_norm": 0.6206487417221069,
-      "learning_rate": 9.316283034953112e-06,
-      "loss": 0.1663,
-      "step": 401
-    },
-    {
-      "epoch": 0.0685403734981485,
-      "grad_norm": 0.9610775113105774,
-      "learning_rate": 9.31457800511509e-06,
-      "loss": 0.2153,
-      "step": 402
-    },
-    {
-      "epoch": 0.06871087193968618,
-      "grad_norm": 0.4641443192958832,
-      "learning_rate": 9.312872975277068e-06,
-      "loss": 0.1053,
-      "step": 403
-    },
-    {
-      "epoch": 0.06888137038122386,
-      "grad_norm": 0.6299108862876892,
-      "learning_rate": 9.311167945439046e-06,
-      "loss": 0.1492,
-      "step": 404
-    },
-    {
-      "epoch": 0.06905186882276154,
-      "grad_norm": 0.42379632592201233,
-      "learning_rate": 9.309462915601024e-06,
-      "loss": 0.0972,
-      "step": 405
-    },
-    {
-      "epoch": 0.06922236726429923,
-      "grad_norm": 1.3610317707061768,
-      "learning_rate": 9.307757885763002e-06,
-      "loss": 0.2578,
-      "step": 406
-    },
-    {
-      "epoch": 0.0693928657058369,
-      "grad_norm": 0.5917848944664001,
-      "learning_rate": 9.30605285592498e-06,
-      "loss": 0.1444,
-      "step": 407
-    },
-    {
-      "epoch": 0.06956336414737459,
-      "grad_norm": 0.9677045345306396,
-      "learning_rate": 9.304347826086956e-06,
-      "loss": 0.2294,
-      "step": 408
-    },
-    {
-      "epoch": 0.06973386258891227,
-      "grad_norm": 0.594524621963501,
-      "learning_rate": 9.302642796248936e-06,
-      "loss": 0.1817,
-      "step": 409
-    },
-    {
-      "epoch": 0.06990436103044996,
-      "grad_norm": 0.571447491645813,
-      "learning_rate": 9.300937766410914e-06,
-      "loss": 0.2015,
-      "step": 410
-    },
-    {
-      "epoch": 0.07007485947198763,
-      "grad_norm": 0.557389497756958,
-      "learning_rate": 9.29923273657289e-06,
-      "loss": 0.1314,
-      "step": 411
-    },
-    {
-      "epoch": 0.07024535791352532,
-      "grad_norm": 0.853410005569458,
-      "learning_rate": 9.29752770673487e-06,
-      "loss": 0.1809,
-      "step": 412
-    },
-    {
-      "epoch": 0.070415856355063,
-      "grad_norm": 1.0953272581100464,
-      "learning_rate": 9.295822676896846e-06,
-      "loss": 0.1144,
-      "step": 413
-    },
-    {
-      "epoch": 0.07058635479660069,
-      "grad_norm": 0.9774150848388672,
-      "learning_rate": 9.294117647058824e-06,
-      "loss": 0.1385,
-      "step": 414
-    },
-    {
-      "epoch": 0.07075685323813836,
-      "grad_norm": 0.7353179454803467,
-      "learning_rate": 9.292412617220802e-06,
-      "loss": 0.243,
-      "step": 415
-    },
-    {
-      "epoch": 0.07092735167967605,
-      "grad_norm": 0.8262443542480469,
-      "learning_rate": 9.29070758738278e-06,
-      "loss": 0.2111,
-      "step": 416
-    },
-    {
-      "epoch": 0.07109785012121374,
-      "grad_norm": 0.6665271520614624,
-      "learning_rate": 9.289002557544758e-06,
-      "loss": 0.1799,
-      "step": 417
-    },
-    {
-      "epoch": 0.07126834856275142,
-      "grad_norm": 0.6875427961349487,
-      "learning_rate": 9.287297527706736e-06,
-      "loss": 0.2445,
-      "step": 418
-    },
-    {
-      "epoch": 0.0714388470042891,
-      "grad_norm": 0.948556661605835,
-      "learning_rate": 9.285592497868713e-06,
-      "loss": 0.1675,
-      "step": 419
-    },
-    {
-      "epoch": 0.07160934544582678,
-      "grad_norm": 0.89045250415802,
-      "learning_rate": 9.283887468030691e-06,
-      "loss": 0.2529,
-      "step": 420
-    },
-    {
-      "epoch": 0.07177984388736447,
-      "grad_norm": 1.0594686269760132,
-      "learning_rate": 9.28218243819267e-06,
-      "loss": 0.1746,
-      "step": 421
-    },
-    {
-      "epoch": 0.07195034232890216,
-      "grad_norm": 0.8712635040283203,
-      "learning_rate": 9.280477408354647e-06,
-      "loss": 0.2074,
-      "step": 422
-    },
-    {
-      "epoch": 0.07212084077043983,
-      "grad_norm": 0.877993643283844,
-      "learning_rate": 9.278772378516625e-06,
-      "loss": 0.1457,
-      "step": 423
-    },
-    {
-      "epoch": 0.07229133921197752,
-      "grad_norm": 1.2433476448059082,
-      "learning_rate": 9.277067348678602e-06,
-      "loss": 0.2237,
-      "step": 424
-    },
-    {
-      "epoch": 0.0724618376535152,
-      "grad_norm": 1.1479034423828125,
-      "learning_rate": 9.275362318840581e-06,
-      "loss": 0.2184,
-      "step": 425
-    },
-    {
-      "epoch": 0.07263233609505289,
-      "grad_norm": 1.4168163537979126,
-      "learning_rate": 9.273657289002557e-06,
-      "loss": 0.1419,
-      "step": 426
-    },
-    {
-      "epoch": 0.07280283453659056,
-      "grad_norm": 0.6461614966392517,
-      "learning_rate": 9.271952259164535e-06,
-      "loss": 0.1649,
-      "step": 427
-    },
-    {
-      "epoch": 0.07297333297812825,
-      "grad_norm": 1.4802743196487427,
-      "learning_rate": 9.270247229326515e-06,
-      "loss": 0.2451,
-      "step": 428
-    },
-    {
-      "epoch": 0.07314383141966593,
-      "grad_norm": 1.2890205383300781,
-      "learning_rate": 9.268542199488491e-06,
-      "loss": 0.1541,
-      "step": 429
-    },
-    {
-      "epoch": 0.0733143298612036,
-      "grad_norm": 0.863760232925415,
-      "learning_rate": 9.26683716965047e-06,
-      "loss": 0.1404,
-      "step": 430
-    },
-    {
-      "epoch": 0.07348482830274129,
-      "grad_norm": 0.9659500122070312,
-      "learning_rate": 9.265132139812447e-06,
-      "loss": 0.2369,
-      "step": 431
-    },
-    {
-      "epoch": 0.07365532674427898,
-      "grad_norm": 1.5475353002548218,
-      "learning_rate": 9.263427109974425e-06,
-      "loss": 0.2093,
-      "step": 432
-    },
-    {
-      "epoch": 0.07382582518581667,
-      "grad_norm": 1.2611944675445557,
-      "learning_rate": 9.261722080136403e-06,
-      "loss": 0.0915,
-      "step": 433
-    },
-    {
-      "epoch": 0.07399632362735434,
-      "grad_norm": 0.6547388434410095,
-      "learning_rate": 9.260017050298381e-06,
-      "loss": 0.1208,
-      "step": 434
-    },
-    {
-      "epoch": 0.07416682206889202,
-      "grad_norm": 2.6680428981781006,
-      "learning_rate": 9.258312020460359e-06,
-      "loss": 0.1942,
-      "step": 435
-    },
-    {
-      "epoch": 0.07433732051042971,
-      "grad_norm": 0.6818041205406189,
-      "learning_rate": 9.256606990622337e-06,
-      "loss": 0.1639,
-      "step": 436
-    },
-    {
-      "epoch": 0.0745078189519674,
-      "grad_norm": 0.8517784476280212,
-      "learning_rate": 9.254901960784315e-06,
-      "loss": 0.1031,
-      "step": 437
-    },
-    {
-      "epoch": 0.07467831739350507,
-      "grad_norm": 0.6099587082862854,
-      "learning_rate": 9.253196930946293e-06,
-      "loss": 0.1316,
-      "step": 438
-    },
-    {
-      "epoch": 0.07484881583504276,
-      "grad_norm": 1.131976842880249,
-      "learning_rate": 9.25149190110827e-06,
-      "loss": 0.1404,
-      "step": 439
-    },
-    {
-      "epoch": 0.07501931427658044,
-      "grad_norm": 0.7698296308517456,
-      "learning_rate": 9.249786871270247e-06,
-      "loss": 0.1313,
-      "step": 440
-    },
-    {
-      "epoch": 0.07518981271811813,
-      "grad_norm": 1.271114468574524,
-      "learning_rate": 9.248081841432227e-06,
-      "loss": 0.1686,
-      "step": 441
-    },
-    {
-      "epoch": 0.0753603111596558,
-      "grad_norm": 0.8797255754470825,
-      "learning_rate": 9.246376811594203e-06,
-      "loss": 0.1667,
-      "step": 442
-    },
-    {
-      "epoch": 0.07553080960119349,
-      "grad_norm": 0.9120625257492065,
-      "learning_rate": 9.24467178175618e-06,
-      "loss": 0.1248,
-      "step": 443
-    },
-    {
-      "epoch": 0.07570130804273117,
-      "grad_norm": 0.8970147967338562,
-      "learning_rate": 9.24296675191816e-06,
-      "loss": 0.1601,
-      "step": 444
-    },
-    {
-      "epoch": 0.07587180648426886,
-      "grad_norm": 1.5787622928619385,
-      "learning_rate": 9.241261722080137e-06,
-      "loss": 0.1132,
-      "step": 445
-    },
-    {
-      "epoch": 0.07604230492580653,
-      "grad_norm": 0.7926347255706787,
-      "learning_rate": 9.239556692242115e-06,
-      "loss": 0.1673,
-      "step": 446
-    },
-    {
-      "epoch": 0.07621280336734422,
-      "grad_norm": 1.0942426919937134,
-      "learning_rate": 9.237851662404093e-06,
-      "loss": 0.1726,
-      "step": 447
-    },
-    {
-      "epoch": 0.0763833018088819,
-      "grad_norm": 0.7824402451515198,
-      "learning_rate": 9.23614663256607e-06,
-      "loss": 0.1421,
-      "step": 448
-    },
-    {
-      "epoch": 0.07655380025041958,
-      "grad_norm": 1.3141944408416748,
-      "learning_rate": 9.234441602728049e-06,
-      "loss": 0.2309,
-      "step": 449
-    },
-    {
-      "epoch": 0.07672429869195727,
-      "grad_norm": 0.86362624168396,
-      "learning_rate": 9.232736572890026e-06,
-      "loss": 0.1154,
-      "step": 450
-    },
-    {
-      "epoch": 0.07689479713349495,
-      "grad_norm": 0.9413621425628662,
-      "learning_rate": 9.231031543052004e-06,
-      "loss": 0.1966,
-      "step": 451
-    },
-    {
-      "epoch": 0.07706529557503264,
-      "grad_norm": 0.6306756734848022,
-      "learning_rate": 9.229326513213982e-06,
-      "loss": 0.1538,
-      "step": 452
-    },
-    {
-      "epoch": 0.07723579401657031,
-      "grad_norm": 1.079569697380066,
-      "learning_rate": 9.22762148337596e-06,
-      "loss": 0.2293,
-      "step": 453
-    },
-    {
-      "epoch": 0.077406292458108,
-      "grad_norm": 1.0969973802566528,
-      "learning_rate": 9.225916453537938e-06,
-      "loss": 0.2179,
-      "step": 454
-    },
-    {
-      "epoch": 0.07757679089964568,
-      "grad_norm": 1.7774721384048462,
-      "learning_rate": 9.224211423699916e-06,
-      "loss": 0.233,
-      "step": 455
-    },
-    {
-      "epoch": 0.07774728934118337,
-      "grad_norm": 0.6311856508255005,
-      "learning_rate": 9.222506393861892e-06,
-      "loss": 0.1904,
-      "step": 456
-    },
-    {
-      "epoch": 0.07791778778272104,
-      "grad_norm": 1.2663410902023315,
-      "learning_rate": 9.220801364023872e-06,
-      "loss": 0.0949,
-      "step": 457
-    },
-    {
-      "epoch": 0.07808828622425873,
-      "grad_norm": 1.3534469604492188,
-      "learning_rate": 9.219096334185848e-06,
-      "loss": 0.1809,
-      "step": 458
-    },
-    {
-      "epoch": 0.07825878466579642,
-      "grad_norm": 0.8532509803771973,
-      "learning_rate": 9.217391304347826e-06,
-      "loss": 0.1482,
-      "step": 459
-    },
-    {
-      "epoch": 0.0784292831073341,
-      "grad_norm": 0.9541242122650146,
-      "learning_rate": 9.215686274509804e-06,
-      "loss": 0.1872,
-      "step": 460
-    },
-    {
-      "epoch": 0.07859978154887177,
-      "grad_norm": 0.6848520040512085,
-      "learning_rate": 9.213981244671782e-06,
-      "loss": 0.1767,
-      "step": 461
-    },
-    {
-      "epoch": 0.07877027999040946,
-      "grad_norm": 0.6618411540985107,
-      "learning_rate": 9.21227621483376e-06,
-      "loss": 0.1802,
-      "step": 462
-    },
-    {
-      "epoch": 0.07894077843194715,
-      "grad_norm": 0.9855371117591858,
-      "learning_rate": 9.210571184995738e-06,
-      "loss": 0.113,
-      "step": 463
-    },
-    {
-      "epoch": 0.07911127687348483,
-      "grad_norm": 0.9223768711090088,
-      "learning_rate": 9.208866155157716e-06,
-      "loss": 0.195,
-      "step": 464
-    },
-    {
-      "epoch": 0.0792817753150225,
-      "grad_norm": 1.2770650386810303,
-      "learning_rate": 9.207161125319694e-06,
-      "loss": 0.2939,
-      "step": 465
-    },
-    {
-      "epoch": 0.0794522737565602,
-      "grad_norm": 0.7938947081565857,
-      "learning_rate": 9.205456095481672e-06,
-      "loss": 0.1819,
-      "step": 466
-    },
-    {
-      "epoch": 0.07962277219809788,
-      "grad_norm": 0.5131012797355652,
-      "learning_rate": 9.20375106564365e-06,
-      "loss": 0.1734,
-      "step": 467
-    },
-    {
-      "epoch": 0.07979327063963557,
-      "grad_norm": 0.7656362652778625,
-      "learning_rate": 9.202046035805628e-06,
-      "loss": 0.2224,
-      "step": 468
-    },
-    {
-      "epoch": 0.07996376908117324,
-      "grad_norm": 0.7925986051559448,
-      "learning_rate": 9.200341005967606e-06,
-      "loss": 0.1276,
-      "step": 469
-    },
-    {
-      "epoch": 0.08013426752271093,
-      "grad_norm": 0.4855455756187439,
-      "learning_rate": 9.198635976129584e-06,
-      "loss": 0.1497,
-      "step": 470
-    },
-    {
-      "epoch": 0.08030476596424861,
-      "grad_norm": 0.7099283337593079,
-      "learning_rate": 9.196930946291562e-06,
-      "loss": 0.2033,
-      "step": 471
-    },
-    {
-      "epoch": 0.08047526440578628,
-      "grad_norm": 0.6167730093002319,
-      "learning_rate": 9.195225916453538e-06,
-      "loss": 0.18,
-      "step": 472
-    },
-    {
-      "epoch": 0.08064576284732397,
-      "grad_norm": 0.8107618689537048,
-      "learning_rate": 9.193520886615518e-06,
-      "loss": 0.2034,
-      "step": 473
-    },
-    {
-      "epoch": 0.08081626128886166,
-      "grad_norm": 0.7333641052246094,
-      "learning_rate": 9.191815856777494e-06,
-      "loss": 0.2379,
-      "step": 474
-    },
-    {
-      "epoch": 0.08098675973039934,
-      "grad_norm": 0.7127386331558228,
-      "learning_rate": 9.190110826939472e-06,
-      "loss": 0.1248,
-      "step": 475
-    },
-    {
-      "epoch": 0.08115725817193702,
-      "grad_norm": 0.801481306552887,
-      "learning_rate": 9.18840579710145e-06,
-      "loss": 0.206,
-      "step": 476
-    },
-    {
-      "epoch": 0.0813277566134747,
-      "grad_norm": 1.2994575500488281,
-      "learning_rate": 9.186700767263428e-06,
-      "loss": 0.0752,
-      "step": 477
-    },
-    {
-      "epoch": 0.08149825505501239,
-      "grad_norm": 0.909691572189331,
-      "learning_rate": 9.184995737425406e-06,
-      "loss": 0.1901,
-      "step": 478
-    },
-    {
-      "epoch": 0.08166875349655008,
-      "grad_norm": 0.850429356098175,
-      "learning_rate": 9.183290707587384e-06,
-      "loss": 0.1715,
-      "step": 479
-    },
-    {
-      "epoch": 0.08183925193808775,
-      "grad_norm": 1.3290419578552246,
-      "learning_rate": 9.181585677749362e-06,
-      "loss": 0.267,
-      "step": 480
-    },
-    {
-      "epoch": 0.08200975037962543,
-      "grad_norm": 0.7993286848068237,
-      "learning_rate": 9.17988064791134e-06,
-      "loss": 0.125,
-      "step": 481
-    },
-    {
-      "epoch": 0.08218024882116312,
-      "grad_norm": 1.093023419380188,
-      "learning_rate": 9.178175618073317e-06,
-      "loss": 0.1835,
-      "step": 482
-    },
-    {
-      "epoch": 0.08235074726270081,
-      "grad_norm": 0.7502027750015259,
-      "learning_rate": 9.176470588235294e-06,
-      "loss": 0.146,
-      "step": 483
-    },
-    {
-      "epoch": 0.08252124570423848,
-      "grad_norm": 0.8021153807640076,
-      "learning_rate": 9.174765558397273e-06,
-      "loss": 0.2086,
-      "step": 484
-    },
-    {
-      "epoch": 0.08269174414577617,
-      "grad_norm": 0.7832606434822083,
-      "learning_rate": 9.17306052855925e-06,
-      "loss": 0.1941,
-      "step": 485
-    },
-    {
-      "epoch": 0.08286224258731385,
-      "grad_norm": 0.7455692887306213,
-      "learning_rate": 9.171355498721227e-06,
-      "loss": 0.119,
-      "step": 486
-    },
-    {
-      "epoch": 0.08303274102885154,
-      "grad_norm": 1.0250868797302246,
-      "learning_rate": 9.169650468883207e-06,
-      "loss": 0.2183,
-      "step": 487
-    },
-    {
-      "epoch": 0.08320323947038921,
-      "grad_norm": 0.5591428875923157,
-      "learning_rate": 9.167945439045183e-06,
-      "loss": 0.1261,
-      "step": 488
-    },
-    {
-      "epoch": 0.0833737379119269,
-      "grad_norm": 0.5469002723693848,
-      "learning_rate": 9.166240409207161e-06,
-      "loss": 0.1285,
-      "step": 489
-    },
-    {
-      "epoch": 0.08354423635346458,
-      "grad_norm": 0.6611772775650024,
-      "learning_rate": 9.16453537936914e-06,
-      "loss": 0.2116,
-      "step": 490
-    },
-    {
-      "epoch": 0.08371473479500227,
-      "grad_norm": 0.9039101004600525,
-      "learning_rate": 9.162830349531117e-06,
-      "loss": 0.1971,
-      "step": 491
-    },
-    {
-      "epoch": 0.08388523323653994,
-      "grad_norm": 1.6236768960952759,
-      "learning_rate": 9.161125319693095e-06,
-      "loss": 0.2645,
-      "step": 492
-    },
-    {
-      "epoch": 0.08405573167807763,
-      "grad_norm": 0.7892815470695496,
-      "learning_rate": 9.159420289855073e-06,
-      "loss": 0.1583,
-      "step": 493
-    },
-    {
-      "epoch": 0.08422623011961532,
-      "grad_norm": 0.7078331708908081,
-      "learning_rate": 9.157715260017051e-06,
-      "loss": 0.1316,
-      "step": 494
-    },
-    {
-      "epoch": 0.08439672856115299,
-      "grad_norm": 0.9892073273658752,
-      "learning_rate": 9.156010230179029e-06,
-      "loss": 0.2271,
-      "step": 495
-    },
-    {
-      "epoch": 0.08456722700269068,
-      "grad_norm": 1.0123225450515747,
-      "learning_rate": 9.154305200341007e-06,
-      "loss": 0.1299,
-      "step": 496
-    },
-    {
-      "epoch": 0.08473772544422836,
-      "grad_norm": 0.7926766276359558,
-      "learning_rate": 9.152600170502985e-06,
-      "loss": 0.1479,
-      "step": 497
-    },
-    {
-      "epoch": 0.08490822388576605,
-      "grad_norm": 1.0032105445861816,
-      "learning_rate": 9.150895140664963e-06,
-      "loss": 0.1157,
-      "step": 498
-    },
-    {
-      "epoch": 0.08507872232730372,
-      "grad_norm": 1.401563286781311,
-      "learning_rate": 9.149190110826939e-06,
-      "loss": 0.2905,
-      "step": 499
-    },
-    {
-      "epoch": 0.08524922076884141,
-      "grad_norm": 0.6407831311225891,
-      "learning_rate": 9.147485080988919e-06,
-      "loss": 0.1407,
-      "step": 500
-    },
-    {
-      "epoch": 0.08524922076884141,
-      "eval_f1_score": 0.028985507246376812,
-      "eval_loss": 0.18852059543132782,
-      "eval_runtime": 183.2722,
-      "eval_samples_per_second": 54.564,
-      "eval_steps_per_second": 3.41,
-      "step": 500
-    },
-    {
-      "epoch": 0.0854197192103791,
-      "grad_norm": 0.6139245629310608,
-      "learning_rate": 9.145780051150895e-06,
-      "loss": 0.1003,
-      "step": 501
-    },
-    {
-      "epoch": 0.08559021765191678,
-      "grad_norm": 0.7585770487785339,
-      "learning_rate": 9.144075021312873e-06,
-      "loss": 0.1214,
-      "step": 502
-    },
-    {
-      "epoch": 0.08576071609345445,
-      "grad_norm": 0.732570469379425,
-      "learning_rate": 9.142369991474853e-06,
-      "loss": 0.1182,
-      "step": 503
-    },
-    {
-      "epoch": 0.08593121453499214,
-      "grad_norm": 0.5371119379997253,
-      "learning_rate": 9.140664961636829e-06,
-      "loss": 0.1725,
-      "step": 504
-    },
-    {
-      "epoch": 0.08610171297652983,
-      "grad_norm": 0.9066203236579895,
-      "learning_rate": 9.138959931798807e-06,
-      "loss": 0.2375,
-      "step": 505
-    },
-    {
-      "epoch": 0.08627221141806751,
-      "grad_norm": 0.5497814416885376,
-      "learning_rate": 9.137254901960785e-06,
-      "loss": 0.1478,
-      "step": 506
-    },
-    {
-      "epoch": 0.08644270985960519,
-      "grad_norm": 1.1947273015975952,
-      "learning_rate": 9.135549872122763e-06,
-      "loss": 0.1718,
-      "step": 507
-    },
-    {
-      "epoch": 0.08661320830114287,
-      "grad_norm": 0.8404238224029541,
-      "learning_rate": 9.13384484228474e-06,
-      "loss": 0.1625,
-      "step": 508
-    },
-    {
-      "epoch": 0.08678370674268056,
-      "grad_norm": 1.0687267780303955,
-      "learning_rate": 9.132139812446719e-06,
-      "loss": 0.2658,
-      "step": 509
-    },
-    {
-      "epoch": 0.08695420518421824,
-      "grad_norm": 0.7937365174293518,
-      "learning_rate": 9.130434782608697e-06,
-      "loss": 0.1944,
-      "step": 510
-    },
-    {
-      "epoch": 0.08712470362575592,
-      "grad_norm": 4.469568252563477,
-      "learning_rate": 9.128729752770674e-06,
-      "loss": 0.1463,
-      "step": 511
-    },
-    {
-      "epoch": 0.0872952020672936,
-      "grad_norm": 1.0447407960891724,
-      "learning_rate": 9.127024722932652e-06,
-      "loss": 0.2103,
-      "step": 512
-    },
-    {
-      "epoch": 0.08746570050883129,
-      "grad_norm": 1.9300107955932617,
-      "learning_rate": 9.12531969309463e-06,
-      "loss": 0.1259,
-      "step": 513
-    },
-    {
-      "epoch": 0.08763619895036896,
-      "grad_norm": 1.7074291706085205,
-      "learning_rate": 9.123614663256608e-06,
-      "loss": 0.1118,
-      "step": 514
-    },
-    {
-      "epoch": 0.08780669739190665,
-      "grad_norm": 1.4637730121612549,
-      "learning_rate": 9.121909633418585e-06,
-      "loss": 0.1045,
-      "step": 515
-    },
-    {
-      "epoch": 0.08797719583344434,
-      "grad_norm": 0.7066337466239929,
-      "learning_rate": 9.120204603580564e-06,
-      "loss": 0.1469,
-      "step": 516
-    },
-    {
-      "epoch": 0.08814769427498202,
-      "grad_norm": 1.2253496646881104,
-      "learning_rate": 9.11849957374254e-06,
-      "loss": 0.2541,
-      "step": 517
-    },
-    {
-      "epoch": 0.0883181927165197,
-      "grad_norm": 0.9337403774261475,
-      "learning_rate": 9.116794543904518e-06,
-      "loss": 0.1464,
-      "step": 518
-    },
-    {
-      "epoch": 0.08848869115805738,
-      "grad_norm": 1.1742297410964966,
-      "learning_rate": 9.115089514066498e-06,
-      "loss": 0.2045,
-      "step": 519
-    },
-    {
-      "epoch": 0.08865918959959507,
-      "grad_norm": 0.9703015089035034,
-      "learning_rate": 9.113384484228474e-06,
-      "loss": 0.0902,
-      "step": 520
-    },
-    {
-      "epoch": 0.08882968804113275,
-      "grad_norm": 0.9036072492599487,
-      "learning_rate": 9.111679454390452e-06,
-      "loss": 0.2117,
-      "step": 521
-    },
-    {
-      "epoch": 0.08900018648267043,
-      "grad_norm": 1.4741570949554443,
-      "learning_rate": 9.10997442455243e-06,
-      "loss": 0.2575,
-      "step": 522
-    },
-    {
-      "epoch": 0.08917068492420811,
-      "grad_norm": 0.7809064388275146,
-      "learning_rate": 9.108269394714408e-06,
-      "loss": 0.1914,
-      "step": 523
-    },
-    {
-      "epoch": 0.0893411833657458,
-      "grad_norm": 0.5947693586349487,
-      "learning_rate": 9.106564364876386e-06,
-      "loss": 0.1615,
-      "step": 524
-    },
-    {
-      "epoch": 0.08951168180728349,
-      "grad_norm": 0.701625406742096,
-      "learning_rate": 9.104859335038364e-06,
-      "loss": 0.1318,
-      "step": 525
-    },
-    {
-      "epoch": 0.08968218024882116,
-      "grad_norm": 0.60597825050354,
-      "learning_rate": 9.103154305200342e-06,
-      "loss": 0.1605,
-      "step": 526
-    },
-    {
-      "epoch": 0.08985267869035884,
-      "grad_norm": 0.8819074034690857,
-      "learning_rate": 9.10144927536232e-06,
-      "loss": 0.2386,
-      "step": 527
-    },
-    {
-      "epoch": 0.09002317713189653,
-      "grad_norm": 0.7760066390037537,
-      "learning_rate": 9.099744245524298e-06,
-      "loss": 0.2037,
-      "step": 528
-    },
-    {
-      "epoch": 0.09019367557343422,
-      "grad_norm": 0.8319275379180908,
-      "learning_rate": 9.098039215686276e-06,
-      "loss": 0.1922,
-      "step": 529
-    },
-    {
-      "epoch": 0.09036417401497189,
-      "grad_norm": 1.5015146732330322,
-      "learning_rate": 9.096334185848254e-06,
-      "loss": 0.2665,
-      "step": 530
-    },
-    {
-      "epoch": 0.09053467245650958,
-      "grad_norm": 1.2790735960006714,
-      "learning_rate": 9.09462915601023e-06,
-      "loss": 0.1451,
-      "step": 531
-    },
-    {
-      "epoch": 0.09070517089804726,
-      "grad_norm": 0.9179703593254089,
-      "learning_rate": 9.09292412617221e-06,
-      "loss": 0.1774,
-      "step": 532
-    },
-    {
-      "epoch": 0.09087566933958495,
-      "grad_norm": 0.9182787537574768,
-      "learning_rate": 9.091219096334186e-06,
-      "loss": 0.1703,
-      "step": 533
-    },
-    {
-      "epoch": 0.09104616778112262,
-      "grad_norm": 0.5663113594055176,
-      "learning_rate": 9.089514066496164e-06,
-      "loss": 0.1502,
-      "step": 534
-    },
-    {
-      "epoch": 0.09121666622266031,
-      "grad_norm": 0.6369578838348389,
-      "learning_rate": 9.087809036658142e-06,
-      "loss": 0.1443,
-      "step": 535
-    },
-    {
-      "epoch": 0.091387164664198,
-      "grad_norm": 0.6523211598396301,
-      "learning_rate": 9.08610400682012e-06,
-      "loss": 0.1666,
-      "step": 536
-    },
-    {
-      "epoch": 0.09155766310573567,
-      "grad_norm": 0.7603857517242432,
-      "learning_rate": 9.084398976982098e-06,
-      "loss": 0.086,
-      "step": 537
-    },
-    {
-      "epoch": 0.09172816154727335,
-      "grad_norm": 1.1395196914672852,
-      "learning_rate": 9.082693947144076e-06,
-      "loss": 0.2465,
-      "step": 538
-    },
-    {
-      "epoch": 0.09189865998881104,
-      "grad_norm": 0.682396650314331,
-      "learning_rate": 9.080988917306054e-06,
-      "loss": 0.0308,
-      "step": 539
-    },
-    {
-      "epoch": 0.09206915843034873,
-      "grad_norm": 1.1837222576141357,
-      "learning_rate": 9.079283887468032e-06,
-      "loss": 0.2423,
-      "step": 540
-    },
-    {
-      "epoch": 0.0922396568718864,
-      "grad_norm": 1.1183197498321533,
-      "learning_rate": 9.07757885763001e-06,
-      "loss": 0.2646,
-      "step": 541
-    },
-    {
-      "epoch": 0.09241015531342409,
-      "grad_norm": 0.6793837547302246,
-      "learning_rate": 9.075873827791987e-06,
-      "loss": 0.1761,
-      "step": 542
-    },
-    {
-      "epoch": 0.09258065375496177,
-      "grad_norm": 0.3091171681880951,
-      "learning_rate": 9.074168797953965e-06,
-      "loss": 0.1189,
-      "step": 543
-    },
-    {
-      "epoch": 0.09275115219649946,
-      "grad_norm": 0.7060979008674622,
-      "learning_rate": 9.072463768115943e-06,
-      "loss": 0.1776,
-      "step": 544
-    },
-    {
-      "epoch": 0.09292165063803713,
-      "grad_norm": 0.3717476427555084,
-      "learning_rate": 9.070758738277921e-06,
-      "loss": 0.1104,
-      "step": 545
-    },
-    {
-      "epoch": 0.09309214907957482,
-      "grad_norm": 0.44111332297325134,
-      "learning_rate": 9.0690537084399e-06,
-      "loss": 0.1127,
-      "step": 546
-    },
-    {
-      "epoch": 0.0932626475211125,
-      "grad_norm": 1.1776901483535767,
-      "learning_rate": 9.067348678601876e-06,
-      "loss": 0.2734,
-      "step": 547
-    },
-    {
-      "epoch": 0.09343314596265019,
-      "grad_norm": 0.5457471609115601,
-      "learning_rate": 9.065643648763855e-06,
-      "loss": 0.1419,
-      "step": 548
-    },
-    {
-      "epoch": 0.09360364440418786,
-      "grad_norm": 0.584895670413971,
-      "learning_rate": 9.063938618925831e-06,
-      "loss": 0.1478,
-      "step": 549
-    },
-    {
-      "epoch": 0.09377414284572555,
-      "grad_norm": 1.4400527477264404,
-      "learning_rate": 9.06223358908781e-06,
-      "loss": 0.1638,
-      "step": 550
-    },
-    {
-      "epoch": 0.09394464128726324,
-      "grad_norm": 0.7518956661224365,
-      "learning_rate": 9.060528559249787e-06,
-      "loss": 0.1243,
-      "step": 551
-    },
-    {
-      "epoch": 0.09411513972880092,
-      "grad_norm": 0.5842875838279724,
-      "learning_rate": 9.058823529411765e-06,
-      "loss": 0.2061,
-      "step": 552
-    },
-    {
-      "epoch": 0.0942856381703386,
-      "grad_norm": 0.8577499985694885,
-      "learning_rate": 9.057118499573743e-06,
-      "loss": 0.25,
-      "step": 553
-    },
-    {
-      "epoch": 0.09445613661187628,
-      "grad_norm": 0.5927447080612183,
-      "learning_rate": 9.055413469735721e-06,
-      "loss": 0.168,
-      "step": 554
-    },
-    {
-      "epoch": 0.09462663505341397,
-      "grad_norm": 1.0270954370498657,
-      "learning_rate": 9.053708439897699e-06,
-      "loss": 0.251,
-      "step": 555
-    },
-    {
-      "epoch": 0.09479713349495164,
-      "grad_norm": 0.89007967710495,
-      "learning_rate": 9.052003410059677e-06,
-      "loss": 0.1797,
-      "step": 556
-    },
-    {
-      "epoch": 0.09496763193648933,
-      "grad_norm": 0.8992642760276794,
-      "learning_rate": 9.050298380221655e-06,
-      "loss": 0.15,
-      "step": 557
-    },
-    {
-      "epoch": 0.09513813037802701,
-      "grad_norm": 0.7758667469024658,
-      "learning_rate": 9.048593350383631e-06,
-      "loss": 0.1533,
-      "step": 558
-    },
-    {
-      "epoch": 0.0953086288195647,
-      "grad_norm": 0.9712894558906555,
-      "learning_rate": 9.046888320545611e-06,
-      "loss": 0.139,
-      "step": 559
-    },
-    {
-      "epoch": 0.09547912726110237,
-      "grad_norm": 0.6283227205276489,
-      "learning_rate": 9.045183290707587e-06,
-      "loss": 0.131,
-      "step": 560
-    },
-    {
-      "epoch": 0.09564962570264006,
-      "grad_norm": 0.8274819254875183,
-      "learning_rate": 9.043478260869565e-06,
-      "loss": 0.1464,
-      "step": 561
-    },
-    {
-      "epoch": 0.09582012414417775,
-      "grad_norm": 0.7347449064254761,
-      "learning_rate": 9.041773231031545e-06,
-      "loss": 0.1941,
-      "step": 562
-    },
-    {
-      "epoch": 0.09599062258571543,
-      "grad_norm": 1.8517053127288818,
-      "learning_rate": 9.040068201193521e-06,
-      "loss": 0.2219,
-      "step": 563
-    },
-    {
-      "epoch": 0.0961611210272531,
-      "grad_norm": 0.6514312624931335,
-      "learning_rate": 9.038363171355499e-06,
-      "loss": 0.1564,
-      "step": 564
-    },
-    {
-      "epoch": 0.09633161946879079,
-      "grad_norm": 1.3228540420532227,
-      "learning_rate": 9.036658141517477e-06,
-      "loss": 0.2409,
-      "step": 565
-    },
-    {
-      "epoch": 0.09650211791032848,
-      "grad_norm": 1.0575270652770996,
-      "learning_rate": 9.034953111679455e-06,
-      "loss": 0.2279,
-      "step": 566
-    },
-    {
-      "epoch": 0.09667261635186616,
-      "grad_norm": 0.650841474533081,
-      "learning_rate": 9.033248081841433e-06,
-      "loss": 0.131,
-      "step": 567
-    },
-    {
-      "epoch": 0.09684311479340384,
-      "grad_norm": 0.8555294871330261,
-      "learning_rate": 9.03154305200341e-06,
-      "loss": 0.1119,
-      "step": 568
-    },
-    {
-      "epoch": 0.09701361323494152,
-      "grad_norm": 0.6819137930870056,
-      "learning_rate": 9.029838022165389e-06,
-      "loss": 0.1066,
-      "step": 569
-    },
-    {
-      "epoch": 0.09718411167647921,
-      "grad_norm": 0.65989750623703,
-      "learning_rate": 9.028132992327367e-06,
-      "loss": 0.1784,
-      "step": 570
-    },
-    {
-      "epoch": 0.0973546101180169,
-      "grad_norm": 0.7284032106399536,
-      "learning_rate": 9.026427962489345e-06,
-      "loss": 0.1243,
-      "step": 571
-    },
-    {
-      "epoch": 0.09752510855955457,
-      "grad_norm": 0.9932384490966797,
-      "learning_rate": 9.024722932651323e-06,
-      "loss": 0.1838,
-      "step": 572
-    },
-    {
-      "epoch": 0.09769560700109226,
-      "grad_norm": 0.9998315572738647,
-      "learning_rate": 9.0230179028133e-06,
-      "loss": 0.2078,
-      "step": 573
-    },
-    {
-      "epoch": 0.09786610544262994,
-      "grad_norm": 0.6617645025253296,
-      "learning_rate": 9.021312872975277e-06,
-      "loss": 0.1751,
-      "step": 574
-    },
-    {
-      "epoch": 0.09803660388416763,
-      "grad_norm": 0.6724914312362671,
-      "learning_rate": 9.019607843137256e-06,
-      "loss": 0.1654,
-      "step": 575
-    },
-    {
-      "epoch": 0.0982071023257053,
-      "grad_norm": 0.7818968296051025,
-      "learning_rate": 9.017902813299233e-06,
-      "loss": 0.1118,
-      "step": 576
-    },
-    {
-      "epoch": 0.09837760076724299,
-      "grad_norm": 1.0563021898269653,
-      "learning_rate": 9.01619778346121e-06,
-      "loss": 0.176,
-      "step": 577
-    },
-    {
-      "epoch": 0.09854809920878067,
-      "grad_norm": 0.7812329530715942,
-      "learning_rate": 9.01449275362319e-06,
-      "loss": 0.1757,
-      "step": 578
-    },
-    {
-      "epoch": 0.09871859765031835,
-      "grad_norm": 0.6268443465232849,
-      "learning_rate": 9.012787723785166e-06,
-      "loss": 0.1592,
-      "step": 579
-    },
-    {
-      "epoch": 0.09888909609185603,
-      "grad_norm": 0.744136393070221,
-      "learning_rate": 9.011082693947144e-06,
-      "loss": 0.2224,
-      "step": 580
-    },
-    {
-      "epoch": 0.09905959453339372,
-      "grad_norm": 0.7699022889137268,
-      "learning_rate": 9.009377664109122e-06,
-      "loss": 0.1461,
-      "step": 581
-    },
-    {
-      "epoch": 0.0992300929749314,
-      "grad_norm": 1.6789628267288208,
-      "learning_rate": 9.0076726342711e-06,
-      "loss": 0.2659,
-      "step": 582
-    },
-    {
-      "epoch": 0.09940059141646908,
-      "grad_norm": 0.9754775762557983,
-      "learning_rate": 9.005967604433078e-06,
-      "loss": 0.0963,
-      "step": 583
-    },
-    {
-      "epoch": 0.09957108985800676,
-      "grad_norm": 0.6113495826721191,
-      "learning_rate": 9.004262574595056e-06,
-      "loss": 0.1764,
-      "step": 584
-    },
-    {
-      "epoch": 0.09974158829954445,
-      "grad_norm": 1.0460819005966187,
-      "learning_rate": 9.002557544757034e-06,
-      "loss": 0.2345,
-      "step": 585
-    },
-    {
-      "epoch": 0.09991208674108214,
-      "grad_norm": 0.9724820256233215,
-      "learning_rate": 9.000852514919012e-06,
-      "loss": 0.1813,
-      "step": 586
-    },
-    {
-      "epoch": 0.10008258518261981,
-      "grad_norm": 0.9336734414100647,
-      "learning_rate": 8.99914748508099e-06,
-      "loss": 0.2385,
-      "step": 587
-    },
-    {
-      "epoch": 0.1002530836241575,
-      "grad_norm": 1.145740270614624,
-      "learning_rate": 8.997442455242968e-06,
-      "loss": 0.1205,
-      "step": 588
-    },
-    {
-      "epoch": 0.10042358206569518,
-      "grad_norm": 1.0796016454696655,
-      "learning_rate": 8.995737425404946e-06,
-      "loss": 0.1873,
-      "step": 589
-    },
-    {
-      "epoch": 0.10059408050723287,
-      "grad_norm": 0.8692519068717957,
-      "learning_rate": 8.994032395566922e-06,
-      "loss": 0.1789,
-      "step": 590
-    },
-    {
-      "epoch": 0.10076457894877054,
-      "grad_norm": 0.9212008118629456,
-      "learning_rate": 8.992327365728902e-06,
-      "loss": 0.2034,
-      "step": 591
-    },
-    {
-      "epoch": 0.10093507739030823,
-      "grad_norm": 0.8814923763275146,
-      "learning_rate": 8.990622335890878e-06,
-      "loss": 0.1145,
-      "step": 592
-    },
-    {
-      "epoch": 0.10110557583184591,
-      "grad_norm": 0.8948725461959839,
-      "learning_rate": 8.988917306052856e-06,
-      "loss": 0.1507,
-      "step": 593
-    },
-    {
-      "epoch": 0.1012760742733836,
-      "grad_norm": 1.134162425994873,
-      "learning_rate": 8.987212276214834e-06,
-      "loss": 0.1286,
-      "step": 594
-    },
-    {
-      "epoch": 0.10144657271492127,
-      "grad_norm": 0.455159068107605,
-      "learning_rate": 8.985507246376812e-06,
-      "loss": 0.1394,
-      "step": 595
-    },
-    {
-      "epoch": 0.10161707115645896,
-      "grad_norm": 1.6202359199523926,
-      "learning_rate": 8.98380221653879e-06,
-      "loss": 0.3456,
-      "step": 596
-    },
-    {
-      "epoch": 0.10178756959799665,
-      "grad_norm": 1.1022000312805176,
-      "learning_rate": 8.982097186700768e-06,
-      "loss": 0.1169,
-      "step": 597
-    },
-    {
-      "epoch": 0.10195806803953433,
-      "grad_norm": 0.6847971677780151,
-      "learning_rate": 8.980392156862746e-06,
-      "loss": 0.0918,
-      "step": 598
-    },
-    {
-      "epoch": 0.102128566481072,
-      "grad_norm": 1.6231807470321655,
-      "learning_rate": 8.978687127024724e-06,
-      "loss": 0.2878,
-      "step": 599
-    },
-    {
-      "epoch": 0.10229906492260969,
-      "grad_norm": 0.6163637042045593,
-      "learning_rate": 8.976982097186702e-06,
-      "loss": 0.1903,
-      "step": 600
-    },
-    {
-      "epoch": 0.10229906492260969,
-      "eval_f1_score": 0.0,
-      "eval_loss": 0.19083337485790253,
-      "eval_runtime": 183.2063,
-      "eval_samples_per_second": 54.583,
-      "eval_steps_per_second": 3.411,
-      "step": 600
-    },
-    {
-      "epoch": 0.10246956336414738,
-      "grad_norm": 0.401183158159256,
-      "learning_rate": 8.97527706734868e-06,
-      "loss": 0.1146,
-      "step": 601
-    },
-    {
-      "epoch": 0.10264006180568505,
-      "grad_norm": 0.8258224725723267,
-      "learning_rate": 8.973572037510658e-06,
-      "loss": 0.2058,
-      "step": 602
-    },
-    {
-      "epoch": 0.10281056024722274,
-      "grad_norm": 0.7540138363838196,
-      "learning_rate": 8.971867007672636e-06,
-      "loss": 0.1825,
-      "step": 603
-    },
-    {
-      "epoch": 0.10298105868876042,
-      "grad_norm": 0.6636288166046143,
-      "learning_rate": 8.970161977834613e-06,
-      "loss": 0.1508,
-      "step": 604
-    },
-    {
-      "epoch": 0.10315155713029811,
-      "grad_norm": 0.6270397305488586,
-      "learning_rate": 8.968456947996591e-06,
-      "loss": 0.1327,
-      "step": 605
-    },
-    {
-      "epoch": 0.10332205557183578,
-      "grad_norm": 0.7056589722633362,
-      "learning_rate": 8.966751918158568e-06,
-      "loss": 0.1897,
-      "step": 606
-    },
-    {
-      "epoch": 0.10349255401337347,
-      "grad_norm": 0.622838020324707,
-      "learning_rate": 8.965046888320547e-06,
-      "loss": 0.1206,
-      "step": 607
-    },
-    {
-      "epoch": 0.10366305245491116,
-      "grad_norm": 0.6201973557472229,
-      "learning_rate": 8.963341858482524e-06,
-      "loss": 0.141,
-      "step": 608
-    },
-    {
-      "epoch": 0.10383355089644884,
-      "grad_norm": 0.6052491068840027,
-      "learning_rate": 8.961636828644502e-06,
-      "loss": 0.1471,
-      "step": 609
-    },
-    {
-      "epoch": 0.10400404933798651,
-      "grad_norm": 0.8596045970916748,
-      "learning_rate": 8.95993179880648e-06,
-      "loss": 0.1085,
-      "step": 610
-    },
-    {
-      "epoch": 0.1041745477795242,
-      "grad_norm": 1.2298681735992432,
-      "learning_rate": 8.958226768968457e-06,
-      "loss": 0.1583,
-      "step": 611
-    },
-    {
-      "epoch": 0.10434504622106189,
-      "grad_norm": 0.83328777551651,
-      "learning_rate": 8.956521739130435e-06,
-      "loss": 0.227,
-      "step": 612
-    },
-    {
-      "epoch": 0.10451554466259957,
-      "grad_norm": 1.138857364654541,
-      "learning_rate": 8.954816709292413e-06,
-      "loss": 0.1638,
-      "step": 613
-    },
-    {
-      "epoch": 0.10468604310413725,
-      "grad_norm": 0.9594923257827759,
-      "learning_rate": 8.953111679454391e-06,
-      "loss": 0.2186,
-      "step": 614
-    },
-    {
-      "epoch": 0.10485654154567493,
-      "grad_norm": 1.0194865465164185,
-      "learning_rate": 8.95140664961637e-06,
-      "loss": 0.1977,
-      "step": 615
-    },
-    {
-      "epoch": 0.10502703998721262,
-      "grad_norm": 0.7128332853317261,
-      "learning_rate": 8.949701619778347e-06,
-      "loss": 0.1246,
-      "step": 616
-    },
-    {
-      "epoch": 0.1051975384287503,
-      "grad_norm": 0.6208592653274536,
-      "learning_rate": 8.947996589940325e-06,
-      "loss": 0.1104,
-      "step": 617
-    },
-    {
-      "epoch": 0.10536803687028798,
-      "grad_norm": 0.8177977204322815,
-      "learning_rate": 8.946291560102303e-06,
-      "loss": 0.1107,
-      "step": 618
-    },
-    {
-      "epoch": 0.10553853531182567,
-      "grad_norm": 0.9365327954292297,
-      "learning_rate": 8.94458653026428e-06,
-      "loss": 0.2365,
-      "step": 619
-    },
-    {
-      "epoch": 0.10570903375336335,
-      "grad_norm": 1.0838879346847534,
-      "learning_rate": 8.942881500426259e-06,
-      "loss": 0.2582,
-      "step": 620
-    },
-    {
-      "epoch": 0.10587953219490102,
-      "grad_norm": 0.9658844470977783,
-      "learning_rate": 8.941176470588237e-06,
-      "loss": 0.0917,
-      "step": 621
-    },
-    {
-      "epoch": 0.10605003063643871,
-      "grad_norm": 1.2021678686141968,
-      "learning_rate": 8.939471440750213e-06,
-      "loss": 0.2124,
-      "step": 622
-    },
-    {
-      "epoch": 0.1062205290779764,
-      "grad_norm": 0.6625456809997559,
-      "learning_rate": 8.937766410912193e-06,
-      "loss": 0.19,
-      "step": 623
-    },
-    {
-      "epoch": 0.10639102751951408,
-      "grad_norm": 0.8603885173797607,
-      "learning_rate": 8.936061381074169e-06,
-      "loss": 0.2054,
-      "step": 624
-    },
-    {
-      "epoch": 0.10656152596105176,
-      "grad_norm": 0.9610936045646667,
-      "learning_rate": 8.934356351236147e-06,
-      "loss": 0.0493,
-      "step": 625
-    },
-    {
-      "epoch": 0.10673202440258944,
-      "grad_norm": 0.7804873585700989,
-      "learning_rate": 8.932651321398125e-06,
-      "loss": 0.1209,
-      "step": 626
-    },
-    {
-      "epoch": 0.10690252284412713,
-      "grad_norm": 0.6967254877090454,
-      "learning_rate": 8.930946291560103e-06,
-      "loss": 0.1907,
-      "step": 627
-    },
-    {
-      "epoch": 0.10707302128566482,
-      "grad_norm": 0.6492006182670593,
-      "learning_rate": 8.92924126172208e-06,
-      "loss": 0.1195,
-      "step": 628
-    },
-    {
-      "epoch": 0.10724351972720249,
-      "grad_norm": 0.7812399864196777,
-      "learning_rate": 8.927536231884059e-06,
-      "loss": 0.0563,
-      "step": 629
-    },
-    {
-      "epoch": 0.10741401816874017,
-      "grad_norm": 0.5185630321502686,
-      "learning_rate": 8.925831202046037e-06,
-      "loss": 0.1077,
-      "step": 630
-    },
-    {
-      "epoch": 0.10758451661027786,
-      "grad_norm": 0.4689163267612457,
-      "learning_rate": 8.924126172208015e-06,
-      "loss": 0.1661,
-      "step": 631
-    },
-    {
-      "epoch": 0.10775501505181555,
-      "grad_norm": 1.0575098991394043,
-      "learning_rate": 8.922421142369993e-06,
-      "loss": 0.2516,
-      "step": 632
-    },
-    {
-      "epoch": 0.10792551349335322,
-      "grad_norm": 0.9641260504722595,
-      "learning_rate": 8.920716112531969e-06,
-      "loss": 0.1721,
-      "step": 633
-    },
-    {
-      "epoch": 0.1080960119348909,
-      "grad_norm": 0.4975630044937134,
-      "learning_rate": 8.919011082693949e-06,
-      "loss": 0.1141,
-      "step": 634
-    },
-    {
-      "epoch": 0.10826651037642859,
-      "grad_norm": 1.1529088020324707,
-      "learning_rate": 8.917306052855925e-06,
-      "loss": 0.1385,
-      "step": 635
-    },
-    {
-      "epoch": 0.10843700881796628,
-      "grad_norm": 0.9603254795074463,
-      "learning_rate": 8.915601023017903e-06,
-      "loss": 0.1957,
-      "step": 636
-    },
-    {
-      "epoch": 0.10860750725950395,
-      "grad_norm": 1.3553906679153442,
-      "learning_rate": 8.913895993179882e-06,
-      "loss": 0.2296,
-      "step": 637
-    },
-    {
-      "epoch": 0.10877800570104164,
-      "grad_norm": 0.6116076707839966,
-      "learning_rate": 8.912190963341859e-06,
-      "loss": 0.0853,
-      "step": 638
-    },
-    {
-      "epoch": 0.10894850414257932,
-      "grad_norm": 0.7998124957084656,
-      "learning_rate": 8.910485933503837e-06,
-      "loss": 0.1961,
-      "step": 639
-    },
-    {
-      "epoch": 0.10911900258411701,
-      "grad_norm": 0.6033914089202881,
-      "learning_rate": 8.908780903665815e-06,
-      "loss": 0.1019,
-      "step": 640
-    },
-    {
-      "epoch": 0.10928950102565468,
-      "grad_norm": 0.8142050504684448,
-      "learning_rate": 8.907075873827792e-06,
-      "loss": 0.0954,
-      "step": 641
-    },
-    {
-      "epoch": 0.10945999946719237,
-      "grad_norm": 1.1877548694610596,
-      "learning_rate": 8.90537084398977e-06,
-      "loss": 0.1673,
-      "step": 642
-    },
-    {
-      "epoch": 0.10963049790873006,
-      "grad_norm": 1.1034671068191528,
-      "learning_rate": 8.903665814151748e-06,
-      "loss": 0.1147,
-      "step": 643
-    },
-    {
-      "epoch": 0.10980099635026773,
-      "grad_norm": 0.755710780620575,
-      "learning_rate": 8.901960784313726e-06,
-      "loss": 0.1529,
-      "step": 644
-    },
-    {
-      "epoch": 0.10997149479180542,
-      "grad_norm": 0.6754464507102966,
-      "learning_rate": 8.900255754475704e-06,
-      "loss": 0.1961,
-      "step": 645
-    },
-    {
-      "epoch": 0.1101419932333431,
-      "grad_norm": 0.7231892943382263,
-      "learning_rate": 8.898550724637682e-06,
-      "loss": 0.1064,
-      "step": 646
-    },
-    {
-      "epoch": 0.11031249167488079,
-      "grad_norm": 1.5384999513626099,
-      "learning_rate": 8.89684569479966e-06,
-      "loss": 0.1895,
-      "step": 647
-    },
-    {
-      "epoch": 0.11048299011641846,
-      "grad_norm": 0.9464960694313049,
-      "learning_rate": 8.895140664961638e-06,
-      "loss": 0.1706,
-      "step": 648
-    },
-    {
-      "epoch": 0.11065348855795615,
-      "grad_norm": 1.2258471250534058,
-      "learning_rate": 8.893435635123614e-06,
-      "loss": 0.2841,
-      "step": 649
-    },
-    {
-      "epoch": 0.11082398699949383,
-      "grad_norm": 0.6537081003189087,
-      "learning_rate": 8.891730605285594e-06,
-      "loss": 0.1588,
-      "step": 650
-    },
-    {
-      "epoch": 0.11099448544103152,
-      "grad_norm": 0.8599593639373779,
-      "learning_rate": 8.89002557544757e-06,
-      "loss": 0.2128,
-      "step": 651
-    },
-    {
-      "epoch": 0.1111649838825692,
-      "grad_norm": 1.066692590713501,
-      "learning_rate": 8.888320545609548e-06,
-      "loss": 0.1172,
-      "step": 652
-    },
-    {
-      "epoch": 0.11133548232410688,
-      "grad_norm": 1.1547163724899292,
-      "learning_rate": 8.886615515771528e-06,
-      "loss": 0.0957,
-      "step": 653
-    },
-    {
-      "epoch": 0.11150598076564457,
-      "grad_norm": 0.9395566582679749,
-      "learning_rate": 8.884910485933504e-06,
-      "loss": 0.1694,
-      "step": 654
-    },
-    {
-      "epoch": 0.11167647920718225,
-      "grad_norm": 0.8956859707832336,
-      "learning_rate": 8.883205456095482e-06,
-      "loss": 0.2008,
-      "step": 655
-    },
-    {
-      "epoch": 0.11184697764871993,
-      "grad_norm": 2.565577507019043,
-      "learning_rate": 8.88150042625746e-06,
-      "loss": 0.159,
-      "step": 656
-    },
-    {
-      "epoch": 0.11201747609025761,
-      "grad_norm": 0.733633279800415,
-      "learning_rate": 8.879795396419438e-06,
-      "loss": 0.1392,
-      "step": 657
-    },
-    {
-      "epoch": 0.1121879745317953,
-      "grad_norm": 1.230839490890503,
-      "learning_rate": 8.878090366581416e-06,
-      "loss": 0.1753,
-      "step": 658
-    },
-    {
-      "epoch": 0.11235847297333298,
-      "grad_norm": 0.8218492269515991,
-      "learning_rate": 8.876385336743394e-06,
-      "loss": 0.1408,
-      "step": 659
-    },
-    {
-      "epoch": 0.11252897141487066,
-      "grad_norm": 0.7273249626159668,
-      "learning_rate": 8.874680306905372e-06,
-      "loss": 0.097,
-      "step": 660
-    },
-    {
-      "epoch": 0.11269946985640834,
-      "grad_norm": 0.7804353833198547,
-      "learning_rate": 8.87297527706735e-06,
-      "loss": 0.1028,
-      "step": 661
-    },
-    {
-      "epoch": 0.11286996829794603,
-      "grad_norm": 1.1182870864868164,
-      "learning_rate": 8.871270247229328e-06,
-      "loss": 0.1596,
-      "step": 662
-    },
-    {
-      "epoch": 0.1130404667394837,
-      "grad_norm": 0.745988130569458,
-      "learning_rate": 8.869565217391306e-06,
-      "loss": 0.1094,
-      "step": 663
-    },
-    {
-      "epoch": 0.11321096518102139,
-      "grad_norm": 0.8780370354652405,
-      "learning_rate": 8.867860187553284e-06,
-      "loss": 0.1809,
-      "step": 664
-    },
-    {
-      "epoch": 0.11338146362255908,
-      "grad_norm": 1.2652925252914429,
-      "learning_rate": 8.86615515771526e-06,
-      "loss": 0.2359,
-      "step": 665
-    },
-    {
-      "epoch": 0.11355196206409676,
-      "grad_norm": 0.6840066313743591,
-      "learning_rate": 8.86445012787724e-06,
-      "loss": 0.1784,
-      "step": 666
-    },
-    {
-      "epoch": 0.11372246050563443,
-      "grad_norm": 0.7085036039352417,
-      "learning_rate": 8.862745098039216e-06,
-      "loss": 0.1476,
-      "step": 667
-    },
-    {
-      "epoch": 0.11389295894717212,
-      "grad_norm": 0.7280018925666809,
-      "learning_rate": 8.861040068201194e-06,
-      "loss": 0.1467,
-      "step": 668
-    },
-    {
-      "epoch": 0.11406345738870981,
-      "grad_norm": 0.8055374026298523,
-      "learning_rate": 8.859335038363172e-06,
-      "loss": 0.1618,
-      "step": 669
-    },
-    {
-      "epoch": 0.1142339558302475,
-      "grad_norm": 0.7427027225494385,
-      "learning_rate": 8.85763000852515e-06,
-      "loss": 0.1322,
-      "step": 670
-    },
-    {
-      "epoch": 0.11440445427178517,
-      "grad_norm": 1.0781701803207397,
-      "learning_rate": 8.855924978687127e-06,
-      "loss": 0.1994,
-      "step": 671
-    },
-    {
-      "epoch": 0.11457495271332285,
-      "grad_norm": 0.5790367126464844,
-      "learning_rate": 8.854219948849105e-06,
-      "loss": 0.1724,
-      "step": 672
-    },
-    {
-      "epoch": 0.11474545115486054,
-      "grad_norm": 1.1667585372924805,
-      "learning_rate": 8.852514919011083e-06,
-      "loss": 0.1996,
-      "step": 673
-    },
-    {
-      "epoch": 0.11491594959639823,
-      "grad_norm": 1.069802165031433,
-      "learning_rate": 8.850809889173061e-06,
-      "loss": 0.2446,
-      "step": 674
-    },
-    {
-      "epoch": 0.1150864480379359,
-      "grad_norm": 0.8994653820991516,
-      "learning_rate": 8.84910485933504e-06,
-      "loss": 0.1097,
-      "step": 675
-    },
-    {
-      "epoch": 0.11525694647947358,
-      "grad_norm": 0.7296671271324158,
-      "learning_rate": 8.847399829497017e-06,
-      "loss": 0.1465,
-      "step": 676
-    },
-    {
-      "epoch": 0.11542744492101127,
-      "grad_norm": 0.7813960909843445,
-      "learning_rate": 8.845694799658995e-06,
-      "loss": 0.1931,
-      "step": 677
-    },
-    {
-      "epoch": 0.11559794336254896,
-      "grad_norm": 0.8685258030891418,
-      "learning_rate": 8.843989769820973e-06,
-      "loss": 0.1205,
-      "step": 678
-    },
-    {
-      "epoch": 0.11576844180408663,
-      "grad_norm": 0.8280768990516663,
-      "learning_rate": 8.842284739982951e-06,
-      "loss": 0.1036,
-      "step": 679
-    },
-    {
-      "epoch": 0.11593894024562432,
-      "grad_norm": 1.205714464187622,
-      "learning_rate": 8.840579710144929e-06,
-      "loss": 0.1832,
-      "step": 680
-    },
-    {
-      "epoch": 0.116109438687162,
-      "grad_norm": 0.6061711311340332,
-      "learning_rate": 8.838874680306905e-06,
-      "loss": 0.1249,
-      "step": 681
-    },
-    {
-      "epoch": 0.11627993712869969,
-      "grad_norm": 0.6693856716156006,
-      "learning_rate": 8.837169650468885e-06,
-      "loss": 0.1709,
-      "step": 682
-    },
-    {
-      "epoch": 0.11645043557023736,
-      "grad_norm": 0.7403165102005005,
-      "learning_rate": 8.835464620630861e-06,
-      "loss": 0.1826,
-      "step": 683
-    },
-    {
-      "epoch": 0.11662093401177505,
-      "grad_norm": 1.0672601461410522,
-      "learning_rate": 8.833759590792839e-06,
-      "loss": 0.187,
-      "step": 684
-    },
-    {
-      "epoch": 0.11679143245331274,
-      "grad_norm": 0.6454370021820068,
-      "learning_rate": 8.832054560954817e-06,
-      "loss": 0.1536,
-      "step": 685
-    },
-    {
-      "epoch": 0.11696193089485041,
-      "grad_norm": 0.7109602689743042,
-      "learning_rate": 8.830349531116795e-06,
-      "loss": 0.1025,
-      "step": 686
-    },
-    {
-      "epoch": 0.1171324293363881,
-      "grad_norm": 1.0405216217041016,
-      "learning_rate": 8.828644501278773e-06,
-      "loss": 0.051,
-      "step": 687
-    },
-    {
-      "epoch": 0.11730292777792578,
-      "grad_norm": 1.011011004447937,
-      "learning_rate": 8.826939471440751e-06,
-      "loss": 0.2072,
-      "step": 688
-    },
-    {
-      "epoch": 0.11747342621946347,
-      "grad_norm": 0.8015874028205872,
-      "learning_rate": 8.825234441602729e-06,
-      "loss": 0.1706,
-      "step": 689
-    },
-    {
-      "epoch": 0.11764392466100114,
-      "grad_norm": 0.603288471698761,
-      "learning_rate": 8.823529411764707e-06,
-      "loss": 0.1464,
-      "step": 690
-    },
-    {
-      "epoch": 0.11781442310253883,
-      "grad_norm": 0.9477997422218323,
-      "learning_rate": 8.821824381926685e-06,
-      "loss": 0.1726,
-      "step": 691
-    },
-    {
-      "epoch": 0.11798492154407651,
-      "grad_norm": 0.5780686736106873,
-      "learning_rate": 8.820119352088663e-06,
-      "loss": 0.1528,
-      "step": 692
-    },
-    {
-      "epoch": 0.1181554199856142,
-      "grad_norm": 0.5830322504043579,
-      "learning_rate": 8.81841432225064e-06,
-      "loss": 0.1505,
-      "step": 693
-    },
-    {
-      "epoch": 0.11832591842715187,
-      "grad_norm": 1.1404770612716675,
-      "learning_rate": 8.816709292412617e-06,
-      "loss": 0.1729,
-      "step": 694
-    },
-    {
-      "epoch": 0.11849641686868956,
-      "grad_norm": 0.8645011782646179,
-      "learning_rate": 8.815004262574597e-06,
-      "loss": 0.1541,
-      "step": 695
-    },
-    {
-      "epoch": 0.11866691531022724,
-      "grad_norm": 0.5606808662414551,
-      "learning_rate": 8.813299232736574e-06,
-      "loss": 0.1299,
-      "step": 696
-    },
-    {
-      "epoch": 0.11883741375176493,
-      "grad_norm": 1.530091643333435,
-      "learning_rate": 8.81159420289855e-06,
-      "loss": 0.1978,
-      "step": 697
-    },
-    {
-      "epoch": 0.1190079121933026,
-      "grad_norm": 0.5594825744628906,
-      "learning_rate": 8.80988917306053e-06,
-      "loss": 0.161,
-      "step": 698
-    },
-    {
-      "epoch": 0.11917841063484029,
-      "grad_norm": 0.5259520411491394,
-      "learning_rate": 8.808184143222507e-06,
-      "loss": 0.083,
-      "step": 699
-    },
-    {
-      "epoch": 0.11934890907637798,
-      "grad_norm": 0.604471743106842,
-      "learning_rate": 8.806479113384485e-06,
-      "loss": 0.0862,
-      "step": 700
-    },
-    {
-      "epoch": 0.11934890907637798,
-      "eval_f1_score": 0.05693950177935943,
-      "eval_loss": 0.1808873862028122,
-      "eval_runtime": 182.561,
-      "eval_samples_per_second": 54.776,
-      "eval_steps_per_second": 3.424,
-      "step": 700
-    },
-    {
-      "epoch": 0.11951940751791566,
-      "grad_norm": 0.7942699193954468,
-      "learning_rate": 8.804774083546463e-06,
-      "loss": 0.149,
-      "step": 701
-    },
-    {
-      "epoch": 0.11968990595945334,
-      "grad_norm": 0.9464733600616455,
-      "learning_rate": 8.80306905370844e-06,
-      "loss": 0.1971,
-      "step": 702
-    },
-    {
-      "epoch": 0.11986040440099102,
-      "grad_norm": 0.8487085103988647,
-      "learning_rate": 8.801364023870418e-06,
-      "loss": 0.0749,
-      "step": 703
-    },
-    {
-      "epoch": 0.12003090284252871,
-      "grad_norm": 0.6099476218223572,
-      "learning_rate": 8.799658994032396e-06,
-      "loss": 0.165,
-      "step": 704
-    },
-    {
-      "epoch": 0.12020140128406638,
-      "grad_norm": 0.7376194000244141,
-      "learning_rate": 8.797953964194374e-06,
-      "loss": 0.1395,
-      "step": 705
-    },
-    {
-      "epoch": 0.12037189972560407,
-      "grad_norm": 0.8289808630943298,
-      "learning_rate": 8.796248934356352e-06,
-      "loss": 0.2172,
-      "step": 706
-    },
-    {
-      "epoch": 0.12054239816714175,
-      "grad_norm": 0.4552036225795746,
-      "learning_rate": 8.79454390451833e-06,
-      "loss": 0.1506,
-      "step": 707
-    },
-    {
-      "epoch": 0.12071289660867944,
-      "grad_norm": 0.5763758420944214,
-      "learning_rate": 8.792838874680306e-06,
-      "loss": 0.1564,
-      "step": 708
-    },
-    {
-      "epoch": 0.12088339505021711,
-      "grad_norm": 0.9908771514892578,
-      "learning_rate": 8.791133844842286e-06,
-      "loss": 0.1016,
-      "step": 709
-    },
-    {
-      "epoch": 0.1210538934917548,
-      "grad_norm": 0.8617611527442932,
-      "learning_rate": 8.789428815004262e-06,
-      "loss": 0.1663,
-      "step": 710
-    },
-    {
-      "epoch": 0.12122439193329249,
-      "grad_norm": 0.6195103526115417,
-      "learning_rate": 8.78772378516624e-06,
-      "loss": 0.0743,
-      "step": 711
-    },
-    {
-      "epoch": 0.12139489037483017,
-      "grad_norm": 0.8117532730102539,
-      "learning_rate": 8.78601875532822e-06,
-      "loss": 0.1828,
-      "step": 712
-    },
-    {
-      "epoch": 0.12156538881636784,
-      "grad_norm": 0.5788756608963013,
-      "learning_rate": 8.784313725490196e-06,
-      "loss": 0.1435,
-      "step": 713
-    },
-    {
-      "epoch": 0.12173588725790553,
-      "grad_norm": 0.6565443277359009,
-      "learning_rate": 8.782608695652174e-06,
-      "loss": 0.1332,
-      "step": 714
-    },
-    {
-      "epoch": 0.12190638569944322,
-      "grad_norm": 0.7393292784690857,
-      "learning_rate": 8.780903665814152e-06,
-      "loss": 0.1103,
-      "step": 715
-    },
-    {
-      "epoch": 0.1220768841409809,
-      "grad_norm": 0.500567615032196,
-      "learning_rate": 8.77919863597613e-06,
-      "loss": 0.0809,
-      "step": 716
-    },
-    {
-      "epoch": 0.12224738258251858,
-      "grad_norm": 1.1821036338806152,
-      "learning_rate": 8.777493606138108e-06,
-      "loss": 0.1491,
-      "step": 717
-    },
-    {
-      "epoch": 0.12241788102405626,
-      "grad_norm": 1.593512773513794,
-      "learning_rate": 8.775788576300086e-06,
-      "loss": 0.1833,
-      "step": 718
-    },
-    {
-      "epoch": 0.12258837946559395,
-      "grad_norm": 0.5709978938102722,
-      "learning_rate": 8.774083546462064e-06,
-      "loss": 0.0911,
-      "step": 719
-    },
-    {
-      "epoch": 0.12275887790713164,
-      "grad_norm": 1.388977289199829,
-      "learning_rate": 8.772378516624042e-06,
-      "loss": 0.2518,
-      "step": 720
-    },
-    {
-      "epoch": 0.12292937634866931,
-      "grad_norm": 0.7217902541160583,
-      "learning_rate": 8.77067348678602e-06,
-      "loss": 0.1836,
-      "step": 721
-    },
-    {
-      "epoch": 0.123099874790207,
-      "grad_norm": 0.7408892512321472,
-      "learning_rate": 8.768968456947998e-06,
-      "loss": 0.208,
-      "step": 722
-    },
-    {
-      "epoch": 0.12327037323174468,
-      "grad_norm": 1.1729018688201904,
-      "learning_rate": 8.767263427109976e-06,
-      "loss": 0.0926,
-      "step": 723
-    },
-    {
-      "epoch": 0.12344087167328237,
-      "grad_norm": 1.1482958793640137,
-      "learning_rate": 8.765558397271952e-06,
-      "loss": 0.2234,
-      "step": 724
-    },
-    {
-      "epoch": 0.12361137011482004,
-      "grad_norm": 1.9159002304077148,
-      "learning_rate": 8.763853367433932e-06,
-      "loss": 0.2116,
-      "step": 725
-    },
-    {
-      "epoch": 0.12378186855635773,
-      "grad_norm": 0.9399540424346924,
-      "learning_rate": 8.762148337595908e-06,
-      "loss": 0.1596,
-      "step": 726
-    },
-    {
-      "epoch": 0.12395236699789541,
-      "grad_norm": 0.8376556038856506,
-      "learning_rate": 8.760443307757886e-06,
-      "loss": 0.2133,
-      "step": 727
-    },
-    {
-      "epoch": 0.12412286543943309,
-      "grad_norm": 0.8435215950012207,
-      "learning_rate": 8.758738277919865e-06,
-      "loss": 0.1411,
-      "step": 728
-    },
-    {
-      "epoch": 0.12429336388097077,
-      "grad_norm": 0.7841005325317383,
-      "learning_rate": 8.757033248081842e-06,
-      "loss": 0.1757,
-      "step": 729
-    },
-    {
-      "epoch": 0.12446386232250846,
-      "grad_norm": 1.5041134357452393,
-      "learning_rate": 8.75532821824382e-06,
-      "loss": 0.2328,
-      "step": 730
-    },
-    {
-      "epoch": 0.12463436076404615,
-      "grad_norm": 1.0415327548980713,
-      "learning_rate": 8.753623188405798e-06,
-      "loss": 0.1609,
-      "step": 731
-    },
-    {
-      "epoch": 0.12480485920558382,
-      "grad_norm": 0.6443787217140198,
-      "learning_rate": 8.751918158567776e-06,
-      "loss": 0.1106,
-      "step": 732
-    },
-    {
-      "epoch": 0.1249753576471215,
-      "grad_norm": 0.5113686919212341,
-      "learning_rate": 8.750213128729753e-06,
-      "loss": 0.1231,
-      "step": 733
-    },
-    {
-      "epoch": 0.1251458560886592,
-      "grad_norm": 0.686532199382782,
-      "learning_rate": 8.748508098891731e-06,
-      "loss": 0.1048,
-      "step": 734
-    },
-    {
-      "epoch": 0.12531635453019688,
-      "grad_norm": 0.7583132982254028,
-      "learning_rate": 8.74680306905371e-06,
-      "loss": 0.1473,
-      "step": 735
-    },
-    {
-      "epoch": 0.12548685297173456,
-      "grad_norm": 0.9985403418540955,
-      "learning_rate": 8.745098039215687e-06,
-      "loss": 0.1695,
-      "step": 736
-    },
-    {
-      "epoch": 0.12565735141327225,
-      "grad_norm": 1.0437244176864624,
-      "learning_rate": 8.743393009377665e-06,
-      "loss": 0.1984,
-      "step": 737
-    },
-    {
-      "epoch": 0.1258278498548099,
-      "grad_norm": 1.27836275100708,
-      "learning_rate": 8.741687979539643e-06,
-      "loss": 0.2065,
-      "step": 738
-    },
-    {
-      "epoch": 0.1259983482963476,
-      "grad_norm": 0.649793267250061,
-      "learning_rate": 8.739982949701621e-06,
-      "loss": 0.0989,
-      "step": 739
-    },
-    {
-      "epoch": 0.12616884673788528,
-      "grad_norm": 0.5728198289871216,
-      "learning_rate": 8.738277919863597e-06,
-      "loss": 0.1628,
-      "step": 740
-    },
-    {
-      "epoch": 0.12633934517942297,
-      "grad_norm": 0.7949087619781494,
-      "learning_rate": 8.736572890025577e-06,
-      "loss": 0.1628,
-      "step": 741
-    },
-    {
-      "epoch": 0.12650984362096065,
-      "grad_norm": 1.0230469703674316,
-      "learning_rate": 8.734867860187553e-06,
-      "loss": 0.1416,
-      "step": 742
-    },
-    {
-      "epoch": 0.12668034206249834,
-      "grad_norm": 0.6708950996398926,
-      "learning_rate": 8.733162830349531e-06,
-      "loss": 0.1691,
-      "step": 743
-    },
-    {
-      "epoch": 0.12685084050403603,
-      "grad_norm": 0.7631368041038513,
-      "learning_rate": 8.73145780051151e-06,
-      "loss": 0.1223,
-      "step": 744
-    },
-    {
-      "epoch": 0.12702133894557371,
-      "grad_norm": 1.418788194656372,
-      "learning_rate": 8.729752770673487e-06,
-      "loss": 0.1503,
-      "step": 745
-    },
-    {
-      "epoch": 0.12719183738711137,
-      "grad_norm": 1.5092041492462158,
-      "learning_rate": 8.728047740835465e-06,
-      "loss": 0.2056,
-      "step": 746
-    },
-    {
-      "epoch": 0.12736233582864906,
-      "grad_norm": 1.0009962320327759,
-      "learning_rate": 8.726342710997443e-06,
-      "loss": 0.1508,
-      "step": 747
-    },
-    {
-      "epoch": 0.12753283427018675,
-      "grad_norm": 1.2288755178451538,
-      "learning_rate": 8.724637681159421e-06,
-      "loss": 0.2004,
-      "step": 748
-    },
-    {
-      "epoch": 0.12770333271172443,
-      "grad_norm": 0.6066832542419434,
-      "learning_rate": 8.722932651321399e-06,
-      "loss": 0.1084,
-      "step": 749
-    },
-    {
-      "epoch": 0.12787383115326212,
-      "grad_norm": 0.8936707973480225,
-      "learning_rate": 8.721227621483377e-06,
-      "loss": 0.1569,
-      "step": 750
-    },
-    {
-      "epoch": 0.1280443295947998,
-      "grad_norm": 1.5922664403915405,
-      "learning_rate": 8.719522591645355e-06,
-      "loss": 0.1859,
-      "step": 751
-    },
-    {
-      "epoch": 0.1282148280363375,
-      "grad_norm": 0.8145460486412048,
-      "learning_rate": 8.717817561807333e-06,
-      "loss": 0.1468,
-      "step": 752
-    },
-    {
-      "epoch": 0.12838532647787515,
-      "grad_norm": 0.8177946209907532,
-      "learning_rate": 8.71611253196931e-06,
-      "loss": 0.1329,
-      "step": 753
-    },
-    {
-      "epoch": 0.12855582491941284,
-      "grad_norm": 0.6831057667732239,
-      "learning_rate": 8.714407502131289e-06,
-      "loss": 0.1655,
-      "step": 754
-    },
-    {
-      "epoch": 0.12872632336095052,
-      "grad_norm": 1.3590229749679565,
-      "learning_rate": 8.712702472293267e-06,
-      "loss": 0.1688,
-      "step": 755
-    },
-    {
-      "epoch": 0.1288968218024882,
-      "grad_norm": 0.8421473503112793,
-      "learning_rate": 8.710997442455243e-06,
-      "loss": 0.1545,
-      "step": 756
-    },
-    {
-      "epoch": 0.1290673202440259,
-      "grad_norm": 0.8665107488632202,
-      "learning_rate": 8.709292412617223e-06,
-      "loss": 0.1541,
-      "step": 757
-    },
-    {
-      "epoch": 0.12923781868556358,
-      "grad_norm": 0.9077842235565186,
-      "learning_rate": 8.707587382779199e-06,
-      "loss": 0.1747,
-      "step": 758
-    },
-    {
-      "epoch": 0.12940831712710127,
-      "grad_norm": 0.773943305015564,
-      "learning_rate": 8.705882352941177e-06,
-      "loss": 0.1386,
-      "step": 759
-    },
-    {
-      "epoch": 0.12957881556863896,
-      "grad_norm": 1.8866589069366455,
-      "learning_rate": 8.704177323103155e-06,
-      "loss": 0.1541,
-      "step": 760
-    },
-    {
-      "epoch": 0.12974931401017661,
-      "grad_norm": 0.8734083771705627,
-      "learning_rate": 8.702472293265133e-06,
-      "loss": 0.0674,
-      "step": 761
-    },
-    {
-      "epoch": 0.1299198124517143,
-      "grad_norm": 0.9331802725791931,
-      "learning_rate": 8.70076726342711e-06,
-      "loss": 0.1726,
-      "step": 762
-    },
-    {
-      "epoch": 0.130090310893252,
-      "grad_norm": 0.7994747757911682,
-      "learning_rate": 8.699062233589089e-06,
-      "loss": 0.1104,
-      "step": 763
-    },
-    {
-      "epoch": 0.13026080933478967,
-      "grad_norm": 0.6318725347518921,
-      "learning_rate": 8.697357203751066e-06,
-      "loss": 0.0885,
-      "step": 764
-    },
-    {
-      "epoch": 0.13043130777632736,
-      "grad_norm": 0.8578609228134155,
-      "learning_rate": 8.695652173913044e-06,
-      "loss": 0.1932,
-      "step": 765
-    },
-    {
-      "epoch": 0.13060180621786505,
-      "grad_norm": 0.5934591293334961,
-      "learning_rate": 8.693947144075022e-06,
-      "loss": 0.1409,
-      "step": 766
-    },
-    {
-      "epoch": 0.13077230465940273,
-      "grad_norm": 0.9246014952659607,
-      "learning_rate": 8.692242114237e-06,
-      "loss": 0.1927,
-      "step": 767
-    },
-    {
-      "epoch": 0.13094280310094042,
-      "grad_norm": 0.7142505049705505,
-      "learning_rate": 8.690537084398978e-06,
-      "loss": 0.1031,
-      "step": 768
-    },
-    {
-      "epoch": 0.13111330154247808,
-      "grad_norm": 1.084960699081421,
-      "learning_rate": 8.688832054560955e-06,
-      "loss": 0.1565,
-      "step": 769
-    },
-    {
-      "epoch": 0.13128379998401576,
-      "grad_norm": 0.9150413274765015,
-      "learning_rate": 8.687127024722934e-06,
-      "loss": 0.183,
-      "step": 770
-    },
-    {
-      "epoch": 0.13145429842555345,
-      "grad_norm": 2.815584421157837,
-      "learning_rate": 8.685421994884912e-06,
-      "loss": 0.3257,
-      "step": 771
-    },
-    {
-      "epoch": 0.13162479686709114,
-      "grad_norm": 0.6796224117279053,
-      "learning_rate": 8.683716965046888e-06,
-      "loss": 0.1414,
-      "step": 772
-    },
-    {
-      "epoch": 0.13179529530862882,
-      "grad_norm": 0.7766322493553162,
-      "learning_rate": 8.682011935208868e-06,
-      "loss": 0.1271,
-      "step": 773
-    },
-    {
-      "epoch": 0.1319657937501665,
-      "grad_norm": 1.089959740638733,
-      "learning_rate": 8.680306905370844e-06,
-      "loss": 0.2125,
-      "step": 774
-    },
-    {
-      "epoch": 0.1321362921917042,
-      "grad_norm": 0.8805111646652222,
-      "learning_rate": 8.678601875532822e-06,
-      "loss": 0.271,
-      "step": 775
-    },
-    {
-      "epoch": 0.13230679063324186,
-      "grad_norm": 0.9524927139282227,
-      "learning_rate": 8.6768968456948e-06,
-      "loss": 0.1699,
-      "step": 776
-    },
-    {
-      "epoch": 0.13247728907477954,
-      "grad_norm": 0.9875715374946594,
-      "learning_rate": 8.675191815856778e-06,
-      "loss": 0.1119,
-      "step": 777
-    },
-    {
-      "epoch": 0.13264778751631723,
-      "grad_norm": 1.037915587425232,
-      "learning_rate": 8.673486786018756e-06,
-      "loss": 0.1976,
-      "step": 778
-    },
-    {
-      "epoch": 0.13281828595785491,
-      "grad_norm": 0.7319202423095703,
-      "learning_rate": 8.671781756180734e-06,
-      "loss": 0.1588,
-      "step": 779
-    },
-    {
-      "epoch": 0.1329887843993926,
-      "grad_norm": 0.7272329330444336,
-      "learning_rate": 8.670076726342712e-06,
-      "loss": 0.1671,
-      "step": 780
-    },
-    {
-      "epoch": 0.1331592828409303,
-      "grad_norm": 2.4887404441833496,
-      "learning_rate": 8.66837169650469e-06,
-      "loss": 0.1451,
-      "step": 781
-    },
-    {
-      "epoch": 0.13332978128246797,
-      "grad_norm": 2.0834434032440186,
-      "learning_rate": 8.666666666666668e-06,
-      "loss": 0.1788,
-      "step": 782
-    },
-    {
-      "epoch": 0.13350027972400566,
-      "grad_norm": 2.073859453201294,
-      "learning_rate": 8.664961636828644e-06,
-      "loss": 0.314,
-      "step": 783
-    },
-    {
-      "epoch": 0.13367077816554332,
-      "grad_norm": 1.1813137531280518,
-      "learning_rate": 8.663256606990624e-06,
-      "loss": 0.1565,
-      "step": 784
-    },
-    {
-      "epoch": 0.133841276607081,
-      "grad_norm": 0.9659483432769775,
-      "learning_rate": 8.6615515771526e-06,
-      "loss": 0.1986,
-      "step": 785
-    },
-    {
-      "epoch": 0.1340117750486187,
-      "grad_norm": 0.7730090618133545,
-      "learning_rate": 8.659846547314578e-06,
-      "loss": 0.1395,
-      "step": 786
-    },
-    {
-      "epoch": 0.13418227349015638,
-      "grad_norm": 1.5010566711425781,
-      "learning_rate": 8.658141517476558e-06,
-      "loss": 0.2804,
-      "step": 787
-    },
-    {
-      "epoch": 0.13435277193169406,
-      "grad_norm": 0.6788565516471863,
-      "learning_rate": 8.656436487638534e-06,
-      "loss": 0.1596,
-      "step": 788
-    },
-    {
-      "epoch": 0.13452327037323175,
-      "grad_norm": 0.6773743629455566,
-      "learning_rate": 8.654731457800512e-06,
-      "loss": 0.1682,
-      "step": 789
-    },
-    {
-      "epoch": 0.13469376881476944,
-      "grad_norm": 0.8464353680610657,
-      "learning_rate": 8.65302642796249e-06,
-      "loss": 0.1261,
-      "step": 790
-    },
-    {
-      "epoch": 0.1348642672563071,
-      "grad_norm": 1.5723978281021118,
-      "learning_rate": 8.651321398124468e-06,
-      "loss": 0.1886,
-      "step": 791
-    },
-    {
-      "epoch": 0.13503476569784478,
-      "grad_norm": 1.281877875328064,
-      "learning_rate": 8.649616368286446e-06,
-      "loss": 0.2739,
-      "step": 792
-    },
-    {
-      "epoch": 0.13520526413938247,
-      "grad_norm": 0.9185252785682678,
-      "learning_rate": 8.647911338448424e-06,
-      "loss": 0.1475,
-      "step": 793
-    },
-    {
-      "epoch": 0.13537576258092016,
-      "grad_norm": 0.8964889049530029,
-      "learning_rate": 8.646206308610402e-06,
-      "loss": 0.2107,
-      "step": 794
-    },
-    {
-      "epoch": 0.13554626102245784,
-      "grad_norm": 0.8502851128578186,
-      "learning_rate": 8.64450127877238e-06,
-      "loss": 0.1214,
-      "step": 795
-    },
-    {
-      "epoch": 0.13571675946399553,
-      "grad_norm": 0.7894145846366882,
-      "learning_rate": 8.642796248934357e-06,
-      "loss": 0.1776,
-      "step": 796
-    },
-    {
-      "epoch": 0.13588725790553322,
-      "grad_norm": 0.8448327779769897,
-      "learning_rate": 8.641091219096335e-06,
-      "loss": 0.1178,
-      "step": 797
-    },
-    {
-      "epoch": 0.1360577563470709,
-      "grad_norm": 0.682449996471405,
-      "learning_rate": 8.639386189258313e-06,
-      "loss": 0.1421,
-      "step": 798
-    },
-    {
-      "epoch": 0.13622825478860856,
-      "grad_norm": 0.7914981842041016,
-      "learning_rate": 8.63768115942029e-06,
-      "loss": 0.1547,
-      "step": 799
-    },
-    {
-      "epoch": 0.13639875323014625,
-      "grad_norm": 1.071673035621643,
-      "learning_rate": 8.63597612958227e-06,
-      "loss": 0.146,
-      "step": 800
-    },
-    {
-      "epoch": 0.13639875323014625,
-      "eval_f1_score": 0.1461794019933555,
-      "eval_loss": 0.17923100292682648,
-      "eval_runtime": 182.6021,
-      "eval_samples_per_second": 54.764,
-      "eval_steps_per_second": 3.423,
-      "step": 800
-    },
-    {
-      "epoch": 0.13656925167168393,
-      "grad_norm": 0.9378244876861572,
-      "learning_rate": 8.634271099744245e-06,
-      "loss": 0.1741,
-      "step": 801
-    },
-    {
-      "epoch": 0.13673975011322162,
-      "grad_norm": 0.5133070945739746,
-      "learning_rate": 8.632566069906223e-06,
-      "loss": 0.098,
-      "step": 802
-    },
-    {
-      "epoch": 0.1369102485547593,
-      "grad_norm": 0.6201533079147339,
-      "learning_rate": 8.630861040068201e-06,
-      "loss": 0.1666,
-      "step": 803
-    },
-    {
-      "epoch": 0.137080746996297,
-      "grad_norm": 0.6914802193641663,
-      "learning_rate": 8.62915601023018e-06,
-      "loss": 0.121,
-      "step": 804
-    },
-    {
-      "epoch": 0.13725124543783468,
-      "grad_norm": 1.078437328338623,
-      "learning_rate": 8.627450980392157e-06,
-      "loss": 0.1503,
-      "step": 805
-    },
-    {
-      "epoch": 0.13742174387937237,
-      "grad_norm": 0.5333152413368225,
-      "learning_rate": 8.625745950554135e-06,
-      "loss": 0.0863,
-      "step": 806
-    },
-    {
-      "epoch": 0.13759224232091002,
-      "grad_norm": 0.8483152389526367,
-      "learning_rate": 8.624040920716113e-06,
-      "loss": 0.1782,
-      "step": 807
-    },
-    {
-      "epoch": 0.1377627407624477,
-      "grad_norm": 1.230548620223999,
-      "learning_rate": 8.622335890878091e-06,
-      "loss": 0.1795,
-      "step": 808
-    },
-    {
-      "epoch": 0.1379332392039854,
-      "grad_norm": 1.096155047416687,
-      "learning_rate": 8.620630861040069e-06,
-      "loss": 0.15,
-      "step": 809
-    },
-    {
-      "epoch": 0.13810373764552308,
-      "grad_norm": 1.0700522661209106,
-      "learning_rate": 8.618925831202047e-06,
-      "loss": 0.1256,
-      "step": 810
-    },
-    {
-      "epoch": 0.13827423608706077,
-      "grad_norm": 0.9942691326141357,
-      "learning_rate": 8.617220801364025e-06,
-      "loss": 0.0621,
-      "step": 811
-    },
-    {
-      "epoch": 0.13844473452859846,
-      "grad_norm": 1.1681605577468872,
-      "learning_rate": 8.615515771526003e-06,
-      "loss": 0.2706,
-      "step": 812
-    },
-    {
-      "epoch": 0.13861523297013614,
-      "grad_norm": 0.6961143612861633,
-      "learning_rate": 8.61381074168798e-06,
-      "loss": 0.1214,
-      "step": 813
-    },
-    {
-      "epoch": 0.1387857314116738,
-      "grad_norm": 1.5237188339233398,
-      "learning_rate": 8.612105711849959e-06,
-      "loss": 0.1471,
-      "step": 814
-    },
-    {
-      "epoch": 0.1389562298532115,
-      "grad_norm": 2.2736763954162598,
-      "learning_rate": 8.610400682011935e-06,
-      "loss": 0.314,
-      "step": 815
-    },
-    {
-      "epoch": 0.13912672829474917,
-      "grad_norm": 0.842560887336731,
-      "learning_rate": 8.608695652173915e-06,
-      "loss": 0.1069,
-      "step": 816
-    },
-    {
-      "epoch": 0.13929722673628686,
-      "grad_norm": 0.838711142539978,
-      "learning_rate": 8.606990622335891e-06,
-      "loss": 0.1392,
-      "step": 817
-    },
-    {
-      "epoch": 0.13946772517782455,
-      "grad_norm": 1.5051636695861816,
-      "learning_rate": 8.605285592497869e-06,
-      "loss": 0.2429,
-      "step": 818
-    },
-    {
-      "epoch": 0.13963822361936223,
-      "grad_norm": 0.9790658950805664,
-      "learning_rate": 8.603580562659847e-06,
-      "loss": 0.1904,
-      "step": 819
-    },
-    {
-      "epoch": 0.13980872206089992,
-      "grad_norm": 1.0247228145599365,
-      "learning_rate": 8.601875532821825e-06,
-      "loss": 0.1407,
-      "step": 820
-    },
-    {
-      "epoch": 0.1399792205024376,
-      "grad_norm": 1.1538825035095215,
-      "learning_rate": 8.600170502983803e-06,
-      "loss": 0.1281,
-      "step": 821
-    },
-    {
-      "epoch": 0.14014971894397527,
-      "grad_norm": 1.488327980041504,
-      "learning_rate": 8.59846547314578e-06,
-      "loss": 0.1372,
-      "step": 822
-    },
-    {
-      "epoch": 0.14032021738551295,
-      "grad_norm": 1.2420367002487183,
-      "learning_rate": 8.596760443307759e-06,
-      "loss": 0.151,
-      "step": 823
-    },
-    {
-      "epoch": 0.14049071582705064,
-      "grad_norm": 1.1011592149734497,
-      "learning_rate": 8.595055413469737e-06,
-      "loss": 0.2513,
-      "step": 824
-    },
-    {
-      "epoch": 0.14066121426858832,
-      "grad_norm": 0.8864753246307373,
-      "learning_rate": 8.593350383631714e-06,
-      "loss": 0.1389,
-      "step": 825
-    },
-    {
-      "epoch": 0.140831712710126,
-      "grad_norm": 1.1462125778198242,
-      "learning_rate": 8.591645353793692e-06,
-      "loss": 0.167,
-      "step": 826
-    },
-    {
-      "epoch": 0.1410022111516637,
-      "grad_norm": 0.8221803903579712,
-      "learning_rate": 8.58994032395567e-06,
-      "loss": 0.1552,
-      "step": 827
-    },
-    {
-      "epoch": 0.14117270959320138,
-      "grad_norm": 0.8550702333450317,
-      "learning_rate": 8.588235294117647e-06,
-      "loss": 0.1512,
-      "step": 828
-    },
-    {
-      "epoch": 0.14134320803473907,
-      "grad_norm": 0.9521147012710571,
-      "learning_rate": 8.586530264279626e-06,
-      "loss": 0.186,
-      "step": 829
-    },
-    {
-      "epoch": 0.14151370647627673,
-      "grad_norm": 1.5711631774902344,
-      "learning_rate": 8.584825234441604e-06,
-      "loss": 0.3002,
-      "step": 830
-    },
-    {
-      "epoch": 0.14168420491781442,
-      "grad_norm": 0.6364051103591919,
-      "learning_rate": 8.58312020460358e-06,
-      "loss": 0.1998,
-      "step": 831
-    },
-    {
-      "epoch": 0.1418547033593521,
-      "grad_norm": 1.021705150604248,
-      "learning_rate": 8.58141517476556e-06,
-      "loss": 0.0973,
-      "step": 832
-    },
-    {
-      "epoch": 0.1420252018008898,
-      "grad_norm": 0.7612175345420837,
-      "learning_rate": 8.579710144927536e-06,
-      "loss": 0.2029,
-      "step": 833
-    },
-    {
-      "epoch": 0.14219570024242748,
-      "grad_norm": 0.8757889270782471,
-      "learning_rate": 8.578005115089514e-06,
-      "loss": 0.2264,
-      "step": 834
-    },
-    {
-      "epoch": 0.14236619868396516,
-      "grad_norm": 0.7471174597740173,
-      "learning_rate": 8.576300085251492e-06,
-      "loss": 0.1473,
-      "step": 835
-    },
-    {
-      "epoch": 0.14253669712550285,
-      "grad_norm": 1.4478951692581177,
-      "learning_rate": 8.57459505541347e-06,
-      "loss": 0.1916,
-      "step": 836
-    },
-    {
-      "epoch": 0.1427071955670405,
-      "grad_norm": 1.1510059833526611,
-      "learning_rate": 8.572890025575448e-06,
-      "loss": 0.1959,
-      "step": 837
-    },
-    {
-      "epoch": 0.1428776940085782,
-      "grad_norm": 0.769268810749054,
-      "learning_rate": 8.571184995737426e-06,
-      "loss": 0.1295,
-      "step": 838
-    },
-    {
-      "epoch": 0.14304819245011588,
-      "grad_norm": 0.692801296710968,
-      "learning_rate": 8.569479965899404e-06,
-      "loss": 0.1456,
-      "step": 839
-    },
-    {
-      "epoch": 0.14321869089165357,
-      "grad_norm": 0.8778480291366577,
-      "learning_rate": 8.567774936061382e-06,
-      "loss": 0.2729,
-      "step": 840
-    },
-    {
-      "epoch": 0.14338918933319125,
-      "grad_norm": 0.716971218585968,
-      "learning_rate": 8.56606990622336e-06,
-      "loss": 0.1546,
-      "step": 841
-    },
-    {
-      "epoch": 0.14355968777472894,
-      "grad_norm": 2.0981221199035645,
-      "learning_rate": 8.564364876385338e-06,
-      "loss": 0.2563,
-      "step": 842
-    },
-    {
-      "epoch": 0.14373018621626663,
-      "grad_norm": 0.6106297969818115,
-      "learning_rate": 8.562659846547316e-06,
-      "loss": 0.1605,
-      "step": 843
-    },
-    {
-      "epoch": 0.1439006846578043,
-      "grad_norm": 0.7936435341835022,
-      "learning_rate": 8.560954816709292e-06,
-      "loss": 0.0751,
-      "step": 844
-    },
-    {
-      "epoch": 0.14407118309934197,
-      "grad_norm": 0.8291981816291809,
-      "learning_rate": 8.559249786871272e-06,
-      "loss": 0.166,
-      "step": 845
-    },
-    {
-      "epoch": 0.14424168154087966,
-      "grad_norm": 0.7863637804985046,
-      "learning_rate": 8.55754475703325e-06,
-      "loss": 0.1637,
-      "step": 846
-    },
-    {
-      "epoch": 0.14441217998241734,
-      "grad_norm": 0.5928487777709961,
-      "learning_rate": 8.555839727195226e-06,
-      "loss": 0.2067,
-      "step": 847
-    },
-    {
-      "epoch": 0.14458267842395503,
-      "grad_norm": 0.7105429172515869,
-      "learning_rate": 8.554134697357206e-06,
-      "loss": 0.1674,
-      "step": 848
-    },
-    {
-      "epoch": 0.14475317686549272,
-      "grad_norm": 0.5402317047119141,
-      "learning_rate": 8.552429667519182e-06,
-      "loss": 0.1455,
-      "step": 849
-    },
-    {
-      "epoch": 0.1449236753070304,
-      "grad_norm": 0.8685750961303711,
-      "learning_rate": 8.55072463768116e-06,
-      "loss": 0.1576,
-      "step": 850
-    },
-    {
-      "epoch": 0.1450941737485681,
-      "grad_norm": 0.6241790652275085,
-      "learning_rate": 8.549019607843138e-06,
-      "loss": 0.1207,
-      "step": 851
-    },
-    {
-      "epoch": 0.14526467219010578,
-      "grad_norm": 0.6889399886131287,
-      "learning_rate": 8.547314578005116e-06,
-      "loss": 0.1418,
-      "step": 852
-    },
-    {
-      "epoch": 0.14543517063164343,
-      "grad_norm": 3.354299306869507,
-      "learning_rate": 8.545609548167094e-06,
-      "loss": 0.126,
-      "step": 853
-    },
-    {
-      "epoch": 0.14560566907318112,
-      "grad_norm": 1.1726495027542114,
-      "learning_rate": 8.543904518329072e-06,
-      "loss": 0.2264,
-      "step": 854
-    },
-    {
-      "epoch": 0.1457761675147188,
-      "grad_norm": 0.8029992580413818,
-      "learning_rate": 8.54219948849105e-06,
-      "loss": 0.1017,
-      "step": 855
-    },
-    {
-      "epoch": 0.1459466659562565,
-      "grad_norm": 0.9463875889778137,
-      "learning_rate": 8.540494458653027e-06,
-      "loss": 0.1911,
-      "step": 856
-    },
-    {
-      "epoch": 0.14611716439779418,
-      "grad_norm": 0.8631307482719421,
-      "learning_rate": 8.538789428815005e-06,
-      "loss": 0.1079,
-      "step": 857
-    },
-    {
-      "epoch": 0.14628766283933187,
-      "grad_norm": 0.9463744759559631,
-      "learning_rate": 8.537084398976982e-06,
-      "loss": 0.2081,
-      "step": 858
-    },
-    {
-      "epoch": 0.14645816128086955,
-      "grad_norm": 0.7745504975318909,
-      "learning_rate": 8.535379369138961e-06,
-      "loss": 0.1143,
-      "step": 859
-    },
-    {
-      "epoch": 0.1466286597224072,
-      "grad_norm": 0.7032846808433533,
-      "learning_rate": 8.533674339300938e-06,
-      "loss": 0.1648,
-      "step": 860
-    },
-    {
-      "epoch": 0.1467991581639449,
-      "grad_norm": 1.1793239116668701,
-      "learning_rate": 8.531969309462916e-06,
-      "loss": 0.2458,
-      "step": 861
-    },
-    {
-      "epoch": 0.14696965660548258,
-      "grad_norm": 0.8669989109039307,
-      "learning_rate": 8.530264279624895e-06,
-      "loss": 0.0996,
-      "step": 862
-    },
-    {
-      "epoch": 0.14714015504702027,
-      "grad_norm": 1.4623005390167236,
-      "learning_rate": 8.528559249786871e-06,
-      "loss": 0.2536,
-      "step": 863
-    },
-    {
-      "epoch": 0.14731065348855796,
-      "grad_norm": 0.7780196070671082,
-      "learning_rate": 8.52685421994885e-06,
-      "loss": 0.1102,
-      "step": 864
-    },
-    {
-      "epoch": 0.14748115193009564,
-      "grad_norm": 0.8474672436714172,
-      "learning_rate": 8.525149190110827e-06,
-      "loss": 0.1611,
-      "step": 865
-    },
-    {
-      "epoch": 0.14765165037163333,
-      "grad_norm": 0.7766311168670654,
-      "learning_rate": 8.523444160272805e-06,
-      "loss": 0.1206,
-      "step": 866
-    },
-    {
-      "epoch": 0.14782214881317102,
-      "grad_norm": 0.6879698038101196,
-      "learning_rate": 8.521739130434783e-06,
-      "loss": 0.1676,
-      "step": 867
-    },
-    {
-      "epoch": 0.14799264725470868,
-      "grad_norm": 0.6485670208930969,
-      "learning_rate": 8.520034100596761e-06,
-      "loss": 0.1672,
-      "step": 868
-    },
-    {
-      "epoch": 0.14816314569624636,
-      "grad_norm": 0.7620902061462402,
-      "learning_rate": 8.518329070758739e-06,
-      "loss": 0.1299,
-      "step": 869
-    },
-    {
-      "epoch": 0.14833364413778405,
-      "grad_norm": 0.9798589944839478,
-      "learning_rate": 8.516624040920717e-06,
-      "loss": 0.0973,
-      "step": 870
-    },
-    {
-      "epoch": 0.14850414257932174,
-      "grad_norm": 0.8749964237213135,
-      "learning_rate": 8.514919011082695e-06,
-      "loss": 0.065,
-      "step": 871
-    },
-    {
-      "epoch": 0.14867464102085942,
-      "grad_norm": 0.6423231959342957,
-      "learning_rate": 8.513213981244673e-06,
-      "loss": 0.1317,
-      "step": 872
-    },
-    {
-      "epoch": 0.1488451394623971,
-      "grad_norm": 0.979634165763855,
-      "learning_rate": 8.511508951406651e-06,
-      "loss": 0.1331,
-      "step": 873
-    },
-    {
-      "epoch": 0.1490156379039348,
-      "grad_norm": 1.4354133605957031,
-      "learning_rate": 8.509803921568627e-06,
-      "loss": 0.1772,
-      "step": 874
-    },
-    {
-      "epoch": 0.14918613634547248,
-      "grad_norm": 0.6209403872489929,
-      "learning_rate": 8.508098891730607e-06,
-      "loss": 0.1024,
-      "step": 875
-    },
-    {
-      "epoch": 0.14935663478701014,
-      "grad_norm": 2.1505284309387207,
-      "learning_rate": 8.506393861892583e-06,
-      "loss": 0.2204,
-      "step": 876
-    },
-    {
-      "epoch": 0.14952713322854783,
-      "grad_norm": 1.8646060228347778,
-      "learning_rate": 8.504688832054561e-06,
-      "loss": 0.288,
-      "step": 877
-    },
-    {
-      "epoch": 0.1496976316700855,
-      "grad_norm": 0.7936485409736633,
-      "learning_rate": 8.502983802216539e-06,
-      "loss": 0.1333,
-      "step": 878
-    },
-    {
-      "epoch": 0.1498681301116232,
-      "grad_norm": 0.8515297174453735,
-      "learning_rate": 8.501278772378517e-06,
-      "loss": 0.1229,
-      "step": 879
-    },
-    {
-      "epoch": 0.15003862855316089,
-      "grad_norm": 0.9410006403923035,
-      "learning_rate": 8.499573742540495e-06,
-      "loss": 0.1328,
-      "step": 880
-    },
-    {
-      "epoch": 0.15020912699469857,
-      "grad_norm": 1.014402985572815,
-      "learning_rate": 8.497868712702473e-06,
-      "loss": 0.0523,
-      "step": 881
-    },
-    {
-      "epoch": 0.15037962543623626,
-      "grad_norm": 1.1425182819366455,
-      "learning_rate": 8.49616368286445e-06,
-      "loss": 0.152,
-      "step": 882
-    },
-    {
-      "epoch": 0.15055012387777392,
-      "grad_norm": 1.3995938301086426,
-      "learning_rate": 8.494458653026429e-06,
-      "loss": 0.2666,
-      "step": 883
-    },
-    {
-      "epoch": 0.1507206223193116,
-      "grad_norm": 0.7522984743118286,
-      "learning_rate": 8.492753623188407e-06,
-      "loss": 0.1231,
-      "step": 884
-    },
-    {
-      "epoch": 0.1508911207608493,
-      "grad_norm": 0.9752036929130554,
-      "learning_rate": 8.491048593350385e-06,
-      "loss": 0.1259,
-      "step": 885
-    },
-    {
-      "epoch": 0.15106161920238698,
-      "grad_norm": 1.2285127639770508,
-      "learning_rate": 8.489343563512363e-06,
-      "loss": 0.0806,
-      "step": 886
-    },
-    {
-      "epoch": 0.15123211764392466,
-      "grad_norm": 0.7990505695343018,
-      "learning_rate": 8.48763853367434e-06,
-      "loss": 0.1139,
-      "step": 887
-    },
-    {
-      "epoch": 0.15140261608546235,
-      "grad_norm": 0.7367463111877441,
-      "learning_rate": 8.485933503836318e-06,
-      "loss": 0.142,
-      "step": 888
-    },
-    {
-      "epoch": 0.15157311452700004,
-      "grad_norm": 0.958259642124176,
-      "learning_rate": 8.484228473998296e-06,
-      "loss": 0.1229,
-      "step": 889
-    },
-    {
-      "epoch": 0.15174361296853772,
-      "grad_norm": 0.8944674134254456,
-      "learning_rate": 8.482523444160273e-06,
-      "loss": 0.1641,
-      "step": 890
-    },
-    {
-      "epoch": 0.15191411141007538,
-      "grad_norm": 1.6810100078582764,
-      "learning_rate": 8.480818414322252e-06,
-      "loss": 0.2561,
-      "step": 891
-    },
-    {
-      "epoch": 0.15208460985161307,
-      "grad_norm": 0.8170380592346191,
-      "learning_rate": 8.479113384484229e-06,
-      "loss": 0.1703,
-      "step": 892
-    },
-    {
-      "epoch": 0.15225510829315075,
-      "grad_norm": 0.9030293226242065,
-      "learning_rate": 8.477408354646206e-06,
-      "loss": 0.0785,
-      "step": 893
-    },
-    {
-      "epoch": 0.15242560673468844,
-      "grad_norm": 1.2546629905700684,
-      "learning_rate": 8.475703324808184e-06,
-      "loss": 0.1747,
-      "step": 894
-    },
-    {
-      "epoch": 0.15259610517622613,
-      "grad_norm": 0.7067754864692688,
-      "learning_rate": 8.473998294970162e-06,
-      "loss": 0.1664,
-      "step": 895
-    },
-    {
-      "epoch": 0.1527666036177638,
-      "grad_norm": 1.1066006422042847,
-      "learning_rate": 8.47229326513214e-06,
-      "loss": 0.1139,
-      "step": 896
-    },
-    {
-      "epoch": 0.1529371020593015,
-      "grad_norm": 0.8568253517150879,
-      "learning_rate": 8.470588235294118e-06,
-      "loss": 0.1815,
-      "step": 897
-    },
-    {
-      "epoch": 0.15310760050083916,
-      "grad_norm": 0.5176457166671753,
-      "learning_rate": 8.468883205456096e-06,
-      "loss": 0.1411,
-      "step": 898
-    },
-    {
-      "epoch": 0.15327809894237684,
-      "grad_norm": 1.1265554428100586,
-      "learning_rate": 8.467178175618074e-06,
-      "loss": 0.1883,
-      "step": 899
-    },
-    {
-      "epoch": 0.15344859738391453,
-      "grad_norm": 1.7193188667297363,
-      "learning_rate": 8.465473145780052e-06,
-      "loss": 0.2669,
-      "step": 900
-    },
-    {
-      "epoch": 0.15344859738391453,
-      "eval_f1_score": 0.2702702702702703,
-      "eval_loss": 0.17168840765953064,
-      "eval_runtime": 182.5767,
-      "eval_samples_per_second": 54.772,
-      "eval_steps_per_second": 3.423,
-      "step": 900
-    },
-    {
-      "epoch": 0.15361909582545222,
-      "grad_norm": 0.8680484890937805,
-      "learning_rate": 8.46376811594203e-06,
-      "loss": 0.1288,
-      "step": 901
-    },
-    {
-      "epoch": 0.1537895942669899,
-      "grad_norm": 1.037379264831543,
-      "learning_rate": 8.462063086104008e-06,
-      "loss": 0.0937,
-      "step": 902
-    },
-    {
-      "epoch": 0.1539600927085276,
-      "grad_norm": 0.8336663246154785,
-      "learning_rate": 8.460358056265984e-06,
-      "loss": 0.2303,
-      "step": 903
-    },
-    {
-      "epoch": 0.15413059115006528,
-      "grad_norm": 1.141602635383606,
-      "learning_rate": 8.458653026427964e-06,
-      "loss": 0.2589,
-      "step": 904
-    },
-    {
-      "epoch": 0.15430108959160296,
-      "grad_norm": 1.1085407733917236,
-      "learning_rate": 8.456947996589942e-06,
-      "loss": 0.2392,
-      "step": 905
-    },
-    {
-      "epoch": 0.15447158803314062,
-      "grad_norm": 0.7251394987106323,
-      "learning_rate": 8.455242966751918e-06,
-      "loss": 0.1609,
-      "step": 906
-    },
-    {
-      "epoch": 0.1546420864746783,
-      "grad_norm": 0.8698781728744507,
-      "learning_rate": 8.453537936913898e-06,
-      "loss": 0.1607,
-      "step": 907
-    },
-    {
-      "epoch": 0.154812584916216,
-      "grad_norm": 0.7384206056594849,
-      "learning_rate": 8.451832907075874e-06,
-      "loss": 0.1801,
-      "step": 908
-    },
-    {
-      "epoch": 0.15498308335775368,
-      "grad_norm": 0.9124801754951477,
-      "learning_rate": 8.450127877237852e-06,
-      "loss": 0.1379,
-      "step": 909
-    },
-    {
-      "epoch": 0.15515358179929137,
-      "grad_norm": 0.6821106672286987,
-      "learning_rate": 8.44842284739983e-06,
-      "loss": 0.1811,
-      "step": 910
-    },
-    {
-      "epoch": 0.15532408024082905,
-      "grad_norm": 0.6709396839141846,
-      "learning_rate": 8.446717817561808e-06,
-      "loss": 0.1118,
-      "step": 911
-    },
-    {
-      "epoch": 0.15549457868236674,
-      "grad_norm": 1.2577794790267944,
-      "learning_rate": 8.445012787723786e-06,
-      "loss": 0.218,
-      "step": 912
-    },
-    {
-      "epoch": 0.15566507712390443,
-      "grad_norm": 0.7951264977455139,
-      "learning_rate": 8.443307757885764e-06,
-      "loss": 0.1542,
-      "step": 913
-    },
-    {
-      "epoch": 0.15583557556544209,
-      "grad_norm": 1.3476436138153076,
-      "learning_rate": 8.441602728047742e-06,
-      "loss": 0.2476,
-      "step": 914
-    },
-    {
-      "epoch": 0.15600607400697977,
-      "grad_norm": 1.033733606338501,
-      "learning_rate": 8.43989769820972e-06,
-      "loss": 0.1993,
-      "step": 915
-    },
-    {
-      "epoch": 0.15617657244851746,
-      "grad_norm": 1.7906930446624756,
-      "learning_rate": 8.438192668371698e-06,
-      "loss": 0.2485,
-      "step": 916
-    },
-    {
-      "epoch": 0.15634707089005515,
-      "grad_norm": 0.7418613433837891,
-      "learning_rate": 8.436487638533676e-06,
-      "loss": 0.1618,
-      "step": 917
-    },
-    {
-      "epoch": 0.15651756933159283,
-      "grad_norm": 1.316089153289795,
-      "learning_rate": 8.434782608695653e-06,
-      "loss": 0.23,
-      "step": 918
-    },
-    {
-      "epoch": 0.15668806777313052,
-      "grad_norm": 1.0219331979751587,
-      "learning_rate": 8.43307757885763e-06,
-      "loss": 0.2063,
-      "step": 919
-    },
-    {
-      "epoch": 0.1568585662146682,
-      "grad_norm": 1.3739581108093262,
-      "learning_rate": 8.43137254901961e-06,
-      "loss": 0.1689,
-      "step": 920
-    },
-    {
-      "epoch": 0.15702906465620586,
-      "grad_norm": 1.3870887756347656,
-      "learning_rate": 8.429667519181587e-06,
-      "loss": 0.2504,
-      "step": 921
-    },
-    {
-      "epoch": 0.15719956309774355,
-      "grad_norm": 1.140902042388916,
-      "learning_rate": 8.427962489343564e-06,
-      "loss": 0.2444,
-      "step": 922
-    },
-    {
-      "epoch": 0.15737006153928124,
-      "grad_norm": 1.5909425020217896,
-      "learning_rate": 8.426257459505543e-06,
-      "loss": 0.221,
-      "step": 923
-    },
-    {
-      "epoch": 0.15754055998081892,
-      "grad_norm": 1.0957505702972412,
-      "learning_rate": 8.42455242966752e-06,
-      "loss": 0.1642,
-      "step": 924
-    },
-    {
-      "epoch": 0.1577110584223566,
-      "grad_norm": 0.7079620361328125,
-      "learning_rate": 8.422847399829497e-06,
-      "loss": 0.1408,
-      "step": 925
-    },
-    {
-      "epoch": 0.1578815568638943,
-      "grad_norm": 0.9637887477874756,
-      "learning_rate": 8.421142369991475e-06,
-      "loss": 0.2089,
-      "step": 926
-    },
-    {
-      "epoch": 0.15805205530543198,
-      "grad_norm": 1.4444074630737305,
-      "learning_rate": 8.419437340153453e-06,
-      "loss": 0.104,
-      "step": 927
-    },
-    {
-      "epoch": 0.15822255374696967,
-      "grad_norm": 1.3934565782546997,
-      "learning_rate": 8.417732310315431e-06,
-      "loss": 0.2762,
-      "step": 928
-    },
-    {
-      "epoch": 0.15839305218850733,
-      "grad_norm": 0.7416796088218689,
-      "learning_rate": 8.41602728047741e-06,
-      "loss": 0.1508,
-      "step": 929
-    },
-    {
-      "epoch": 0.158563550630045,
-      "grad_norm": 0.5937446355819702,
-      "learning_rate": 8.414322250639387e-06,
-      "loss": 0.0793,
-      "step": 930
-    },
-    {
-      "epoch": 0.1587340490715827,
-      "grad_norm": 0.728579580783844,
-      "learning_rate": 8.412617220801365e-06,
-      "loss": 0.1153,
-      "step": 931
-    },
-    {
-      "epoch": 0.1589045475131204,
-      "grad_norm": 0.7820244431495667,
-      "learning_rate": 8.410912190963343e-06,
-      "loss": 0.066,
-      "step": 932
-    },
-    {
-      "epoch": 0.15907504595465807,
-      "grad_norm": 1.6547861099243164,
-      "learning_rate": 8.40920716112532e-06,
-      "loss": 0.2295,
-      "step": 933
-    },
-    {
-      "epoch": 0.15924554439619576,
-      "grad_norm": 1.0257244110107422,
-      "learning_rate": 8.407502131287299e-06,
-      "loss": 0.1739,
-      "step": 934
-    },
-    {
-      "epoch": 0.15941604283773345,
-      "grad_norm": 1.2456021308898926,
-      "learning_rate": 8.405797101449275e-06,
-      "loss": 0.2205,
-      "step": 935
-    },
-    {
-      "epoch": 0.15958654127927113,
-      "grad_norm": 0.9185108542442322,
-      "learning_rate": 8.404092071611253e-06,
-      "loss": 0.1653,
-      "step": 936
-    },
-    {
-      "epoch": 0.1597570397208088,
-      "grad_norm": 1.904354453086853,
-      "learning_rate": 8.402387041773231e-06,
-      "loss": 0.1273,
-      "step": 937
-    },
-    {
-      "epoch": 0.15992753816234648,
-      "grad_norm": 0.9777550101280212,
-      "learning_rate": 8.400682011935209e-06,
-      "loss": 0.254,
-      "step": 938
-    },
-    {
-      "epoch": 0.16009803660388416,
-      "grad_norm": 1.0197429656982422,
-      "learning_rate": 8.398976982097187e-06,
-      "loss": 0.1463,
-      "step": 939
-    },
-    {
-      "epoch": 0.16026853504542185,
-      "grad_norm": 0.6910821795463562,
-      "learning_rate": 8.397271952259165e-06,
-      "loss": 0.1425,
-      "step": 940
-    },
-    {
-      "epoch": 0.16043903348695954,
-      "grad_norm": 1.0888043642044067,
-      "learning_rate": 8.395566922421143e-06,
-      "loss": 0.1992,
-      "step": 941
-    },
-    {
-      "epoch": 0.16060953192849722,
-      "grad_norm": 0.6020128130912781,
-      "learning_rate": 8.39386189258312e-06,
-      "loss": 0.167,
-      "step": 942
-    },
-    {
-      "epoch": 0.1607800303700349,
-      "grad_norm": 0.7391735315322876,
-      "learning_rate": 8.392156862745099e-06,
-      "loss": 0.1455,
-      "step": 943
-    },
-    {
-      "epoch": 0.16095052881157257,
-      "grad_norm": 0.7194268703460693,
-      "learning_rate": 8.390451832907077e-06,
-      "loss": 0.1916,
-      "step": 944
-    },
-    {
-      "epoch": 0.16112102725311025,
-      "grad_norm": 1.303993582725525,
-      "learning_rate": 8.388746803069055e-06,
-      "loss": 0.0889,
-      "step": 945
-    },
-    {
-      "epoch": 0.16129152569464794,
-      "grad_norm": 0.8701795339584351,
-      "learning_rate": 8.387041773231033e-06,
-      "loss": 0.1036,
-      "step": 946
-    },
-    {
-      "epoch": 0.16146202413618563,
-      "grad_norm": 0.8625507354736328,
-      "learning_rate": 8.38533674339301e-06,
-      "loss": 0.1888,
-      "step": 947
-    },
-    {
-      "epoch": 0.16163252257772331,
-      "grad_norm": 0.8963658809661865,
-      "learning_rate": 8.383631713554989e-06,
-      "loss": 0.0969,
-      "step": 948
-    },
-    {
-      "epoch": 0.161803021019261,
-      "grad_norm": 0.7759039998054504,
-      "learning_rate": 8.381926683716965e-06,
-      "loss": 0.098,
-      "step": 949
-    },
-    {
-      "epoch": 0.1619735194607987,
-      "grad_norm": 0.7342348694801331,
-      "learning_rate": 8.380221653878944e-06,
-      "loss": 0.1258,
-      "step": 950
-    },
-    {
-      "epoch": 0.16214401790233637,
-      "grad_norm": 0.6597558856010437,
-      "learning_rate": 8.37851662404092e-06,
-      "loss": 0.1128,
-      "step": 951
-    },
-    {
-      "epoch": 0.16231451634387403,
-      "grad_norm": 1.2019399404525757,
-      "learning_rate": 8.376811594202899e-06,
-      "loss": 0.2601,
-      "step": 952
-    },
-    {
-      "epoch": 0.16248501478541172,
-      "grad_norm": 0.915368914604187,
-      "learning_rate": 8.375106564364877e-06,
-      "loss": 0.121,
-      "step": 953
-    },
-    {
-      "epoch": 0.1626555132269494,
-      "grad_norm": 0.9407950043678284,
-      "learning_rate": 8.373401534526855e-06,
-      "loss": 0.143,
-      "step": 954
-    },
-    {
-      "epoch": 0.1628260116684871,
-      "grad_norm": 0.8434804081916809,
-      "learning_rate": 8.371696504688832e-06,
-      "loss": 0.1648,
-      "step": 955
-    },
-    {
-      "epoch": 0.16299651011002478,
-      "grad_norm": 1.3952746391296387,
-      "learning_rate": 8.36999147485081e-06,
-      "loss": 0.1728,
-      "step": 956
-    },
-    {
-      "epoch": 0.16316700855156246,
-      "grad_norm": 0.6356755495071411,
-      "learning_rate": 8.368286445012788e-06,
-      "loss": 0.1485,
-      "step": 957
-    },
-    {
-      "epoch": 0.16333750699310015,
-      "grad_norm": 0.6824244260787964,
-      "learning_rate": 8.366581415174766e-06,
-      "loss": 0.1239,
-      "step": 958
-    },
-    {
-      "epoch": 0.16350800543463784,
-      "grad_norm": 0.8188114166259766,
-      "learning_rate": 8.364876385336744e-06,
-      "loss": 0.1449,
-      "step": 959
-    },
-    {
-      "epoch": 0.1636785038761755,
-      "grad_norm": 0.781184196472168,
-      "learning_rate": 8.363171355498722e-06,
-      "loss": 0.1508,
-      "step": 960
-    },
-    {
-      "epoch": 0.16384900231771318,
-      "grad_norm": 0.7655654549598694,
-      "learning_rate": 8.3614663256607e-06,
-      "loss": 0.0807,
-      "step": 961
-    },
-    {
-      "epoch": 0.16401950075925087,
-      "grad_norm": 1.431767463684082,
-      "learning_rate": 8.359761295822676e-06,
-      "loss": 0.187,
-      "step": 962
-    },
-    {
-      "epoch": 0.16418999920078856,
-      "grad_norm": 1.4487807750701904,
-      "learning_rate": 8.358056265984656e-06,
-      "loss": 0.2303,
-      "step": 963
-    },
-    {
-      "epoch": 0.16436049764232624,
-      "grad_norm": 0.8229055404663086,
-      "learning_rate": 8.356351236146634e-06,
-      "loss": 0.1904,
-      "step": 964
-    },
-    {
-      "epoch": 0.16453099608386393,
-      "grad_norm": 1.2401492595672607,
-      "learning_rate": 8.35464620630861e-06,
-      "loss": 0.0672,
-      "step": 965
-    },
-    {
-      "epoch": 0.16470149452540161,
-      "grad_norm": 0.9173046350479126,
-      "learning_rate": 8.35294117647059e-06,
-      "loss": 0.1383,
-      "step": 966
-    },
-    {
-      "epoch": 0.16487199296693927,
-      "grad_norm": 0.7897533774375916,
-      "learning_rate": 8.351236146632566e-06,
-      "loss": 0.183,
-      "step": 967
-    },
-    {
-      "epoch": 0.16504249140847696,
-      "grad_norm": 0.8701984882354736,
-      "learning_rate": 8.349531116794544e-06,
-      "loss": 0.156,
-      "step": 968
-    },
-    {
-      "epoch": 0.16521298985001465,
-      "grad_norm": 0.7386003732681274,
-      "learning_rate": 8.347826086956522e-06,
-      "loss": 0.1993,
-      "step": 969
-    },
-    {
-      "epoch": 0.16538348829155233,
-      "grad_norm": 0.9146957993507385,
-      "learning_rate": 8.3461210571185e-06,
-      "loss": 0.1883,
-      "step": 970
-    },
-    {
-      "epoch": 0.16555398673309002,
-      "grad_norm": 1.4515008926391602,
-      "learning_rate": 8.344416027280478e-06,
-      "loss": 0.2608,
-      "step": 971
-    },
-    {
-      "epoch": 0.1657244851746277,
-      "grad_norm": 0.6535465121269226,
-      "learning_rate": 8.342710997442456e-06,
-      "loss": 0.1371,
-      "step": 972
-    },
-    {
-      "epoch": 0.1658949836161654,
-      "grad_norm": 0.9747045040130615,
-      "learning_rate": 8.341005967604434e-06,
-      "loss": 0.2147,
-      "step": 973
-    },
-    {
-      "epoch": 0.16606548205770308,
-      "grad_norm": 0.5989007353782654,
-      "learning_rate": 8.339300937766412e-06,
-      "loss": 0.1839,
-      "step": 974
-    },
-    {
-      "epoch": 0.16623598049924074,
-      "grad_norm": 0.6900016069412231,
-      "learning_rate": 8.33759590792839e-06,
-      "loss": 0.1518,
-      "step": 975
-    },
-    {
-      "epoch": 0.16640647894077842,
-      "grad_norm": 0.7895710468292236,
-      "learning_rate": 8.335890878090368e-06,
-      "loss": 0.1331,
-      "step": 976
-    },
-    {
-      "epoch": 0.1665769773823161,
-      "grad_norm": 0.867023229598999,
-      "learning_rate": 8.334185848252346e-06,
-      "loss": 0.2256,
-      "step": 977
-    },
-    {
-      "epoch": 0.1667474758238538,
-      "grad_norm": 0.797188401222229,
-      "learning_rate": 8.332480818414322e-06,
-      "loss": 0.0746,
-      "step": 978
-    },
-    {
-      "epoch": 0.16691797426539148,
-      "grad_norm": 0.8969321846961975,
-      "learning_rate": 8.330775788576302e-06,
-      "loss": 0.1294,
-      "step": 979
-    },
-    {
-      "epoch": 0.16708847270692917,
-      "grad_norm": 0.5373475551605225,
-      "learning_rate": 8.32907075873828e-06,
-      "loss": 0.0832,
-      "step": 980
-    },
-    {
-      "epoch": 0.16725897114846686,
-      "grad_norm": 0.7001715302467346,
-      "learning_rate": 8.327365728900256e-06,
-      "loss": 0.0546,
-      "step": 981
-    },
-    {
-      "epoch": 0.16742946959000454,
-      "grad_norm": 0.8702039122581482,
-      "learning_rate": 8.325660699062235e-06,
-      "loss": 0.1504,
-      "step": 982
-    },
-    {
-      "epoch": 0.1675999680315422,
-      "grad_norm": 0.7910013794898987,
-      "learning_rate": 8.323955669224212e-06,
-      "loss": 0.1656,
-      "step": 983
-    },
-    {
-      "epoch": 0.1677704664730799,
-      "grad_norm": 0.6668055653572083,
-      "learning_rate": 8.32225063938619e-06,
-      "loss": 0.1416,
-      "step": 984
-    },
-    {
-      "epoch": 0.16794096491461757,
-      "grad_norm": 0.9219913482666016,
-      "learning_rate": 8.320545609548167e-06,
-      "loss": 0.1858,
-      "step": 985
-    },
-    {
-      "epoch": 0.16811146335615526,
-      "grad_norm": 1.0230827331542969,
-      "learning_rate": 8.318840579710145e-06,
-      "loss": 0.1463,
-      "step": 986
-    },
-    {
-      "epoch": 0.16828196179769295,
-      "grad_norm": 0.9198185801506042,
-      "learning_rate": 8.317135549872123e-06,
-      "loss": 0.1762,
-      "step": 987
-    },
-    {
-      "epoch": 0.16845246023923063,
-      "grad_norm": 0.6115066409111023,
-      "learning_rate": 8.315430520034101e-06,
-      "loss": 0.1259,
-      "step": 988
-    },
-    {
-      "epoch": 0.16862295868076832,
-      "grad_norm": 0.7364334464073181,
-      "learning_rate": 8.31372549019608e-06,
-      "loss": 0.2297,
-      "step": 989
-    },
-    {
-      "epoch": 0.16879345712230598,
-      "grad_norm": 0.6792196035385132,
-      "learning_rate": 8.312020460358057e-06,
-      "loss": 0.1877,
-      "step": 990
-    },
-    {
-      "epoch": 0.16896395556384367,
-      "grad_norm": 0.6307271718978882,
-      "learning_rate": 8.310315430520035e-06,
-      "loss": 0.112,
-      "step": 991
-    },
-    {
-      "epoch": 0.16913445400538135,
-      "grad_norm": 0.7789974212646484,
-      "learning_rate": 8.308610400682013e-06,
-      "loss": 0.1531,
-      "step": 992
-    },
-    {
-      "epoch": 0.16930495244691904,
-      "grad_norm": 0.7418656945228577,
-      "learning_rate": 8.306905370843991e-06,
-      "loss": 0.0761,
-      "step": 993
-    },
-    {
-      "epoch": 0.16947545088845672,
-      "grad_norm": 0.6087162494659424,
-      "learning_rate": 8.305200341005967e-06,
-      "loss": 0.1508,
-      "step": 994
-    },
-    {
-      "epoch": 0.1696459493299944,
-      "grad_norm": 0.7893383502960205,
-      "learning_rate": 8.303495311167947e-06,
-      "loss": 0.159,
-      "step": 995
-    },
-    {
-      "epoch": 0.1698164477715321,
-      "grad_norm": 0.6431482434272766,
-      "learning_rate": 8.301790281329925e-06,
-      "loss": 0.1404,
-      "step": 996
-    },
-    {
-      "epoch": 0.16998694621306978,
-      "grad_norm": 0.6812745928764343,
-      "learning_rate": 8.300085251491901e-06,
-      "loss": 0.1582,
-      "step": 997
-    },
-    {
-      "epoch": 0.17015744465460744,
-      "grad_norm": 0.7111532092094421,
-      "learning_rate": 8.29838022165388e-06,
-      "loss": 0.1712,
-      "step": 998
-    },
-    {
-      "epoch": 0.17032794309614513,
-      "grad_norm": 0.8176578283309937,
-      "learning_rate": 8.296675191815857e-06,
-      "loss": 0.169,
-      "step": 999
-    },
-    {
-      "epoch": 0.17049844153768282,
-      "grad_norm": 0.8261346220970154,
-      "learning_rate": 8.294970161977835e-06,
-      "loss": 0.1914,
-      "step": 1000
-    },
-    {
-      "epoch": 0.17049844153768282,
-      "eval_f1_score": 0.18012422360248448,
-      "eval_loss": 0.16470393538475037,
-      "eval_runtime": 182.6084,
-      "eval_samples_per_second": 54.762,
-      "eval_steps_per_second": 3.423,
-      "step": 1000
-    },
-    {
-      "epoch": 0.1706689399792205,
-      "grad_norm": 0.8544616103172302,
-      "learning_rate": 8.293265132139813e-06,
-      "loss": 0.1978,
-      "step": 1001
-    },
-    {
-      "epoch": 0.1708394384207582,
-      "grad_norm": 0.7983449101448059,
-      "learning_rate": 8.291560102301791e-06,
-      "loss": 0.1464,
-      "step": 1002
-    },
-    {
-      "epoch": 0.17100993686229587,
-      "grad_norm": 0.998521625995636,
-      "learning_rate": 8.289855072463769e-06,
-      "loss": 0.1008,
-      "step": 1003
-    },
-    {
-      "epoch": 0.17118043530383356,
-      "grad_norm": 0.9781310558319092,
-      "learning_rate": 8.288150042625747e-06,
-      "loss": 0.2678,
-      "step": 1004
-    },
-    {
-      "epoch": 0.17135093374537122,
-      "grad_norm": 0.8729251623153687,
-      "learning_rate": 8.286445012787725e-06,
-      "loss": 0.144,
-      "step": 1005
-    },
-    {
-      "epoch": 0.1715214321869089,
-      "grad_norm": 0.8118807673454285,
-      "learning_rate": 8.284739982949703e-06,
-      "loss": 0.1468,
-      "step": 1006
-    },
-    {
-      "epoch": 0.1716919306284466,
-      "grad_norm": 0.9156134128570557,
-      "learning_rate": 8.28303495311168e-06,
-      "loss": 0.2034,
-      "step": 1007
-    },
-    {
-      "epoch": 0.17186242906998428,
-      "grad_norm": 0.6133968234062195,
-      "learning_rate": 8.281329923273657e-06,
-      "loss": 0.1561,
-      "step": 1008
-    },
-    {
-      "epoch": 0.17203292751152197,
-      "grad_norm": 0.9954972863197327,
-      "learning_rate": 8.279624893435637e-06,
-      "loss": 0.1995,
-      "step": 1009
-    },
-    {
-      "epoch": 0.17220342595305965,
-      "grad_norm": 0.7595720887184143,
-      "learning_rate": 8.277919863597613e-06,
-      "loss": 0.1625,
-      "step": 1010
-    },
-    {
-      "epoch": 0.17237392439459734,
-      "grad_norm": 0.7384895086288452,
-      "learning_rate": 8.27621483375959e-06,
-      "loss": 0.1656,
-      "step": 1011
-    },
-    {
-      "epoch": 0.17254442283613503,
-      "grad_norm": 1.3423556089401245,
-      "learning_rate": 8.274509803921569e-06,
-      "loss": 0.1169,
-      "step": 1012
-    },
-    {
-      "epoch": 0.17271492127767268,
-      "grad_norm": 0.6995768547058105,
-      "learning_rate": 8.272804774083547e-06,
-      "loss": 0.1639,
-      "step": 1013
-    },
-    {
-      "epoch": 0.17288541971921037,
-      "grad_norm": 1.036745548248291,
-      "learning_rate": 8.271099744245525e-06,
-      "loss": 0.2056,
-      "step": 1014
-    },
-    {
-      "epoch": 0.17305591816074806,
-      "grad_norm": 0.8006750345230103,
-      "learning_rate": 8.269394714407503e-06,
-      "loss": 0.1981,
-      "step": 1015
-    },
-    {
-      "epoch": 0.17322641660228574,
-      "grad_norm": 1.0786309242248535,
-      "learning_rate": 8.26768968456948e-06,
-      "loss": 0.2123,
-      "step": 1016
-    },
-    {
-      "epoch": 0.17339691504382343,
-      "grad_norm": 0.6563388705253601,
-      "learning_rate": 8.265984654731458e-06,
-      "loss": 0.121,
-      "step": 1017
-    },
-    {
-      "epoch": 0.17356741348536112,
-      "grad_norm": 0.6930649876594543,
-      "learning_rate": 8.264279624893436e-06,
-      "loss": 0.1634,
-      "step": 1018
-    },
-    {
-      "epoch": 0.1737379119268988,
-      "grad_norm": 0.7806385159492493,
-      "learning_rate": 8.262574595055414e-06,
-      "loss": 0.0994,
-      "step": 1019
-    },
-    {
-      "epoch": 0.1739084103684365,
-      "grad_norm": 0.6290420889854431,
-      "learning_rate": 8.260869565217392e-06,
-      "loss": 0.1625,
-      "step": 1020
-    },
-    {
-      "epoch": 0.17407890880997415,
-      "grad_norm": 0.5905277132987976,
-      "learning_rate": 8.25916453537937e-06,
-      "loss": 0.1627,
-      "step": 1021
-    },
-    {
-      "epoch": 0.17424940725151183,
-      "grad_norm": 0.7448222637176514,
-      "learning_rate": 8.257459505541348e-06,
-      "loss": 0.1767,
-      "step": 1022
-    },
-    {
-      "epoch": 0.17441990569304952,
-      "grad_norm": 0.7371039390563965,
-      "learning_rate": 8.255754475703326e-06,
-      "loss": 0.0966,
-      "step": 1023
-    },
-    {
-      "epoch": 0.1745904041345872,
-      "grad_norm": 0.8105472326278687,
-      "learning_rate": 8.254049445865302e-06,
-      "loss": 0.1085,
-      "step": 1024
-    },
-    {
-      "epoch": 0.1747609025761249,
-      "grad_norm": 0.6953812837600708,
-      "learning_rate": 8.252344416027282e-06,
-      "loss": 0.1188,
-      "step": 1025
-    },
-    {
-      "epoch": 0.17493140101766258,
-      "grad_norm": 0.595607340335846,
-      "learning_rate": 8.250639386189258e-06,
-      "loss": 0.1055,
-      "step": 1026
-    },
-    {
-      "epoch": 0.17510189945920027,
-      "grad_norm": 1.1801515817642212,
-      "learning_rate": 8.248934356351236e-06,
-      "loss": 0.2072,
-      "step": 1027
-    },
-    {
-      "epoch": 0.17527239790073793,
-      "grad_norm": 0.7170021533966064,
-      "learning_rate": 8.247229326513214e-06,
-      "loss": 0.1377,
-      "step": 1028
-    },
-    {
-      "epoch": 0.1754428963422756,
-      "grad_norm": 0.524588406085968,
-      "learning_rate": 8.245524296675192e-06,
-      "loss": 0.0811,
-      "step": 1029
-    },
-    {
-      "epoch": 0.1756133947838133,
-      "grad_norm": 1.3629536628723145,
-      "learning_rate": 8.24381926683717e-06,
-      "loss": 0.1628,
-      "step": 1030
-    },
-    {
-      "epoch": 0.17578389322535098,
-      "grad_norm": 1.0692698955535889,
-      "learning_rate": 8.242114236999148e-06,
-      "loss": 0.1084,
-      "step": 1031
-    },
-    {
-      "epoch": 0.17595439166688867,
-      "grad_norm": 0.9809920787811279,
-      "learning_rate": 8.240409207161126e-06,
-      "loss": 0.168,
-      "step": 1032
-    },
-    {
-      "epoch": 0.17612489010842636,
-      "grad_norm": 0.5582424998283386,
-      "learning_rate": 8.238704177323104e-06,
-      "loss": 0.1203,
-      "step": 1033
-    },
-    {
-      "epoch": 0.17629538854996404,
-      "grad_norm": 0.6211906671524048,
-      "learning_rate": 8.236999147485082e-06,
-      "loss": 0.1139,
-      "step": 1034
-    },
-    {
-      "epoch": 0.17646588699150173,
-      "grad_norm": 0.8978925347328186,
-      "learning_rate": 8.23529411764706e-06,
-      "loss": 0.0401,
-      "step": 1035
-    },
-    {
-      "epoch": 0.1766363854330394,
-      "grad_norm": 0.7331124544143677,
-      "learning_rate": 8.233589087809038e-06,
-      "loss": 0.149,
-      "step": 1036
-    },
-    {
-      "epoch": 0.17680688387457708,
-      "grad_norm": 0.9506707787513733,
-      "learning_rate": 8.231884057971014e-06,
-      "loss": 0.1644,
-      "step": 1037
-    },
-    {
-      "epoch": 0.17697738231611476,
-      "grad_norm": 0.7511488199234009,
-      "learning_rate": 8.230179028132994e-06,
-      "loss": 0.1275,
-      "step": 1038
-    },
-    {
-      "epoch": 0.17714788075765245,
-      "grad_norm": 0.7483866214752197,
-      "learning_rate": 8.228473998294972e-06,
-      "loss": 0.1603,
-      "step": 1039
-    },
-    {
-      "epoch": 0.17731837919919013,
-      "grad_norm": 0.7479255795478821,
-      "learning_rate": 8.226768968456948e-06,
-      "loss": 0.1182,
-      "step": 1040
-    },
-    {
-      "epoch": 0.17748887764072782,
-      "grad_norm": 0.6701338291168213,
-      "learning_rate": 8.225063938618927e-06,
-      "loss": 0.114,
-      "step": 1041
-    },
-    {
-      "epoch": 0.1776593760822655,
-      "grad_norm": 1.5791823863983154,
-      "learning_rate": 8.223358908780904e-06,
-      "loss": 0.2531,
-      "step": 1042
-    },
-    {
-      "epoch": 0.1778298745238032,
-      "grad_norm": 0.8722712397575378,
-      "learning_rate": 8.221653878942882e-06,
-      "loss": 0.163,
-      "step": 1043
-    },
-    {
-      "epoch": 0.17800037296534085,
-      "grad_norm": 0.8049382567405701,
-      "learning_rate": 8.21994884910486e-06,
-      "loss": 0.1425,
-      "step": 1044
-    },
-    {
-      "epoch": 0.17817087140687854,
-      "grad_norm": 0.8700260519981384,
-      "learning_rate": 8.218243819266838e-06,
-      "loss": 0.1701,
-      "step": 1045
-    },
-    {
-      "epoch": 0.17834136984841623,
-      "grad_norm": 0.7408254146575928,
-      "learning_rate": 8.216538789428816e-06,
-      "loss": 0.1226,
-      "step": 1046
-    },
-    {
-      "epoch": 0.1785118682899539,
-      "grad_norm": 1.1590758562088013,
-      "learning_rate": 8.214833759590793e-06,
-      "loss": 0.2347,
-      "step": 1047
-    },
-    {
-      "epoch": 0.1786823667314916,
-      "grad_norm": 0.891322135925293,
-      "learning_rate": 8.213128729752771e-06,
-      "loss": 0.1725,
-      "step": 1048
-    },
-    {
-      "epoch": 0.17885286517302929,
-      "grad_norm": 0.9721522927284241,
-      "learning_rate": 8.21142369991475e-06,
-      "loss": 0.1578,
-      "step": 1049
-    },
-    {
-      "epoch": 0.17902336361456697,
-      "grad_norm": 1.037187933921814,
-      "learning_rate": 8.209718670076727e-06,
-      "loss": 0.1349,
-      "step": 1050
-    },
-    {
-      "epoch": 0.17919386205610463,
-      "grad_norm": 0.9142031073570251,
-      "learning_rate": 8.208013640238705e-06,
-      "loss": 0.1507,
-      "step": 1051
-    },
-    {
-      "epoch": 0.17936436049764232,
-      "grad_norm": 0.7952942848205566,
-      "learning_rate": 8.206308610400683e-06,
-      "loss": 0.0867,
-      "step": 1052
-    },
-    {
-      "epoch": 0.17953485893918,
-      "grad_norm": 1.0088366270065308,
-      "learning_rate": 8.20460358056266e-06,
-      "loss": 0.235,
-      "step": 1053
-    },
-    {
-      "epoch": 0.1797053573807177,
-      "grad_norm": 1.3762648105621338,
-      "learning_rate": 8.202898550724639e-06,
-      "loss": 0.1695,
-      "step": 1054
-    },
-    {
-      "epoch": 0.17987585582225538,
-      "grad_norm": 0.5625283122062683,
-      "learning_rate": 8.201193520886617e-06,
-      "loss": 0.0914,
-      "step": 1055
-    },
-    {
-      "epoch": 0.18004635426379306,
-      "grad_norm": 1.4387606382369995,
-      "learning_rate": 8.199488491048593e-06,
-      "loss": 0.162,
-      "step": 1056
-    },
-    {
-      "epoch": 0.18021685270533075,
-      "grad_norm": 0.8674620389938354,
-      "learning_rate": 8.197783461210573e-06,
-      "loss": 0.1298,
-      "step": 1057
-    },
-    {
-      "epoch": 0.18038735114686844,
-      "grad_norm": 0.7806848287582397,
-      "learning_rate": 8.19607843137255e-06,
-      "loss": 0.1555,
-      "step": 1058
-    },
-    {
-      "epoch": 0.1805578495884061,
-      "grad_norm": 0.9684147834777832,
-      "learning_rate": 8.194373401534527e-06,
-      "loss": 0.1036,
-      "step": 1059
-    },
-    {
-      "epoch": 0.18072834802994378,
-      "grad_norm": 0.6939026117324829,
-      "learning_rate": 8.192668371696505e-06,
-      "loss": 0.1149,
-      "step": 1060
-    },
-    {
-      "epoch": 0.18089884647148147,
-      "grad_norm": 1.6228690147399902,
-      "learning_rate": 8.190963341858483e-06,
-      "loss": 0.2124,
-      "step": 1061
-    },
-    {
-      "epoch": 0.18106934491301915,
-      "grad_norm": 0.9765108227729797,
-      "learning_rate": 8.189258312020461e-06,
-      "loss": 0.1768,
-      "step": 1062
-    },
-    {
-      "epoch": 0.18123984335455684,
-      "grad_norm": 0.7821102142333984,
-      "learning_rate": 8.187553282182439e-06,
-      "loss": 0.1132,
-      "step": 1063
-    },
-    {
-      "epoch": 0.18141034179609453,
-      "grad_norm": 1.0715186595916748,
-      "learning_rate": 8.185848252344417e-06,
-      "loss": 0.1132,
-      "step": 1064
-    },
-    {
-      "epoch": 0.1815808402376322,
-      "grad_norm": 1.13173508644104,
-      "learning_rate": 8.184143222506395e-06,
-      "loss": 0.0617,
-      "step": 1065
-    },
-    {
-      "epoch": 0.1817513386791699,
-      "grad_norm": 0.7820847630500793,
-      "learning_rate": 8.182438192668373e-06,
-      "loss": 0.1588,
-      "step": 1066
-    },
-    {
-      "epoch": 0.18192183712070756,
-      "grad_norm": 0.8238593935966492,
-      "learning_rate": 8.18073316283035e-06,
-      "loss": 0.1767,
-      "step": 1067
-    },
-    {
-      "epoch": 0.18209233556224524,
-      "grad_norm": 1.0191346406936646,
-      "learning_rate": 8.179028132992329e-06,
-      "loss": 0.2245,
-      "step": 1068
-    },
-    {
-      "epoch": 0.18226283400378293,
-      "grad_norm": 1.069532871246338,
-      "learning_rate": 8.177323103154305e-06,
-      "loss": 0.0601,
-      "step": 1069
-    },
-    {
-      "epoch": 0.18243333244532062,
-      "grad_norm": 1.0120583772659302,
-      "learning_rate": 8.175618073316285e-06,
-      "loss": 0.1654,
-      "step": 1070
-    },
-    {
-      "epoch": 0.1826038308868583,
-      "grad_norm": 2.169363021850586,
-      "learning_rate": 8.173913043478263e-06,
-      "loss": 0.2481,
-      "step": 1071
-    },
-    {
-      "epoch": 0.182774329328396,
-      "grad_norm": 1.7898081541061401,
-      "learning_rate": 8.172208013640239e-06,
-      "loss": 0.2461,
-      "step": 1072
-    },
-    {
-      "epoch": 0.18294482776993368,
-      "grad_norm": 0.9205228090286255,
-      "learning_rate": 8.170502983802218e-06,
-      "loss": 0.1533,
-      "step": 1073
-    },
-    {
-      "epoch": 0.18311532621147134,
-      "grad_norm": 0.8017996549606323,
-      "learning_rate": 8.168797953964195e-06,
-      "loss": 0.1057,
-      "step": 1074
-    },
-    {
-      "epoch": 0.18328582465300902,
-      "grad_norm": 0.9081313610076904,
-      "learning_rate": 8.167092924126173e-06,
-      "loss": 0.1907,
-      "step": 1075
-    },
-    {
-      "epoch": 0.1834563230945467,
-      "grad_norm": 1.5040651559829712,
-      "learning_rate": 8.16538789428815e-06,
-      "loss": 0.2621,
-      "step": 1076
-    },
-    {
-      "epoch": 0.1836268215360844,
-      "grad_norm": 0.8921488523483276,
-      "learning_rate": 8.163682864450129e-06,
-      "loss": 0.1452,
-      "step": 1077
-    },
-    {
-      "epoch": 0.18379731997762208,
-      "grad_norm": 1.1706310510635376,
-      "learning_rate": 8.161977834612106e-06,
-      "loss": 0.0867,
-      "step": 1078
-    },
-    {
-      "epoch": 0.18396781841915977,
-      "grad_norm": 1.15962553024292,
-      "learning_rate": 8.160272804774084e-06,
-      "loss": 0.2057,
-      "step": 1079
-    },
-    {
-      "epoch": 0.18413831686069745,
-      "grad_norm": 1.0124270915985107,
-      "learning_rate": 8.158567774936062e-06,
-      "loss": 0.1282,
-      "step": 1080
-    },
-    {
-      "epoch": 0.18430881530223514,
-      "grad_norm": 0.5859152674674988,
-      "learning_rate": 8.15686274509804e-06,
-      "loss": 0.1282,
-      "step": 1081
-    },
-    {
-      "epoch": 0.1844793137437728,
-      "grad_norm": 0.691265344619751,
-      "learning_rate": 8.155157715260018e-06,
-      "loss": 0.1481,
-      "step": 1082
-    },
-    {
-      "epoch": 0.18464981218531049,
-      "grad_norm": 0.9174695014953613,
-      "learning_rate": 8.153452685421995e-06,
-      "loss": 0.1903,
-      "step": 1083
-    },
-    {
-      "epoch": 0.18482031062684817,
-      "grad_norm": 1.0907355546951294,
-      "learning_rate": 8.151747655583974e-06,
-      "loss": 0.1909,
-      "step": 1084
-    },
-    {
-      "epoch": 0.18499080906838586,
-      "grad_norm": 0.8020131587982178,
-      "learning_rate": 8.15004262574595e-06,
-      "loss": 0.1207,
-      "step": 1085
-    },
-    {
-      "epoch": 0.18516130750992355,
-      "grad_norm": 0.6175723075866699,
-      "learning_rate": 8.148337595907928e-06,
-      "loss": 0.1203,
-      "step": 1086
-    },
-    {
-      "epoch": 0.18533180595146123,
-      "grad_norm": 1.7859039306640625,
-      "learning_rate": 8.146632566069906e-06,
-      "loss": 0.2715,
-      "step": 1087
-    },
-    {
-      "epoch": 0.18550230439299892,
-      "grad_norm": 0.9180811643600464,
-      "learning_rate": 8.144927536231884e-06,
-      "loss": 0.1733,
-      "step": 1088
-    },
-    {
-      "epoch": 0.1856728028345366,
-      "grad_norm": 0.9217977523803711,
-      "learning_rate": 8.143222506393862e-06,
-      "loss": 0.151,
-      "step": 1089
-    },
-    {
-      "epoch": 0.18584330127607426,
-      "grad_norm": 0.5744757652282715,
-      "learning_rate": 8.14151747655584e-06,
-      "loss": 0.1393,
-      "step": 1090
-    },
-    {
-      "epoch": 0.18601379971761195,
-      "grad_norm": 1.082798957824707,
-      "learning_rate": 8.139812446717818e-06,
-      "loss": 0.0939,
-      "step": 1091
-    },
-    {
-      "epoch": 0.18618429815914964,
-      "grad_norm": 0.8758176565170288,
-      "learning_rate": 8.138107416879796e-06,
-      "loss": 0.1736,
-      "step": 1092
-    },
-    {
-      "epoch": 0.18635479660068732,
-      "grad_norm": 0.8503788113594055,
-      "learning_rate": 8.136402387041774e-06,
-      "loss": 0.1174,
-      "step": 1093
-    },
-    {
-      "epoch": 0.186525295042225,
-      "grad_norm": 0.8520966172218323,
-      "learning_rate": 8.134697357203752e-06,
-      "loss": 0.1138,
-      "step": 1094
-    },
-    {
-      "epoch": 0.1866957934837627,
-      "grad_norm": 0.6943926215171814,
-      "learning_rate": 8.13299232736573e-06,
-      "loss": 0.1266,
-      "step": 1095
-    },
-    {
-      "epoch": 0.18686629192530038,
-      "grad_norm": 1.430769920349121,
-      "learning_rate": 8.131287297527706e-06,
-      "loss": 0.2663,
-      "step": 1096
-    },
-    {
-      "epoch": 0.18703679036683804,
-      "grad_norm": 0.6373170018196106,
-      "learning_rate": 8.129582267689686e-06,
-      "loss": 0.1744,
-      "step": 1097
-    },
-    {
-      "epoch": 0.18720728880837573,
-      "grad_norm": 1.0063538551330566,
-      "learning_rate": 8.127877237851664e-06,
-      "loss": 0.185,
-      "step": 1098
-    },
-    {
-      "epoch": 0.1873777872499134,
-      "grad_norm": 0.666248083114624,
-      "learning_rate": 8.12617220801364e-06,
-      "loss": 0.1649,
-      "step": 1099
-    },
-    {
-      "epoch": 0.1875482856914511,
-      "grad_norm": 0.635853111743927,
-      "learning_rate": 8.12446717817562e-06,
-      "loss": 0.1536,
-      "step": 1100
-    },
-    {
-      "epoch": 0.1875482856914511,
-      "eval_f1_score": 0.22028985507246376,
-      "eval_loss": 0.1652476042509079,
-      "eval_runtime": 182.5656,
-      "eval_samples_per_second": 54.775,
-      "eval_steps_per_second": 3.423,
-      "step": 1100
-    },
-    {
-      "epoch": 0.1877187841329888,
-      "grad_norm": 0.6525788307189941,
-      "learning_rate": 8.122762148337596e-06,
-      "loss": 0.1528,
-      "step": 1101
-    },
-    {
-      "epoch": 0.18788928257452647,
-      "grad_norm": 0.8407154083251953,
-      "learning_rate": 8.121057118499574e-06,
-      "loss": 0.107,
-      "step": 1102
-    },
-    {
-      "epoch": 0.18805978101606416,
-      "grad_norm": 1.0600398778915405,
-      "learning_rate": 8.119352088661552e-06,
-      "loss": 0.1299,
-      "step": 1103
-    },
-    {
-      "epoch": 0.18823027945760185,
-      "grad_norm": 0.7606744766235352,
-      "learning_rate": 8.11764705882353e-06,
-      "loss": 0.1808,
-      "step": 1104
-    },
-    {
-      "epoch": 0.1884007778991395,
-      "grad_norm": 0.7155107259750366,
-      "learning_rate": 8.115942028985508e-06,
-      "loss": 0.1413,
-      "step": 1105
-    },
-    {
-      "epoch": 0.1885712763406772,
-      "grad_norm": 1.1344115734100342,
-      "learning_rate": 8.114236999147486e-06,
-      "loss": 0.2271,
-      "step": 1106
-    },
-    {
-      "epoch": 0.18874177478221488,
-      "grad_norm": 0.6760721802711487,
-      "learning_rate": 8.112531969309464e-06,
-      "loss": 0.1453,
-      "step": 1107
-    },
-    {
-      "epoch": 0.18891227322375256,
-      "grad_norm": 0.6441182494163513,
-      "learning_rate": 8.110826939471442e-06,
-      "loss": 0.0994,
-      "step": 1108
-    },
-    {
-      "epoch": 0.18908277166529025,
-      "grad_norm": 0.8685035109519958,
-      "learning_rate": 8.10912190963342e-06,
-      "loss": 0.1631,
-      "step": 1109
-    },
-    {
-      "epoch": 0.18925327010682794,
-      "grad_norm": 1.2088656425476074,
-      "learning_rate": 8.107416879795397e-06,
-      "loss": 0.204,
-      "step": 1110
-    },
-    {
-      "epoch": 0.18942376854836562,
-      "grad_norm": 0.5793415307998657,
-      "learning_rate": 8.105711849957375e-06,
-      "loss": 0.0975,
-      "step": 1111
-    },
-    {
-      "epoch": 0.18959426698990328,
-      "grad_norm": 0.8224422931671143,
-      "learning_rate": 8.104006820119352e-06,
-      "loss": 0.1729,
-      "step": 1112
-    },
-    {
-      "epoch": 0.18976476543144097,
-      "grad_norm": 1.1413917541503906,
-      "learning_rate": 8.102301790281331e-06,
-      "loss": 0.2047,
-      "step": 1113
-    },
-    {
-      "epoch": 0.18993526387297865,
-      "grad_norm": 0.9919366240501404,
-      "learning_rate": 8.10059676044331e-06,
-      "loss": 0.104,
-      "step": 1114
-    },
-    {
-      "epoch": 0.19010576231451634,
-      "grad_norm": 0.8470364809036255,
-      "learning_rate": 8.098891730605285e-06,
-      "loss": 0.2055,
-      "step": 1115
-    },
-    {
-      "epoch": 0.19027626075605403,
-      "grad_norm": 0.8210625052452087,
-      "learning_rate": 8.097186700767265e-06,
-      "loss": 0.1939,
-      "step": 1116
-    },
-    {
-      "epoch": 0.19044675919759171,
-      "grad_norm": 0.8386514186859131,
-      "learning_rate": 8.095481670929241e-06,
-      "loss": 0.1323,
-      "step": 1117
-    },
-    {
-      "epoch": 0.1906172576391294,
-      "grad_norm": 0.8340263962745667,
-      "learning_rate": 8.09377664109122e-06,
-      "loss": 0.1597,
-      "step": 1118
-    },
-    {
-      "epoch": 0.1907877560806671,
-      "grad_norm": 0.7370643615722656,
-      "learning_rate": 8.092071611253197e-06,
-      "loss": 0.1838,
-      "step": 1119
-    },
-    {
-      "epoch": 0.19095825452220475,
-      "grad_norm": 0.8056231737136841,
-      "learning_rate": 8.090366581415175e-06,
-      "loss": 0.1611,
-      "step": 1120
-    },
-    {
-      "epoch": 0.19112875296374243,
-      "grad_norm": 0.7211765050888062,
-      "learning_rate": 8.088661551577153e-06,
-      "loss": 0.1507,
-      "step": 1121
-    },
-    {
-      "epoch": 0.19129925140528012,
-      "grad_norm": 1.055554986000061,
-      "learning_rate": 8.086956521739131e-06,
-      "loss": 0.1991,
-      "step": 1122
-    },
-    {
-      "epoch": 0.1914697498468178,
-      "grad_norm": 0.8282398581504822,
-      "learning_rate": 8.085251491901109e-06,
-      "loss": 0.1819,
-      "step": 1123
-    },
-    {
-      "epoch": 0.1916402482883555,
-      "grad_norm": 0.9935968518257141,
-      "learning_rate": 8.083546462063087e-06,
-      "loss": 0.1149,
-      "step": 1124
-    },
-    {
-      "epoch": 0.19181074672989318,
-      "grad_norm": 0.8828055262565613,
-      "learning_rate": 8.081841432225065e-06,
-      "loss": 0.1656,
-      "step": 1125
-    },
-    {
-      "epoch": 0.19198124517143086,
-      "grad_norm": 0.7493565678596497,
-      "learning_rate": 8.080136402387043e-06,
-      "loss": 0.1505,
-      "step": 1126
-    },
-    {
-      "epoch": 0.19215174361296855,
-      "grad_norm": 0.8418991565704346,
-      "learning_rate": 8.07843137254902e-06,
-      "loss": 0.1045,
-      "step": 1127
-    },
-    {
-      "epoch": 0.1923222420545062,
-      "grad_norm": 0.8248168230056763,
-      "learning_rate": 8.076726342710997e-06,
-      "loss": 0.2225,
-      "step": 1128
-    },
-    {
-      "epoch": 0.1924927404960439,
-      "grad_norm": 0.6494383215904236,
-      "learning_rate": 8.075021312872977e-06,
-      "loss": 0.1073,
-      "step": 1129
-    },
-    {
-      "epoch": 0.19266323893758158,
-      "grad_norm": 0.6220464110374451,
-      "learning_rate": 8.073316283034955e-06,
-      "loss": 0.089,
-      "step": 1130
-    },
-    {
-      "epoch": 0.19283373737911927,
-      "grad_norm": 0.7500420212745667,
-      "learning_rate": 8.071611253196931e-06,
-      "loss": 0.1275,
-      "step": 1131
-    },
-    {
-      "epoch": 0.19300423582065696,
-      "grad_norm": 1.1788630485534668,
-      "learning_rate": 8.06990622335891e-06,
-      "loss": 0.2489,
-      "step": 1132
-    },
-    {
-      "epoch": 0.19317473426219464,
-      "grad_norm": 0.9187591671943665,
-      "learning_rate": 8.068201193520887e-06,
-      "loss": 0.223,
-      "step": 1133
-    },
-    {
-      "epoch": 0.19334523270373233,
-      "grad_norm": 0.5593259930610657,
-      "learning_rate": 8.066496163682865e-06,
-      "loss": 0.1245,
-      "step": 1134
-    },
-    {
-      "epoch": 0.19351573114527,
-      "grad_norm": 0.9649487733840942,
-      "learning_rate": 8.064791133844843e-06,
-      "loss": 0.1953,
-      "step": 1135
-    },
-    {
-      "epoch": 0.19368622958680767,
-      "grad_norm": 0.993848443031311,
-      "learning_rate": 8.06308610400682e-06,
-      "loss": 0.2279,
-      "step": 1136
-    },
-    {
-      "epoch": 0.19385672802834536,
-      "grad_norm": 0.593940794467926,
-      "learning_rate": 8.061381074168799e-06,
-      "loss": 0.1847,
-      "step": 1137
-    },
-    {
-      "epoch": 0.19402722646988305,
-      "grad_norm": 0.6698784828186035,
-      "learning_rate": 8.059676044330777e-06,
-      "loss": 0.12,
-      "step": 1138
-    },
-    {
-      "epoch": 0.19419772491142073,
-      "grad_norm": 0.9543096423149109,
-      "learning_rate": 8.057971014492754e-06,
-      "loss": 0.1031,
-      "step": 1139
-    },
-    {
-      "epoch": 0.19436822335295842,
-      "grad_norm": 0.6808764934539795,
-      "learning_rate": 8.056265984654732e-06,
-      "loss": 0.1851,
-      "step": 1140
-    },
-    {
-      "epoch": 0.1945387217944961,
-      "grad_norm": 0.9641972780227661,
-      "learning_rate": 8.05456095481671e-06,
-      "loss": 0.2253,
-      "step": 1141
-    },
-    {
-      "epoch": 0.1947092202360338,
-      "grad_norm": 0.6276617646217346,
-      "learning_rate": 8.052855924978688e-06,
-      "loss": 0.2123,
-      "step": 1142
-    },
-    {
-      "epoch": 0.19487971867757145,
-      "grad_norm": 0.778846025466919,
-      "learning_rate": 8.051150895140666e-06,
-      "loss": 0.1573,
-      "step": 1143
-    },
-    {
-      "epoch": 0.19505021711910914,
-      "grad_norm": 0.6329131722450256,
-      "learning_rate": 8.049445865302643e-06,
-      "loss": 0.1638,
-      "step": 1144
-    },
-    {
-      "epoch": 0.19522071556064682,
-      "grad_norm": 0.7255607843399048,
-      "learning_rate": 8.047740835464622e-06,
-      "loss": 0.1964,
-      "step": 1145
-    },
-    {
-      "epoch": 0.1953912140021845,
-      "grad_norm": 0.7392814755439758,
-      "learning_rate": 8.046035805626598e-06,
-      "loss": 0.1729,
-      "step": 1146
-    },
-    {
-      "epoch": 0.1955617124437222,
-      "grad_norm": 0.5993164777755737,
-      "learning_rate": 8.044330775788576e-06,
-      "loss": 0.1532,
-      "step": 1147
-    },
-    {
-      "epoch": 0.19573221088525988,
-      "grad_norm": 0.6985999345779419,
-      "learning_rate": 8.042625745950556e-06,
-      "loss": 0.1615,
-      "step": 1148
-    },
-    {
-      "epoch": 0.19590270932679757,
-      "grad_norm": 0.7796421051025391,
-      "learning_rate": 8.040920716112532e-06,
-      "loss": 0.0907,
-      "step": 1149
-    },
-    {
-      "epoch": 0.19607320776833526,
-      "grad_norm": 0.5680537819862366,
-      "learning_rate": 8.03921568627451e-06,
-      "loss": 0.1613,
-      "step": 1150
-    },
-    {
-      "epoch": 0.19624370620987291,
-      "grad_norm": 0.5613918900489807,
-      "learning_rate": 8.037510656436488e-06,
-      "loss": 0.1104,
-      "step": 1151
-    },
-    {
-      "epoch": 0.1964142046514106,
-      "grad_norm": 0.660589873790741,
-      "learning_rate": 8.035805626598466e-06,
-      "loss": 0.1217,
-      "step": 1152
-    },
-    {
-      "epoch": 0.1965847030929483,
-      "grad_norm": 0.8582456707954407,
-      "learning_rate": 8.034100596760444e-06,
-      "loss": 0.0883,
-      "step": 1153
-    },
-    {
-      "epoch": 0.19675520153448597,
-      "grad_norm": 1.0223498344421387,
-      "learning_rate": 8.032395566922422e-06,
-      "loss": 0.2005,
-      "step": 1154
-    },
-    {
-      "epoch": 0.19692569997602366,
-      "grad_norm": 0.7808375954627991,
-      "learning_rate": 8.0306905370844e-06,
-      "loss": 0.1288,
-      "step": 1155
-    },
-    {
-      "epoch": 0.19709619841756135,
-      "grad_norm": 0.7909091711044312,
-      "learning_rate": 8.028985507246378e-06,
-      "loss": 0.1117,
-      "step": 1156
-    },
-    {
-      "epoch": 0.19726669685909903,
-      "grad_norm": 1.5872262716293335,
-      "learning_rate": 8.027280477408356e-06,
-      "loss": 0.2244,
-      "step": 1157
-    },
-    {
-      "epoch": 0.1974371953006367,
-      "grad_norm": 0.5757585763931274,
-      "learning_rate": 8.025575447570332e-06,
-      "loss": 0.1102,
-      "step": 1158
-    },
-    {
-      "epoch": 0.19760769374217438,
-      "grad_norm": 0.6899182200431824,
-      "learning_rate": 8.023870417732312e-06,
-      "loss": 0.0771,
-      "step": 1159
-    },
-    {
-      "epoch": 0.19777819218371206,
-      "grad_norm": 1.3838332891464233,
-      "learning_rate": 8.022165387894288e-06,
-      "loss": 0.2105,
-      "step": 1160
-    },
-    {
-      "epoch": 0.19794869062524975,
-      "grad_norm": 1.5885940790176392,
-      "learning_rate": 8.020460358056266e-06,
-      "loss": 0.2264,
-      "step": 1161
-    },
-    {
-      "epoch": 0.19811918906678744,
-      "grad_norm": 1.0133886337280273,
-      "learning_rate": 8.018755328218244e-06,
-      "loss": 0.193,
-      "step": 1162
-    },
-    {
-      "epoch": 0.19828968750832512,
-      "grad_norm": 0.8545687794685364,
-      "learning_rate": 8.017050298380222e-06,
-      "loss": 0.1345,
-      "step": 1163
-    },
-    {
-      "epoch": 0.1984601859498628,
-      "grad_norm": 0.8866453170776367,
-      "learning_rate": 8.0153452685422e-06,
-      "loss": 0.1932,
-      "step": 1164
-    },
-    {
-      "epoch": 0.1986306843914005,
-      "grad_norm": 1.2009869813919067,
-      "learning_rate": 8.013640238704178e-06,
-      "loss": 0.1389,
-      "step": 1165
-    },
-    {
-      "epoch": 0.19880118283293816,
-      "grad_norm": 1.0642001628875732,
-      "learning_rate": 8.011935208866156e-06,
-      "loss": 0.1595,
-      "step": 1166
-    },
-    {
-      "epoch": 0.19897168127447584,
-      "grad_norm": 0.9353564977645874,
-      "learning_rate": 8.010230179028134e-06,
-      "loss": 0.17,
-      "step": 1167
-    },
-    {
-      "epoch": 0.19914217971601353,
-      "grad_norm": 1.202636480331421,
-      "learning_rate": 8.008525149190112e-06,
-      "loss": 0.2489,
-      "step": 1168
-    },
-    {
-      "epoch": 0.19931267815755122,
-      "grad_norm": 0.769627034664154,
-      "learning_rate": 8.00682011935209e-06,
-      "loss": 0.1945,
-      "step": 1169
-    },
-    {
-      "epoch": 0.1994831765990889,
-      "grad_norm": 0.6681427955627441,
-      "learning_rate": 8.005115089514067e-06,
-      "loss": 0.1682,
-      "step": 1170
-    },
-    {
-      "epoch": 0.1996536750406266,
-      "grad_norm": 1.5832921266555786,
-      "learning_rate": 8.003410059676044e-06,
-      "loss": 0.2114,
-      "step": 1171
-    },
-    {
-      "epoch": 0.19982417348216427,
-      "grad_norm": 0.896315336227417,
-      "learning_rate": 8.001705029838023e-06,
-      "loss": 0.084,
-      "step": 1172
-    },
-    {
-      "epoch": 0.19999467192370196,
-      "grad_norm": 1.0023328065872192,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 0.2149,
-      "step": 1173
-    },
-    {
-      "epoch": 0.20016517036523962,
-      "grad_norm": 0.7651743292808533,
-      "learning_rate": 7.998294970161978e-06,
-      "loss": 0.0709,
-      "step": 1174
-    },
-    {
-      "epoch": 0.2003356688067773,
-      "grad_norm": 1.0578312873840332,
-      "learning_rate": 7.996589940323957e-06,
-      "loss": 0.1952,
-      "step": 1175
-    },
-    {
-      "epoch": 0.200506167248315,
-      "grad_norm": 1.1662102937698364,
-      "learning_rate": 7.994884910485933e-06,
-      "loss": 0.227,
-      "step": 1176
-    },
-    {
-      "epoch": 0.20067666568985268,
-      "grad_norm": 0.852552056312561,
-      "learning_rate": 7.993179880647911e-06,
-      "loss": 0.193,
-      "step": 1177
-    },
-    {
-      "epoch": 0.20084716413139037,
-      "grad_norm": 0.8441449999809265,
-      "learning_rate": 7.99147485080989e-06,
-      "loss": 0.1953,
-      "step": 1178
-    },
-    {
-      "epoch": 0.20101766257292805,
-      "grad_norm": 1.0427707433700562,
-      "learning_rate": 7.989769820971867e-06,
-      "loss": 0.0567,
-      "step": 1179
-    },
-    {
-      "epoch": 0.20118816101446574,
-      "grad_norm": 0.5938249230384827,
-      "learning_rate": 7.988064791133845e-06,
-      "loss": 0.137,
-      "step": 1180
-    },
-    {
-      "epoch": 0.2013586594560034,
-      "grad_norm": 0.6687734723091125,
-      "learning_rate": 7.986359761295823e-06,
-      "loss": 0.1103,
-      "step": 1181
-    },
-    {
-      "epoch": 0.20152915789754108,
-      "grad_norm": 0.6688919067382812,
-      "learning_rate": 7.984654731457801e-06,
-      "loss": 0.1135,
-      "step": 1182
-    },
-    {
-      "epoch": 0.20169965633907877,
-      "grad_norm": 0.7219515442848206,
-      "learning_rate": 7.982949701619779e-06,
-      "loss": 0.0858,
-      "step": 1183
-    },
-    {
-      "epoch": 0.20187015478061646,
-      "grad_norm": 0.4615936279296875,
-      "learning_rate": 7.981244671781757e-06,
-      "loss": 0.0989,
-      "step": 1184
-    },
-    {
-      "epoch": 0.20204065322215414,
-      "grad_norm": 0.6289270520210266,
-      "learning_rate": 7.979539641943735e-06,
-      "loss": 0.1416,
-      "step": 1185
-    },
-    {
-      "epoch": 0.20221115166369183,
-      "grad_norm": 0.5620813965797424,
-      "learning_rate": 7.977834612105713e-06,
-      "loss": 0.1293,
-      "step": 1186
-    },
-    {
-      "epoch": 0.20238165010522952,
-      "grad_norm": 0.8259469866752625,
-      "learning_rate": 7.97612958226769e-06,
-      "loss": 0.1645,
-      "step": 1187
-    },
-    {
-      "epoch": 0.2025521485467672,
-      "grad_norm": 0.4567446708679199,
-      "learning_rate": 7.974424552429669e-06,
-      "loss": 0.1148,
-      "step": 1188
-    },
-    {
-      "epoch": 0.20272264698830486,
-      "grad_norm": 0.7620810866355896,
-      "learning_rate": 7.972719522591647e-06,
-      "loss": 0.1775,
-      "step": 1189
-    },
-    {
-      "epoch": 0.20289314542984255,
-      "grad_norm": 1.8133633136749268,
-      "learning_rate": 7.971014492753623e-06,
-      "loss": 0.3183,
-      "step": 1190
-    },
-    {
-      "epoch": 0.20306364387138023,
-      "grad_norm": 0.7991937398910522,
-      "learning_rate": 7.969309462915603e-06,
-      "loss": 0.182,
-      "step": 1191
-    },
-    {
-      "epoch": 0.20323414231291792,
-      "grad_norm": 1.1091400384902954,
-      "learning_rate": 7.967604433077579e-06,
-      "loss": 0.2114,
-      "step": 1192
-    },
-    {
-      "epoch": 0.2034046407544556,
-      "grad_norm": 0.5558500289916992,
-      "learning_rate": 7.965899403239557e-06,
-      "loss": 0.1173,
-      "step": 1193
-    },
-    {
-      "epoch": 0.2035751391959933,
-      "grad_norm": 0.6900790929794312,
-      "learning_rate": 7.964194373401535e-06,
-      "loss": 0.1523,
-      "step": 1194
-    },
-    {
-      "epoch": 0.20374563763753098,
-      "grad_norm": 0.806940495967865,
-      "learning_rate": 7.962489343563513e-06,
-      "loss": 0.2267,
-      "step": 1195
-    },
-    {
-      "epoch": 0.20391613607906867,
-      "grad_norm": 0.8060824871063232,
-      "learning_rate": 7.96078431372549e-06,
-      "loss": 0.1432,
-      "step": 1196
-    },
-    {
-      "epoch": 0.20408663452060632,
-      "grad_norm": 0.842232882976532,
-      "learning_rate": 7.959079283887469e-06,
-      "loss": 0.1044,
-      "step": 1197
-    },
-    {
-      "epoch": 0.204257132962144,
-      "grad_norm": 0.7650852203369141,
-      "learning_rate": 7.957374254049447e-06,
-      "loss": 0.157,
-      "step": 1198
-    },
-    {
-      "epoch": 0.2044276314036817,
-      "grad_norm": 0.7800074219703674,
-      "learning_rate": 7.955669224211425e-06,
-      "loss": 0.1424,
-      "step": 1199
-    },
-    {
-      "epoch": 0.20459812984521938,
-      "grad_norm": 0.8121132850646973,
-      "learning_rate": 7.953964194373403e-06,
-      "loss": 0.1909,
-      "step": 1200
-    },
-    {
-      "epoch": 0.20459812984521938,
-      "eval_f1_score": 0.2485207100591716,
-      "eval_loss": 0.1650407910346985,
-      "eval_runtime": 182.5967,
-      "eval_samples_per_second": 54.765,
-      "eval_steps_per_second": 3.423,
-      "step": 1200
-    },
-    {
-      "epoch": 0.20476862828675707,
-      "grad_norm": 0.9365339875221252,
-      "learning_rate": 7.95225916453538e-06,
-      "loss": 0.1804,
-      "step": 1201
-    },
-    {
-      "epoch": 0.20493912672829476,
-      "grad_norm": 0.4943313002586365,
-      "learning_rate": 7.950554134697358e-06,
-      "loss": 0.1368,
-      "step": 1202
-    },
-    {
-      "epoch": 0.20510962516983244,
-      "grad_norm": 0.6111050248146057,
-      "learning_rate": 7.948849104859335e-06,
-      "loss": 0.1726,
-      "step": 1203
-    },
-    {
-      "epoch": 0.2052801236113701,
-      "grad_norm": 0.8806681036949158,
-      "learning_rate": 7.947144075021314e-06,
-      "loss": 0.1393,
-      "step": 1204
-    },
-    {
-      "epoch": 0.2054506220529078,
-      "grad_norm": 0.8702719807624817,
-      "learning_rate": 7.945439045183292e-06,
-      "loss": 0.2051,
-      "step": 1205
-    },
-    {
-      "epoch": 0.20562112049444548,
-      "grad_norm": 0.5891739130020142,
-      "learning_rate": 7.943734015345269e-06,
-      "loss": 0.1182,
-      "step": 1206
-    },
-    {
-      "epoch": 0.20579161893598316,
-      "grad_norm": 0.7840372920036316,
-      "learning_rate": 7.942028985507248e-06,
-      "loss": 0.176,
-      "step": 1207
-    },
-    {
-      "epoch": 0.20596211737752085,
-      "grad_norm": 0.7101555466651917,
-      "learning_rate": 7.940323955669224e-06,
-      "loss": 0.2072,
-      "step": 1208
-    },
-    {
-      "epoch": 0.20613261581905853,
-      "grad_norm": 0.7602161765098572,
-      "learning_rate": 7.938618925831202e-06,
-      "loss": 0.1198,
-      "step": 1209
-    },
-    {
-      "epoch": 0.20630311426059622,
-      "grad_norm": 0.7662124037742615,
-      "learning_rate": 7.93691389599318e-06,
-      "loss": 0.1544,
-      "step": 1210
-    },
-    {
-      "epoch": 0.2064736127021339,
-      "grad_norm": 0.6383134722709656,
-      "learning_rate": 7.935208866155158e-06,
-      "loss": 0.1298,
-      "step": 1211
-    },
-    {
-      "epoch": 0.20664411114367157,
-      "grad_norm": 0.9750174283981323,
-      "learning_rate": 7.933503836317136e-06,
-      "loss": 0.142,
-      "step": 1212
-    },
-    {
-      "epoch": 0.20681460958520925,
-      "grad_norm": 0.7169644832611084,
-      "learning_rate": 7.931798806479114e-06,
-      "loss": 0.1538,
-      "step": 1213
-    },
-    {
-      "epoch": 0.20698510802674694,
-      "grad_norm": 0.7920701503753662,
-      "learning_rate": 7.930093776641092e-06,
-      "loss": 0.1591,
-      "step": 1214
-    },
-    {
-      "epoch": 0.20715560646828463,
-      "grad_norm": 0.7697597742080688,
-      "learning_rate": 7.92838874680307e-06,
-      "loss": 0.1361,
-      "step": 1215
-    },
-    {
-      "epoch": 0.2073261049098223,
-      "grad_norm": 0.8267337083816528,
-      "learning_rate": 7.926683716965048e-06,
-      "loss": 0.1605,
-      "step": 1216
-    },
-    {
-      "epoch": 0.20749660335136,
-      "grad_norm": 1.0727347135543823,
-      "learning_rate": 7.924978687127026e-06,
-      "loss": 0.2054,
-      "step": 1217
-    },
-    {
-      "epoch": 0.20766710179289768,
-      "grad_norm": 0.9135003089904785,
-      "learning_rate": 7.923273657289004e-06,
-      "loss": 0.1439,
-      "step": 1218
-    },
-    {
-      "epoch": 0.20783760023443534,
-      "grad_norm": 0.9521437287330627,
-      "learning_rate": 7.92156862745098e-06,
-      "loss": 0.1991,
-      "step": 1219
-    },
-    {
-      "epoch": 0.20800809867597303,
-      "grad_norm": 0.8825364708900452,
-      "learning_rate": 7.91986359761296e-06,
-      "loss": 0.1299,
-      "step": 1220
-    },
-    {
-      "epoch": 0.20817859711751072,
-      "grad_norm": 1.088287353515625,
-      "learning_rate": 7.918158567774936e-06,
-      "loss": 0.1093,
-      "step": 1221
-    },
-    {
-      "epoch": 0.2083490955590484,
-      "grad_norm": 1.09437894821167,
-      "learning_rate": 7.916453537936914e-06,
-      "loss": 0.0805,
-      "step": 1222
-    },
-    {
-      "epoch": 0.2085195940005861,
-      "grad_norm": 0.9781740307807922,
-      "learning_rate": 7.914748508098894e-06,
-      "loss": 0.1763,
-      "step": 1223
-    },
-    {
-      "epoch": 0.20869009244212378,
-      "grad_norm": 1.0949674844741821,
-      "learning_rate": 7.91304347826087e-06,
-      "loss": 0.1578,
-      "step": 1224
-    },
-    {
-      "epoch": 0.20886059088366146,
-      "grad_norm": 0.9407842755317688,
-      "learning_rate": 7.911338448422848e-06,
-      "loss": 0.0841,
-      "step": 1225
-    },
-    {
-      "epoch": 0.20903108932519915,
-      "grad_norm": 1.0195931196212769,
-      "learning_rate": 7.909633418584826e-06,
-      "loss": 0.1611,
-      "step": 1226
-    },
-    {
-      "epoch": 0.2092015877667368,
-      "grad_norm": 0.7501747608184814,
-      "learning_rate": 7.907928388746804e-06,
-      "loss": 0.0916,
-      "step": 1227
-    },
-    {
-      "epoch": 0.2093720862082745,
-      "grad_norm": 0.7305780053138733,
-      "learning_rate": 7.906223358908782e-06,
-      "loss": 0.1143,
-      "step": 1228
-    },
-    {
-      "epoch": 0.20954258464981218,
-      "grad_norm": 0.8773810267448425,
-      "learning_rate": 7.90451832907076e-06,
-      "loss": 0.1699,
-      "step": 1229
-    },
-    {
-      "epoch": 0.20971308309134987,
-      "grad_norm": 0.929328441619873,
-      "learning_rate": 7.902813299232738e-06,
-      "loss": 0.1352,
-      "step": 1230
-    },
-    {
-      "epoch": 0.20988358153288755,
-      "grad_norm": 0.6584373712539673,
-      "learning_rate": 7.901108269394716e-06,
-      "loss": 0.057,
-      "step": 1231
-    },
-    {
-      "epoch": 0.21005407997442524,
-      "grad_norm": 0.7133747935295105,
-      "learning_rate": 7.899403239556693e-06,
-      "loss": 0.1141,
-      "step": 1232
-    },
-    {
-      "epoch": 0.21022457841596293,
-      "grad_norm": 0.6574261784553528,
-      "learning_rate": 7.89769820971867e-06,
-      "loss": 0.1312,
-      "step": 1233
-    },
-    {
-      "epoch": 0.2103950768575006,
-      "grad_norm": 0.8198479413986206,
-      "learning_rate": 7.89599317988065e-06,
-      "loss": 0.1476,
-      "step": 1234
-    },
-    {
-      "epoch": 0.21056557529903827,
-      "grad_norm": 1.679710030555725,
-      "learning_rate": 7.894288150042626e-06,
-      "loss": 0.2136,
-      "step": 1235
-    },
-    {
-      "epoch": 0.21073607374057596,
-      "grad_norm": 0.8997982144355774,
-      "learning_rate": 7.892583120204604e-06,
-      "loss": 0.1446,
-      "step": 1236
-    },
-    {
-      "epoch": 0.21090657218211364,
-      "grad_norm": 0.5984455943107605,
-      "learning_rate": 7.890878090366582e-06,
-      "loss": 0.1091,
-      "step": 1237
-    },
-    {
-      "epoch": 0.21107707062365133,
-      "grad_norm": 1.0783398151397705,
-      "learning_rate": 7.88917306052856e-06,
-      "loss": 0.1465,
-      "step": 1238
-    },
-    {
-      "epoch": 0.21124756906518902,
-      "grad_norm": 1.564841389656067,
-      "learning_rate": 7.887468030690537e-06,
-      "loss": 0.284,
-      "step": 1239
-    },
-    {
-      "epoch": 0.2114180675067267,
-      "grad_norm": 0.936279833316803,
-      "learning_rate": 7.885763000852515e-06,
-      "loss": 0.0737,
-      "step": 1240
-    },
-    {
-      "epoch": 0.2115885659482644,
-      "grad_norm": 1.0812268257141113,
-      "learning_rate": 7.884057971014493e-06,
-      "loss": 0.1423,
-      "step": 1241
-    },
-    {
-      "epoch": 0.21175906438980205,
-      "grad_norm": 0.9714178442955017,
-      "learning_rate": 7.882352941176471e-06,
-      "loss": 0.1452,
-      "step": 1242
-    },
-    {
-      "epoch": 0.21192956283133974,
-      "grad_norm": 0.7834417819976807,
-      "learning_rate": 7.88064791133845e-06,
-      "loss": 0.1637,
-      "step": 1243
-    },
-    {
-      "epoch": 0.21210006127287742,
-      "grad_norm": 0.6951526999473572,
-      "learning_rate": 7.878942881500427e-06,
-      "loss": 0.1474,
-      "step": 1244
-    },
-    {
-      "epoch": 0.2122705597144151,
-      "grad_norm": 0.8583109974861145,
-      "learning_rate": 7.877237851662405e-06,
-      "loss": 0.171,
-      "step": 1245
-    },
-    {
-      "epoch": 0.2124410581559528,
-      "grad_norm": 1.11123788356781,
-      "learning_rate": 7.875532821824381e-06,
-      "loss": 0.2114,
-      "step": 1246
-    },
-    {
-      "epoch": 0.21261155659749048,
-      "grad_norm": 0.8634263873100281,
-      "learning_rate": 7.873827791986361e-06,
-      "loss": 0.1309,
-      "step": 1247
-    },
-    {
-      "epoch": 0.21278205503902817,
-      "grad_norm": 1.0174041986465454,
-      "learning_rate": 7.872122762148339e-06,
-      "loss": 0.2006,
-      "step": 1248
-    },
-    {
-      "epoch": 0.21295255348056585,
-      "grad_norm": 0.8859142661094666,
-      "learning_rate": 7.870417732310315e-06,
-      "loss": 0.1345,
-      "step": 1249
-    },
-    {
-      "epoch": 0.2131230519221035,
-      "grad_norm": 0.9608497023582458,
-      "learning_rate": 7.868712702472295e-06,
-      "loss": 0.1496,
-      "step": 1250
-    },
-    {
-      "epoch": 0.2132935503636412,
-      "grad_norm": 0.843275785446167,
-      "learning_rate": 7.867007672634271e-06,
-      "loss": 0.1602,
-      "step": 1251
-    },
-    {
-      "epoch": 0.21346404880517889,
-      "grad_norm": 0.6963731646537781,
-      "learning_rate": 7.865302642796249e-06,
-      "loss": 0.1346,
-      "step": 1252
-    },
-    {
-      "epoch": 0.21363454724671657,
-      "grad_norm": 1.220220685005188,
-      "learning_rate": 7.863597612958227e-06,
-      "loss": 0.1669,
-      "step": 1253
-    },
-    {
-      "epoch": 0.21380504568825426,
-      "grad_norm": 0.7139214277267456,
-      "learning_rate": 7.861892583120205e-06,
-      "loss": 0.1291,
-      "step": 1254
-    },
-    {
-      "epoch": 0.21397554412979194,
-      "grad_norm": 0.7002556920051575,
-      "learning_rate": 7.860187553282183e-06,
-      "loss": 0.122,
-      "step": 1255
-    },
-    {
-      "epoch": 0.21414604257132963,
-      "grad_norm": 1.3244677782058716,
-      "learning_rate": 7.85848252344416e-06,
-      "loss": 0.2133,
-      "step": 1256
-    },
-    {
-      "epoch": 0.21431654101286732,
-      "grad_norm": 0.5991846919059753,
-      "learning_rate": 7.856777493606139e-06,
-      "loss": 0.1207,
-      "step": 1257
-    },
-    {
-      "epoch": 0.21448703945440498,
-      "grad_norm": 0.9258220195770264,
-      "learning_rate": 7.855072463768117e-06,
-      "loss": 0.1126,
-      "step": 1258
-    },
-    {
-      "epoch": 0.21465753789594266,
-      "grad_norm": 0.80934739112854,
-      "learning_rate": 7.853367433930095e-06,
-      "loss": 0.0496,
-      "step": 1259
-    },
-    {
-      "epoch": 0.21482803633748035,
-      "grad_norm": 0.6553164720535278,
-      "learning_rate": 7.851662404092073e-06,
-      "loss": 0.1773,
-      "step": 1260
-    },
-    {
-      "epoch": 0.21499853477901804,
-      "grad_norm": 1.0420881509780884,
-      "learning_rate": 7.84995737425405e-06,
-      "loss": 0.133,
-      "step": 1261
-    },
-    {
-      "epoch": 0.21516903322055572,
-      "grad_norm": 0.9948811531066895,
-      "learning_rate": 7.848252344416027e-06,
-      "loss": 0.1441,
-      "step": 1262
-    },
-    {
-      "epoch": 0.2153395316620934,
-      "grad_norm": 1.6052919626235962,
-      "learning_rate": 7.846547314578006e-06,
-      "loss": 0.2128,
-      "step": 1263
-    },
-    {
-      "epoch": 0.2155100301036311,
-      "grad_norm": 1.10100519657135,
-      "learning_rate": 7.844842284739984e-06,
-      "loss": 0.1032,
-      "step": 1264
-    },
-    {
-      "epoch": 0.21568052854516875,
-      "grad_norm": 0.7574976682662964,
-      "learning_rate": 7.84313725490196e-06,
-      "loss": 0.1129,
-      "step": 1265
-    },
-    {
-      "epoch": 0.21585102698670644,
-      "grad_norm": 1.1044847965240479,
-      "learning_rate": 7.84143222506394e-06,
-      "loss": 0.1948,
-      "step": 1266
-    },
-    {
-      "epoch": 0.21602152542824413,
-      "grad_norm": 0.8704826831817627,
-      "learning_rate": 7.839727195225917e-06,
-      "loss": 0.1208,
-      "step": 1267
-    },
-    {
-      "epoch": 0.2161920238697818,
-      "grad_norm": 1.2776570320129395,
-      "learning_rate": 7.838022165387895e-06,
-      "loss": 0.1804,
-      "step": 1268
-    },
-    {
-      "epoch": 0.2163625223113195,
-      "grad_norm": 0.8619382381439209,
-      "learning_rate": 7.836317135549872e-06,
-      "loss": 0.142,
-      "step": 1269
-    },
-    {
-      "epoch": 0.21653302075285719,
-      "grad_norm": 0.6960404515266418,
-      "learning_rate": 7.83461210571185e-06,
-      "loss": 0.0856,
-      "step": 1270
-    },
-    {
-      "epoch": 0.21670351919439487,
-      "grad_norm": 0.700543999671936,
-      "learning_rate": 7.832907075873828e-06,
-      "loss": 0.1139,
-      "step": 1271
-    },
-    {
-      "epoch": 0.21687401763593256,
-      "grad_norm": 0.5665769577026367,
-      "learning_rate": 7.831202046035806e-06,
-      "loss": 0.1006,
-      "step": 1272
-    },
-    {
-      "epoch": 0.21704451607747022,
-      "grad_norm": 0.911679208278656,
-      "learning_rate": 7.829497016197784e-06,
-      "loss": 0.166,
-      "step": 1273
-    },
-    {
-      "epoch": 0.2172150145190079,
-      "grad_norm": 0.9011550545692444,
-      "learning_rate": 7.827791986359762e-06,
-      "loss": 0.161,
-      "step": 1274
-    },
-    {
-      "epoch": 0.2173855129605456,
-      "grad_norm": 1.1256483793258667,
-      "learning_rate": 7.82608695652174e-06,
-      "loss": 0.072,
-      "step": 1275
-    },
-    {
-      "epoch": 0.21755601140208328,
-      "grad_norm": 0.7013976573944092,
-      "learning_rate": 7.824381926683718e-06,
-      "loss": 0.1177,
-      "step": 1276
-    },
-    {
-      "epoch": 0.21772650984362096,
-      "grad_norm": 0.8687089681625366,
-      "learning_rate": 7.822676896845696e-06,
-      "loss": 0.1332,
-      "step": 1277
-    },
-    {
-      "epoch": 0.21789700828515865,
-      "grad_norm": 0.7818564772605896,
-      "learning_rate": 7.820971867007672e-06,
-      "loss": 0.1016,
-      "step": 1278
-    },
-    {
-      "epoch": 0.21806750672669634,
-      "grad_norm": 1.2303472757339478,
-      "learning_rate": 7.819266837169652e-06,
-      "loss": 0.2575,
-      "step": 1279
-    },
-    {
-      "epoch": 0.21823800516823402,
-      "grad_norm": 0.6944379210472107,
-      "learning_rate": 7.817561807331628e-06,
-      "loss": 0.1291,
-      "step": 1280
-    },
-    {
-      "epoch": 0.21840850360977168,
-      "grad_norm": 0.9652915596961975,
-      "learning_rate": 7.815856777493606e-06,
-      "loss": 0.1347,
-      "step": 1281
-    },
-    {
-      "epoch": 0.21857900205130937,
-      "grad_norm": 0.9090012311935425,
-      "learning_rate": 7.814151747655586e-06,
-      "loss": 0.1583,
-      "step": 1282
-    },
-    {
-      "epoch": 0.21874950049284705,
-      "grad_norm": 0.774336576461792,
-      "learning_rate": 7.812446717817562e-06,
-      "loss": 0.0878,
-      "step": 1283
-    },
-    {
-      "epoch": 0.21891999893438474,
-      "grad_norm": 1.003588318824768,
-      "learning_rate": 7.81074168797954e-06,
-      "loss": 0.1432,
-      "step": 1284
-    },
-    {
-      "epoch": 0.21909049737592243,
-      "grad_norm": 1.4490317106246948,
-      "learning_rate": 7.809036658141518e-06,
-      "loss": 0.2245,
-      "step": 1285
-    },
-    {
-      "epoch": 0.2192609958174601,
-      "grad_norm": 1.0642396211624146,
-      "learning_rate": 7.807331628303496e-06,
-      "loss": 0.1324,
-      "step": 1286
-    },
-    {
-      "epoch": 0.2194314942589978,
-      "grad_norm": 0.7888051271438599,
-      "learning_rate": 7.805626598465474e-06,
-      "loss": 0.0722,
-      "step": 1287
-    },
-    {
-      "epoch": 0.21960199270053546,
-      "grad_norm": 1.4737591743469238,
-      "learning_rate": 7.803921568627452e-06,
-      "loss": 0.1382,
-      "step": 1288
-    },
-    {
-      "epoch": 0.21977249114207315,
-      "grad_norm": 1.012850284576416,
-      "learning_rate": 7.80221653878943e-06,
-      "loss": 0.1952,
-      "step": 1289
-    },
-    {
-      "epoch": 0.21994298958361083,
-      "grad_norm": 1.8369516134262085,
-      "learning_rate": 7.800511508951408e-06,
-      "loss": 0.2694,
-      "step": 1290
-    },
-    {
-      "epoch": 0.22011348802514852,
-      "grad_norm": 0.830058217048645,
-      "learning_rate": 7.798806479113386e-06,
-      "loss": 0.1163,
-      "step": 1291
-    },
-    {
-      "epoch": 0.2202839864666862,
-      "grad_norm": 0.9844965934753418,
-      "learning_rate": 7.797101449275364e-06,
-      "loss": 0.1419,
-      "step": 1292
-    },
-    {
-      "epoch": 0.2204544849082239,
-      "grad_norm": 1.0009138584136963,
-      "learning_rate": 7.795396419437342e-06,
-      "loss": 0.2142,
-      "step": 1293
-    },
-    {
-      "epoch": 0.22062498334976158,
-      "grad_norm": 1.0360790491104126,
-      "learning_rate": 7.793691389599318e-06,
-      "loss": 0.1299,
-      "step": 1294
-    },
-    {
-      "epoch": 0.22079548179129926,
-      "grad_norm": 0.9678265452384949,
-      "learning_rate": 7.791986359761297e-06,
-      "loss": 0.1877,
-      "step": 1295
-    },
-    {
-      "epoch": 0.22096598023283692,
-      "grad_norm": 1.2459754943847656,
-      "learning_rate": 7.790281329923274e-06,
-      "loss": 0.075,
-      "step": 1296
-    },
-    {
-      "epoch": 0.2211364786743746,
-      "grad_norm": 0.7200685143470764,
-      "learning_rate": 7.788576300085252e-06,
-      "loss": 0.1422,
-      "step": 1297
-    },
-    {
-      "epoch": 0.2213069771159123,
-      "grad_norm": 0.9024417996406555,
-      "learning_rate": 7.786871270247231e-06,
-      "loss": 0.1777,
-      "step": 1298
-    },
-    {
-      "epoch": 0.22147747555744998,
-      "grad_norm": 1.0395989418029785,
-      "learning_rate": 7.785166240409207e-06,
-      "loss": 0.1318,
-      "step": 1299
-    },
-    {
-      "epoch": 0.22164797399898767,
-      "grad_norm": 0.775626003742218,
-      "learning_rate": 7.783461210571185e-06,
-      "loss": 0.107,
-      "step": 1300
-    },
-    {
-      "epoch": 0.22164797399898767,
-      "eval_f1_score": 0.2802197802197802,
-      "eval_loss": 0.16321277618408203,
-      "eval_runtime": 182.6892,
-      "eval_samples_per_second": 54.738,
-      "eval_steps_per_second": 3.421,
-      "step": 1300
-    },
-    {
-      "epoch": 0.22181847244052535,
-      "grad_norm": 0.8368775248527527,
-      "learning_rate": 7.781756180733163e-06,
-      "loss": 0.133,
-      "step": 1301
-    },
-    {
-      "epoch": 0.22198897088206304,
-      "grad_norm": 0.7635461091995239,
-      "learning_rate": 7.780051150895141e-06,
-      "loss": 0.1335,
-      "step": 1302
-    },
-    {
-      "epoch": 0.2221594693236007,
-      "grad_norm": 1.5250236988067627,
-      "learning_rate": 7.77834612105712e-06,
-      "loss": 0.2301,
-      "step": 1303
-    },
-    {
-      "epoch": 0.2223299677651384,
-      "grad_norm": 0.7985216379165649,
-      "learning_rate": 7.776641091219097e-06,
-      "loss": 0.1333,
-      "step": 1304
-    },
-    {
-      "epoch": 0.22250046620667607,
-      "grad_norm": 0.7213661074638367,
-      "learning_rate": 7.774936061381073e-06,
-      "loss": 0.1145,
-      "step": 1305
-    },
-    {
-      "epoch": 0.22267096464821376,
-      "grad_norm": 0.763440728187561,
-      "learning_rate": 7.773231031543053e-06,
-      "loss": 0.1564,
-      "step": 1306
-    },
-    {
-      "epoch": 0.22284146308975145,
-      "grad_norm": 0.8962368369102478,
-      "learning_rate": 7.771526001705031e-06,
-      "loss": 0.1195,
-      "step": 1307
-    },
-    {
-      "epoch": 0.22301196153128913,
-      "grad_norm": 0.9387539029121399,
-      "learning_rate": 7.769820971867007e-06,
-      "loss": 0.1495,
-      "step": 1308
-    },
-    {
-      "epoch": 0.22318245997282682,
-      "grad_norm": 0.8244006037712097,
-      "learning_rate": 7.768115942028987e-06,
-      "loss": 0.1314,
-      "step": 1309
-    },
-    {
-      "epoch": 0.2233529584143645,
-      "grad_norm": 0.7632812261581421,
-      "learning_rate": 7.766410912190963e-06,
-      "loss": 0.1104,
-      "step": 1310
-    },
-    {
-      "epoch": 0.22352345685590216,
-      "grad_norm": 0.7748917937278748,
-      "learning_rate": 7.764705882352941e-06,
-      "loss": 0.1194,
-      "step": 1311
-    },
-    {
-      "epoch": 0.22369395529743985,
-      "grad_norm": 0.8611569404602051,
-      "learning_rate": 7.763000852514919e-06,
-      "loss": 0.1094,
-      "step": 1312
-    },
-    {
-      "epoch": 0.22386445373897754,
-      "grad_norm": 1.057794213294983,
-      "learning_rate": 7.761295822676897e-06,
-      "loss": 0.1883,
-      "step": 1313
-    },
-    {
-      "epoch": 0.22403495218051522,
-      "grad_norm": 0.7631834149360657,
-      "learning_rate": 7.759590792838875e-06,
-      "loss": 0.1968,
-      "step": 1314
-    },
-    {
-      "epoch": 0.2242054506220529,
-      "grad_norm": 0.7893649339675903,
-      "learning_rate": 7.757885763000853e-06,
-      "loss": 0.0726,
-      "step": 1315
-    },
-    {
-      "epoch": 0.2243759490635906,
-      "grad_norm": 0.6442000865936279,
-      "learning_rate": 7.756180733162831e-06,
-      "loss": 0.085,
-      "step": 1316
-    },
-    {
-      "epoch": 0.22454644750512828,
-      "grad_norm": 0.4879632592201233,
-      "learning_rate": 7.754475703324809e-06,
-      "loss": 0.1163,
-      "step": 1317
-    },
-    {
-      "epoch": 0.22471694594666597,
-      "grad_norm": 1.018438696861267,
-      "learning_rate": 7.752770673486787e-06,
-      "loss": 0.1591,
-      "step": 1318
-    },
-    {
-      "epoch": 0.22488744438820363,
-      "grad_norm": 1.1671644449234009,
-      "learning_rate": 7.751065643648765e-06,
-      "loss": 0.1739,
-      "step": 1319
-    },
-    {
-      "epoch": 0.22505794282974131,
-      "grad_norm": 0.8683679699897766,
-      "learning_rate": 7.749360613810743e-06,
-      "loss": 0.1497,
-      "step": 1320
-    },
-    {
-      "epoch": 0.225228441271279,
-      "grad_norm": 1.1401067972183228,
-      "learning_rate": 7.747655583972719e-06,
-      "loss": 0.1526,
-      "step": 1321
-    },
-    {
-      "epoch": 0.2253989397128167,
-      "grad_norm": 1.02974271774292,
-      "learning_rate": 7.745950554134699e-06,
-      "loss": 0.2012,
-      "step": 1322
-    },
-    {
-      "epoch": 0.22556943815435437,
-      "grad_norm": 0.8591212034225464,
-      "learning_rate": 7.744245524296677e-06,
-      "loss": 0.1126,
-      "step": 1323
-    },
-    {
-      "epoch": 0.22573993659589206,
-      "grad_norm": 1.0445603132247925,
-      "learning_rate": 7.742540494458653e-06,
-      "loss": 0.1644,
-      "step": 1324
-    },
-    {
-      "epoch": 0.22591043503742975,
-      "grad_norm": 0.6658400297164917,
-      "learning_rate": 7.740835464620632e-06,
-      "loss": 0.1212,
-      "step": 1325
-    },
-    {
-      "epoch": 0.2260809334789674,
-      "grad_norm": 0.87973552942276,
-      "learning_rate": 7.739130434782609e-06,
-      "loss": 0.1966,
-      "step": 1326
-    },
-    {
-      "epoch": 0.2262514319205051,
-      "grad_norm": 0.8204584717750549,
-      "learning_rate": 7.737425404944587e-06,
-      "loss": 0.1921,
-      "step": 1327
-    },
-    {
-      "epoch": 0.22642193036204278,
-      "grad_norm": 1.1901997327804565,
-      "learning_rate": 7.735720375106565e-06,
-      "loss": 0.2525,
-      "step": 1328
-    },
-    {
-      "epoch": 0.22659242880358046,
-      "grad_norm": 1.110565423965454,
-      "learning_rate": 7.734015345268543e-06,
-      "loss": 0.085,
-      "step": 1329
-    },
-    {
-      "epoch": 0.22676292724511815,
-      "grad_norm": 0.9930194020271301,
-      "learning_rate": 7.73231031543052e-06,
-      "loss": 0.1188,
-      "step": 1330
-    },
-    {
-      "epoch": 0.22693342568665584,
-      "grad_norm": 1.0601656436920166,
-      "learning_rate": 7.730605285592498e-06,
-      "loss": 0.0828,
-      "step": 1331
-    },
-    {
-      "epoch": 0.22710392412819352,
-      "grad_norm": 0.7120490074157715,
-      "learning_rate": 7.728900255754476e-06,
-      "loss": 0.1459,
-      "step": 1332
-    },
-    {
-      "epoch": 0.2272744225697312,
-      "grad_norm": 0.9619858264923096,
-      "learning_rate": 7.727195225916454e-06,
-      "loss": 0.1637,
-      "step": 1333
-    },
-    {
-      "epoch": 0.22744492101126887,
-      "grad_norm": 0.9100483655929565,
-      "learning_rate": 7.725490196078432e-06,
-      "loss": 0.1962,
-      "step": 1334
-    },
-    {
-      "epoch": 0.22761541945280656,
-      "grad_norm": 0.6363540291786194,
-      "learning_rate": 7.72378516624041e-06,
-      "loss": 0.1219,
-      "step": 1335
-    },
-    {
-      "epoch": 0.22778591789434424,
-      "grad_norm": 0.9817243218421936,
-      "learning_rate": 7.722080136402388e-06,
-      "loss": 0.2153,
-      "step": 1336
-    },
-    {
-      "epoch": 0.22795641633588193,
-      "grad_norm": 0.878165602684021,
-      "learning_rate": 7.720375106564364e-06,
-      "loss": 0.1994,
-      "step": 1337
-    },
-    {
-      "epoch": 0.22812691477741961,
-      "grad_norm": 0.9704774618148804,
-      "learning_rate": 7.718670076726344e-06,
-      "loss": 0.1689,
-      "step": 1338
-    },
-    {
-      "epoch": 0.2282974132189573,
-      "grad_norm": 1.0748244524002075,
-      "learning_rate": 7.716965046888322e-06,
-      "loss": 0.2116,
-      "step": 1339
-    },
-    {
-      "epoch": 0.228467911660495,
-      "grad_norm": 0.9812359809875488,
-      "learning_rate": 7.715260017050298e-06,
-      "loss": 0.0619,
-      "step": 1340
-    },
-    {
-      "epoch": 0.22863841010203267,
-      "grad_norm": 1.5158941745758057,
-      "learning_rate": 7.713554987212278e-06,
-      "loss": 0.2373,
-      "step": 1341
-    },
-    {
-      "epoch": 0.22880890854357033,
-      "grad_norm": 0.8915465474128723,
-      "learning_rate": 7.711849957374254e-06,
-      "loss": 0.1591,
-      "step": 1342
-    },
-    {
-      "epoch": 0.22897940698510802,
-      "grad_norm": 0.6390937566757202,
-      "learning_rate": 7.710144927536232e-06,
-      "loss": 0.1521,
-      "step": 1343
-    },
-    {
-      "epoch": 0.2291499054266457,
-      "grad_norm": 0.7857467532157898,
-      "learning_rate": 7.70843989769821e-06,
-      "loss": 0.1484,
-      "step": 1344
-    },
-    {
-      "epoch": 0.2293204038681834,
-      "grad_norm": 0.6358761191368103,
-      "learning_rate": 7.706734867860188e-06,
-      "loss": 0.1284,
-      "step": 1345
-    },
-    {
-      "epoch": 0.22949090230972108,
-      "grad_norm": 0.9754729270935059,
-      "learning_rate": 7.705029838022166e-06,
-      "loss": 0.1711,
-      "step": 1346
-    },
-    {
-      "epoch": 0.22966140075125877,
-      "grad_norm": 0.9040155410766602,
-      "learning_rate": 7.703324808184144e-06,
-      "loss": 0.1411,
-      "step": 1347
-    },
-    {
-      "epoch": 0.22983189919279645,
-      "grad_norm": 1.1135698556900024,
-      "learning_rate": 7.701619778346122e-06,
-      "loss": 0.2615,
-      "step": 1348
-    },
-    {
-      "epoch": 0.2300023976343341,
-      "grad_norm": 0.6657915115356445,
-      "learning_rate": 7.6999147485081e-06,
-      "loss": 0.1502,
-      "step": 1349
-    },
-    {
-      "epoch": 0.2301728960758718,
-      "grad_norm": 0.8232541084289551,
-      "learning_rate": 7.698209718670078e-06,
-      "loss": 0.1593,
-      "step": 1350
-    },
-    {
-      "epoch": 0.23034339451740948,
-      "grad_norm": 0.9148602485656738,
-      "learning_rate": 7.696504688832056e-06,
-      "loss": 0.1059,
-      "step": 1351
-    },
-    {
-      "epoch": 0.23051389295894717,
-      "grad_norm": 0.8433572053909302,
-      "learning_rate": 7.694799658994034e-06,
-      "loss": 0.1235,
-      "step": 1352
-    },
-    {
-      "epoch": 0.23068439140048486,
-      "grad_norm": 0.9784965515136719,
-      "learning_rate": 7.69309462915601e-06,
-      "loss": 0.1795,
-      "step": 1353
-    },
-    {
-      "epoch": 0.23085488984202254,
-      "grad_norm": 0.5294272899627686,
-      "learning_rate": 7.69138959931799e-06,
-      "loss": 0.1123,
-      "step": 1354
-    },
-    {
-      "epoch": 0.23102538828356023,
-      "grad_norm": 0.5891151428222656,
-      "learning_rate": 7.689684569479966e-06,
-      "loss": 0.1402,
-      "step": 1355
-    },
-    {
-      "epoch": 0.23119588672509792,
-      "grad_norm": 0.6787668466567993,
-      "learning_rate": 7.687979539641944e-06,
-      "loss": 0.1335,
-      "step": 1356
-    },
-    {
-      "epoch": 0.23136638516663557,
-      "grad_norm": 0.49819067120552063,
-      "learning_rate": 7.686274509803923e-06,
-      "loss": 0.0955,
-      "step": 1357
-    },
-    {
-      "epoch": 0.23153688360817326,
-      "grad_norm": 0.7876557111740112,
-      "learning_rate": 7.6845694799659e-06,
-      "loss": 0.0986,
-      "step": 1358
-    },
-    {
-      "epoch": 0.23170738204971095,
-      "grad_norm": 0.745729386806488,
-      "learning_rate": 7.682864450127878e-06,
-      "loss": 0.0595,
-      "step": 1359
-    },
-    {
-      "epoch": 0.23187788049124863,
-      "grad_norm": 0.979210376739502,
-      "learning_rate": 7.681159420289856e-06,
-      "loss": 0.1627,
-      "step": 1360
-    },
-    {
-      "epoch": 0.23204837893278632,
-      "grad_norm": 0.6400150060653687,
-      "learning_rate": 7.679454390451833e-06,
-      "loss": 0.1346,
-      "step": 1361
-    },
-    {
-      "epoch": 0.232218877374324,
-      "grad_norm": 1.5608742237091064,
-      "learning_rate": 7.677749360613811e-06,
-      "loss": 0.2716,
-      "step": 1362
-    },
-    {
-      "epoch": 0.2323893758158617,
-      "grad_norm": 1.263974666595459,
-      "learning_rate": 7.67604433077579e-06,
-      "loss": 0.1644,
-      "step": 1363
-    },
-    {
-      "epoch": 0.23255987425739938,
-      "grad_norm": 0.7581167221069336,
-      "learning_rate": 7.674339300937767e-06,
-      "loss": 0.1086,
-      "step": 1364
-    },
-    {
-      "epoch": 0.23273037269893704,
-      "grad_norm": 1.1444523334503174,
-      "learning_rate": 7.672634271099745e-06,
-      "loss": 0.1173,
-      "step": 1365
-    },
-    {
-      "epoch": 0.23290087114047472,
-      "grad_norm": 0.6748406291007996,
-      "learning_rate": 7.670929241261723e-06,
-      "loss": 0.1271,
-      "step": 1366
-    },
-    {
-      "epoch": 0.2330713695820124,
-      "grad_norm": 0.9369028806686401,
-      "learning_rate": 7.669224211423701e-06,
-      "loss": 0.1096,
-      "step": 1367
-    },
-    {
-      "epoch": 0.2332418680235501,
-      "grad_norm": 1.4002010822296143,
-      "learning_rate": 7.667519181585679e-06,
-      "loss": 0.192,
-      "step": 1368
-    },
-    {
-      "epoch": 0.23341236646508778,
-      "grad_norm": 1.4191194772720337,
-      "learning_rate": 7.665814151747655e-06,
-      "loss": 0.1895,
-      "step": 1369
-    },
-    {
-      "epoch": 0.23358286490662547,
-      "grad_norm": 0.6959466338157654,
-      "learning_rate": 7.664109121909635e-06,
-      "loss": 0.0942,
-      "step": 1370
-    },
-    {
-      "epoch": 0.23375336334816316,
-      "grad_norm": 0.9197830557823181,
-      "learning_rate": 7.662404092071611e-06,
-      "loss": 0.1313,
-      "step": 1371
-    },
-    {
-      "epoch": 0.23392386178970082,
-      "grad_norm": 0.7821164727210999,
-      "learning_rate": 7.66069906223359e-06,
-      "loss": 0.1471,
-      "step": 1372
-    },
-    {
-      "epoch": 0.2340943602312385,
-      "grad_norm": 0.7130285501480103,
-      "learning_rate": 7.658994032395569e-06,
-      "loss": 0.1401,
-      "step": 1373
-    },
-    {
-      "epoch": 0.2342648586727762,
-      "grad_norm": 1.06864333152771,
-      "learning_rate": 7.657289002557545e-06,
-      "loss": 0.2456,
-      "step": 1374
-    },
-    {
-      "epoch": 0.23443535711431387,
-      "grad_norm": 1.1891578435897827,
-      "learning_rate": 7.655583972719523e-06,
-      "loss": 0.1243,
-      "step": 1375
-    },
-    {
-      "epoch": 0.23460585555585156,
-      "grad_norm": 1.2399449348449707,
-      "learning_rate": 7.653878942881501e-06,
-      "loss": 0.1893,
-      "step": 1376
-    },
-    {
-      "epoch": 0.23477635399738925,
-      "grad_norm": 0.650579035282135,
-      "learning_rate": 7.652173913043479e-06,
-      "loss": 0.0773,
-      "step": 1377
-    },
-    {
-      "epoch": 0.23494685243892693,
-      "grad_norm": 0.7715778946876526,
-      "learning_rate": 7.650468883205457e-06,
-      "loss": 0.1176,
-      "step": 1378
-    },
-    {
-      "epoch": 0.23511735088046462,
-      "grad_norm": 0.9234492182731628,
-      "learning_rate": 7.648763853367435e-06,
-      "loss": 0.1242,
-      "step": 1379
-    },
-    {
-      "epoch": 0.23528784932200228,
-      "grad_norm": 0.6156173944473267,
-      "learning_rate": 7.647058823529411e-06,
-      "loss": 0.1881,
-      "step": 1380
-    },
-    {
-      "epoch": 0.23545834776353997,
-      "grad_norm": 0.7190359830856323,
-      "learning_rate": 7.64535379369139e-06,
-      "loss": 0.1079,
-      "step": 1381
-    },
-    {
-      "epoch": 0.23562884620507765,
-      "grad_norm": 0.9452102780342102,
-      "learning_rate": 7.643648763853369e-06,
-      "loss": 0.1618,
-      "step": 1382
-    },
-    {
-      "epoch": 0.23579934464661534,
-      "grad_norm": 0.6474363803863525,
-      "learning_rate": 7.641943734015345e-06,
-      "loss": 0.1629,
-      "step": 1383
-    },
-    {
-      "epoch": 0.23596984308815303,
-      "grad_norm": 1.203114628791809,
-      "learning_rate": 7.640238704177325e-06,
-      "loss": 0.1824,
-      "step": 1384
-    },
-    {
-      "epoch": 0.2361403415296907,
-      "grad_norm": 1.1076685190200806,
-      "learning_rate": 7.6385336743393e-06,
-      "loss": 0.1822,
-      "step": 1385
-    },
-    {
-      "epoch": 0.2363108399712284,
-      "grad_norm": 0.3934735655784607,
-      "learning_rate": 7.636828644501279e-06,
-      "loss": 0.0704,
-      "step": 1386
-    },
-    {
-      "epoch": 0.23648133841276608,
-      "grad_norm": 0.8384243845939636,
-      "learning_rate": 7.635123614663257e-06,
-      "loss": 0.1534,
-      "step": 1387
-    },
-    {
-      "epoch": 0.23665183685430374,
-      "grad_norm": 1.733253002166748,
-      "learning_rate": 7.633418584825235e-06,
-      "loss": 0.2937,
-      "step": 1388
-    },
-    {
-      "epoch": 0.23682233529584143,
-      "grad_norm": 0.605530321598053,
-      "learning_rate": 7.631713554987213e-06,
-      "loss": 0.0652,
-      "step": 1389
-    },
-    {
-      "epoch": 0.23699283373737912,
-      "grad_norm": 0.5830997228622437,
-      "learning_rate": 7.63000852514919e-06,
-      "loss": 0.1229,
-      "step": 1390
-    },
-    {
-      "epoch": 0.2371633321789168,
-      "grad_norm": 0.6826112866401672,
-      "learning_rate": 7.628303495311169e-06,
-      "loss": 0.1483,
-      "step": 1391
-    },
-    {
-      "epoch": 0.2373338306204545,
-      "grad_norm": 0.91994708776474,
-      "learning_rate": 7.6265984654731465e-06,
-      "loss": 0.1845,
-      "step": 1392
-    },
-    {
-      "epoch": 0.23750432906199218,
-      "grad_norm": 0.8456718921661377,
-      "learning_rate": 7.624893435635124e-06,
-      "loss": 0.1556,
-      "step": 1393
-    },
-    {
-      "epoch": 0.23767482750352986,
-      "grad_norm": 0.9300505518913269,
-      "learning_rate": 7.6231884057971015e-06,
-      "loss": 0.2081,
-      "step": 1394
-    },
-    {
-      "epoch": 0.23784532594506752,
-      "grad_norm": 0.6279098391532898,
-      "learning_rate": 7.62148337595908e-06,
-      "loss": 0.1495,
-      "step": 1395
-    },
-    {
-      "epoch": 0.2380158243866052,
-      "grad_norm": 0.5213572978973389,
-      "learning_rate": 7.619778346121057e-06,
-      "loss": 0.1036,
-      "step": 1396
-    },
-    {
-      "epoch": 0.2381863228281429,
-      "grad_norm": 0.6799693703651428,
-      "learning_rate": 7.618073316283035e-06,
-      "loss": 0.0962,
-      "step": 1397
-    },
-    {
-      "epoch": 0.23835682126968058,
-      "grad_norm": 0.8099979758262634,
-      "learning_rate": 7.616368286445014e-06,
-      "loss": 0.1187,
-      "step": 1398
-    },
-    {
-      "epoch": 0.23852731971121827,
-      "grad_norm": 0.7557656168937683,
-      "learning_rate": 7.614663256606991e-06,
-      "loss": 0.1806,
-      "step": 1399
-    },
-    {
-      "epoch": 0.23869781815275595,
-      "grad_norm": 0.753513514995575,
-      "learning_rate": 7.612958226768969e-06,
-      "loss": 0.0898,
-      "step": 1400
-    },
-    {
-      "epoch": 0.23869781815275595,
-      "eval_f1_score": 0.17363344051446947,
-      "eval_loss": 0.16346921026706696,
-      "eval_runtime": 182.6256,
-      "eval_samples_per_second": 54.757,
-      "eval_steps_per_second": 3.422,
-      "step": 1400
-    },
-    {
-      "epoch": 0.23886831659429364,
-      "grad_norm": 0.6202503442764282,
-      "learning_rate": 7.611253196930946e-06,
-      "loss": 0.1279,
-      "step": 1401
-    },
-    {
-      "epoch": 0.23903881503583133,
-      "grad_norm": 1.2577037811279297,
-      "learning_rate": 7.609548167092925e-06,
-      "loss": 0.1738,
-      "step": 1402
-    },
-    {
-      "epoch": 0.23920931347736898,
-      "grad_norm": 0.665626049041748,
-      "learning_rate": 7.607843137254902e-06,
-      "loss": 0.1537,
-      "step": 1403
-    },
-    {
-      "epoch": 0.23937981191890667,
-      "grad_norm": 0.6698447465896606,
-      "learning_rate": 7.60613810741688e-06,
-      "loss": 0.0781,
-      "step": 1404
-    },
-    {
-      "epoch": 0.23955031036044436,
-      "grad_norm": 0.7142729163169861,
-      "learning_rate": 7.604433077578858e-06,
-      "loss": 0.1481,
-      "step": 1405
-    },
-    {
-      "epoch": 0.23972080880198204,
-      "grad_norm": 0.6046807765960693,
-      "learning_rate": 7.602728047740836e-06,
-      "loss": 0.1471,
-      "step": 1406
-    },
-    {
-      "epoch": 0.23989130724351973,
-      "grad_norm": 0.7382557988166809,
-      "learning_rate": 7.601023017902814e-06,
-      "loss": 0.1355,
-      "step": 1407
-    },
-    {
-      "epoch": 0.24006180568505742,
-      "grad_norm": 0.6939229369163513,
-      "learning_rate": 7.599317988064792e-06,
-      "loss": 0.0534,
-      "step": 1408
-    },
-    {
-      "epoch": 0.2402323041265951,
-      "grad_norm": 1.3162113428115845,
-      "learning_rate": 7.59761295822677e-06,
-      "loss": 0.2284,
-      "step": 1409
-    },
-    {
-      "epoch": 0.24040280256813276,
-      "grad_norm": 0.7624379396438599,
-      "learning_rate": 7.595907928388747e-06,
-      "loss": 0.1293,
-      "step": 1410
-    },
-    {
-      "epoch": 0.24057330100967045,
-      "grad_norm": 0.676016628742218,
-      "learning_rate": 7.594202898550726e-06,
-      "loss": 0.1288,
-      "step": 1411
-    },
-    {
-      "epoch": 0.24074379945120813,
-      "grad_norm": 0.7912443280220032,
-      "learning_rate": 7.592497868712703e-06,
-      "loss": 0.0694,
-      "step": 1412
-    },
-    {
-      "epoch": 0.24091429789274582,
-      "grad_norm": 0.8490058183670044,
-      "learning_rate": 7.590792838874681e-06,
-      "loss": 0.0542,
-      "step": 1413
-    },
-    {
-      "epoch": 0.2410847963342835,
-      "grad_norm": 1.5905382633209229,
-      "learning_rate": 7.58908780903666e-06,
-      "loss": 0.2503,
-      "step": 1414
-    },
-    {
-      "epoch": 0.2412552947758212,
-      "grad_norm": 0.7132392525672913,
-      "learning_rate": 7.587382779198637e-06,
-      "loss": 0.0873,
-      "step": 1415
-    },
-    {
-      "epoch": 0.24142579321735888,
-      "grad_norm": 0.8989769220352173,
-      "learning_rate": 7.585677749360615e-06,
-      "loss": 0.1376,
-      "step": 1416
-    },
-    {
-      "epoch": 0.24159629165889657,
-      "grad_norm": 0.7588139772415161,
-      "learning_rate": 7.583972719522592e-06,
-      "loss": 0.0944,
-      "step": 1417
-    },
-    {
-      "epoch": 0.24176679010043423,
-      "grad_norm": 0.5211461186408997,
-      "learning_rate": 7.5822676896845706e-06,
-      "loss": 0.0748,
-      "step": 1418
-    },
-    {
-      "epoch": 0.2419372885419719,
-      "grad_norm": 0.6701415181159973,
-      "learning_rate": 7.580562659846548e-06,
-      "loss": 0.0785,
-      "step": 1419
-    },
-    {
-      "epoch": 0.2421077869835096,
-      "grad_norm": 1.8684077262878418,
-      "learning_rate": 7.578857630008526e-06,
-      "loss": 0.2471,
-      "step": 1420
-    },
-    {
-      "epoch": 0.24227828542504729,
-      "grad_norm": 0.7341625690460205,
-      "learning_rate": 7.577152600170503e-06,
-      "loss": 0.1161,
-      "step": 1421
-    },
-    {
-      "epoch": 0.24244878386658497,
-      "grad_norm": 0.9016053676605225,
-      "learning_rate": 7.5754475703324815e-06,
-      "loss": 0.1613,
-      "step": 1422
-    },
-    {
-      "epoch": 0.24261928230812266,
-      "grad_norm": 0.8441269993782043,
-      "learning_rate": 7.5737425404944595e-06,
-      "loss": 0.1145,
-      "step": 1423
-    },
-    {
-      "epoch": 0.24278978074966034,
-      "grad_norm": 0.6860020756721497,
-      "learning_rate": 7.572037510656437e-06,
-      "loss": 0.199,
-      "step": 1424
-    },
-    {
-      "epoch": 0.24296027919119803,
-      "grad_norm": 1.288138747215271,
-      "learning_rate": 7.570332480818415e-06,
-      "loss": 0.1958,
-      "step": 1425
-    },
-    {
-      "epoch": 0.2431307776327357,
-      "grad_norm": 1.4583251476287842,
-      "learning_rate": 7.5686274509803925e-06,
-      "loss": 0.2127,
-      "step": 1426
-    },
-    {
-      "epoch": 0.24330127607427338,
-      "grad_norm": 0.8146334886550903,
-      "learning_rate": 7.566922421142371e-06,
-      "loss": 0.1469,
-      "step": 1427
-    },
-    {
-      "epoch": 0.24347177451581106,
-      "grad_norm": 0.7682924866676331,
-      "learning_rate": 7.565217391304348e-06,
-      "loss": 0.1244,
-      "step": 1428
-    },
-    {
-      "epoch": 0.24364227295734875,
-      "grad_norm": 0.9778890609741211,
-      "learning_rate": 7.563512361466326e-06,
-      "loss": 0.0938,
-      "step": 1429
-    },
-    {
-      "epoch": 0.24381277139888644,
-      "grad_norm": 1.0876981019973755,
-      "learning_rate": 7.561807331628303e-06,
-      "loss": 0.2666,
-      "step": 1430
-    },
-    {
-      "epoch": 0.24398326984042412,
-      "grad_norm": 0.7767715454101562,
-      "learning_rate": 7.560102301790282e-06,
-      "loss": 0.1377,
-      "step": 1431
-    },
-    {
-      "epoch": 0.2441537682819618,
-      "grad_norm": 0.878875195980072,
-      "learning_rate": 7.55839727195226e-06,
-      "loss": 0.184,
-      "step": 1432
-    },
-    {
-      "epoch": 0.24432426672349947,
-      "grad_norm": 1.333276391029358,
-      "learning_rate": 7.556692242114237e-06,
-      "loss": 0.1676,
-      "step": 1433
-    },
-    {
-      "epoch": 0.24449476516503715,
-      "grad_norm": 0.685215413570404,
-      "learning_rate": 7.554987212276216e-06,
-      "loss": 0.1374,
-      "step": 1434
-    },
-    {
-      "epoch": 0.24466526360657484,
-      "grad_norm": 1.133496880531311,
-      "learning_rate": 7.553282182438193e-06,
-      "loss": 0.0851,
-      "step": 1435
-    },
-    {
-      "epoch": 0.24483576204811253,
-      "grad_norm": 0.6287438869476318,
-      "learning_rate": 7.551577152600171e-06,
-      "loss": 0.1637,
-      "step": 1436
-    },
-    {
-      "epoch": 0.2450062604896502,
-      "grad_norm": 0.9850641489028931,
-      "learning_rate": 7.549872122762148e-06,
-      "loss": 0.1658,
-      "step": 1437
-    },
-    {
-      "epoch": 0.2451767589311879,
-      "grad_norm": 0.7101774215698242,
-      "learning_rate": 7.548167092924127e-06,
-      "loss": 0.102,
-      "step": 1438
-    },
-    {
-      "epoch": 0.24534725737272559,
-      "grad_norm": 1.6058562994003296,
-      "learning_rate": 7.546462063086104e-06,
-      "loss": 0.2678,
-      "step": 1439
-    },
-    {
-      "epoch": 0.24551775581426327,
-      "grad_norm": 1.1810576915740967,
-      "learning_rate": 7.544757033248082e-06,
-      "loss": 0.2186,
-      "step": 1440
-    },
-    {
-      "epoch": 0.24568825425580093,
-      "grad_norm": 1.7627534866333008,
-      "learning_rate": 7.543052003410061e-06,
-      "loss": 0.311,
-      "step": 1441
-    },
-    {
-      "epoch": 0.24585875269733862,
-      "grad_norm": 1.5101150274276733,
-      "learning_rate": 7.541346973572038e-06,
-      "loss": 0.2104,
-      "step": 1442
-    },
-    {
-      "epoch": 0.2460292511388763,
-      "grad_norm": 0.5594928860664368,
-      "learning_rate": 7.539641943734016e-06,
-      "loss": 0.1275,
-      "step": 1443
-    },
-    {
-      "epoch": 0.246199749580414,
-      "grad_norm": 0.6583553552627563,
-      "learning_rate": 7.537936913895994e-06,
-      "loss": 0.1235,
-      "step": 1444
-    },
-    {
-      "epoch": 0.24637024802195168,
-      "grad_norm": 0.7115300893783569,
-      "learning_rate": 7.536231884057972e-06,
-      "loss": 0.1055,
-      "step": 1445
-    },
-    {
-      "epoch": 0.24654074646348936,
-      "grad_norm": 0.8568644523620605,
-      "learning_rate": 7.534526854219949e-06,
-      "loss": 0.0792,
-      "step": 1446
-    },
-    {
-      "epoch": 0.24671124490502705,
-      "grad_norm": 0.6825379133224487,
-      "learning_rate": 7.532821824381928e-06,
-      "loss": 0.1291,
-      "step": 1447
-    },
-    {
-      "epoch": 0.24688174334656474,
-      "grad_norm": 0.6188522577285767,
-      "learning_rate": 7.531116794543906e-06,
-      "loss": 0.094,
-      "step": 1448
-    },
-    {
-      "epoch": 0.2470522417881024,
-      "grad_norm": 1.4240697622299194,
-      "learning_rate": 7.529411764705883e-06,
-      "loss": 0.2407,
-      "step": 1449
-    },
-    {
-      "epoch": 0.24722274022964008,
-      "grad_norm": 0.8678609728813171,
-      "learning_rate": 7.5277067348678615e-06,
-      "loss": 0.1461,
-      "step": 1450
-    },
-    {
-      "epoch": 0.24739323867117777,
-      "grad_norm": 0.8159744143486023,
-      "learning_rate": 7.526001705029839e-06,
-      "loss": 0.1565,
-      "step": 1451
-    },
-    {
-      "epoch": 0.24756373711271545,
-      "grad_norm": 1.4507590532302856,
-      "learning_rate": 7.5242966751918166e-06,
-      "loss": 0.2225,
-      "step": 1452
-    },
-    {
-      "epoch": 0.24773423555425314,
-      "grad_norm": 0.7991291880607605,
-      "learning_rate": 7.522591645353794e-06,
-      "loss": 0.1174,
-      "step": 1453
-    },
-    {
-      "epoch": 0.24790473399579083,
-      "grad_norm": 0.8276837468147278,
-      "learning_rate": 7.5208866155157724e-06,
-      "loss": 0.0945,
-      "step": 1454
-    },
-    {
-      "epoch": 0.2480752324373285,
-      "grad_norm": 0.8337952494621277,
-      "learning_rate": 7.5191815856777495e-06,
-      "loss": 0.184,
-      "step": 1455
-    },
-    {
-      "epoch": 0.24824573087886617,
-      "grad_norm": 0.7352725863456726,
-      "learning_rate": 7.5174765558397275e-06,
-      "loss": 0.0979,
-      "step": 1456
-    },
-    {
-      "epoch": 0.24841622932040386,
-      "grad_norm": 0.8864614367485046,
-      "learning_rate": 7.515771526001706e-06,
-      "loss": 0.0645,
-      "step": 1457
-    },
-    {
-      "epoch": 0.24858672776194154,
-      "grad_norm": 0.9972084760665894,
-      "learning_rate": 7.514066496163683e-06,
-      "loss": 0.1962,
-      "step": 1458
-    },
-    {
-      "epoch": 0.24875722620347923,
-      "grad_norm": 1.0494168996810913,
-      "learning_rate": 7.512361466325661e-06,
-      "loss": 0.2017,
-      "step": 1459
-    },
-    {
-      "epoch": 0.24892772464501692,
-      "grad_norm": 0.9285470843315125,
-      "learning_rate": 7.510656436487639e-06,
-      "loss": 0.1574,
-      "step": 1460
-    },
-    {
-      "epoch": 0.2490982230865546,
-      "grad_norm": 0.5147951245307922,
-      "learning_rate": 7.508951406649617e-06,
-      "loss": 0.0829,
-      "step": 1461
-    },
-    {
-      "epoch": 0.2492687215280923,
-      "grad_norm": 0.733745813369751,
-      "learning_rate": 7.507246376811594e-06,
-      "loss": 0.0967,
-      "step": 1462
-    },
-    {
-      "epoch": 0.24943921996962998,
-      "grad_norm": 1.0238525867462158,
-      "learning_rate": 7.505541346973573e-06,
-      "loss": 0.1784,
-      "step": 1463
-    },
-    {
-      "epoch": 0.24960971841116764,
-      "grad_norm": 0.6515784859657288,
-      "learning_rate": 7.50383631713555e-06,
-      "loss": 0.1098,
-      "step": 1464
-    },
-    {
-      "epoch": 0.24978021685270532,
-      "grad_norm": 0.833389401435852,
-      "learning_rate": 7.502131287297528e-06,
-      "loss": 0.1741,
-      "step": 1465
-    },
-    {
-      "epoch": 0.249950715294243,
-      "grad_norm": 1.6986770629882812,
-      "learning_rate": 7.500426257459507e-06,
-      "loss": 0.2424,
-      "step": 1466
-    },
-    {
-      "epoch": 0.2501212137357807,
-      "grad_norm": 0.6558035016059875,
-      "learning_rate": 7.498721227621484e-06,
-      "loss": 0.0959,
-      "step": 1467
-    },
-    {
-      "epoch": 0.2502917121773184,
-      "grad_norm": 0.7692930102348328,
-      "learning_rate": 7.497016197783462e-06,
-      "loss": 0.1215,
-      "step": 1468
-    },
-    {
-      "epoch": 0.25046221061885604,
-      "grad_norm": 0.6363939046859741,
-      "learning_rate": 7.495311167945439e-06,
-      "loss": 0.1522,
-      "step": 1469
-    },
-    {
-      "epoch": 0.25063270906039375,
-      "grad_norm": 0.7761989831924438,
-      "learning_rate": 7.493606138107418e-06,
-      "loss": 0.1029,
-      "step": 1470
-    },
-    {
-      "epoch": 0.2508032075019314,
-      "grad_norm": 0.8327432870864868,
-      "learning_rate": 7.491901108269395e-06,
-      "loss": 0.1437,
-      "step": 1471
-    },
-    {
-      "epoch": 0.2509737059434691,
-      "grad_norm": 0.7670499682426453,
-      "learning_rate": 7.490196078431373e-06,
-      "loss": 0.1171,
-      "step": 1472
-    },
-    {
-      "epoch": 0.2511442043850068,
-      "grad_norm": 0.7924665808677673,
-      "learning_rate": 7.488491048593352e-06,
-      "loss": 0.1267,
-      "step": 1473
-    },
-    {
-      "epoch": 0.2513147028265445,
-      "grad_norm": 0.6221386790275574,
-      "learning_rate": 7.486786018755329e-06,
-      "loss": 0.1241,
-      "step": 1474
-    },
-    {
-      "epoch": 0.25148520126808216,
-      "grad_norm": 0.6074143648147583,
-      "learning_rate": 7.485080988917307e-06,
-      "loss": 0.0999,
-      "step": 1475
-    },
-    {
-      "epoch": 0.2516556997096198,
-      "grad_norm": 0.8128476738929749,
-      "learning_rate": 7.483375959079284e-06,
-      "loss": 0.1176,
-      "step": 1476
-    },
-    {
-      "epoch": 0.25182619815115753,
-      "grad_norm": 0.7725571393966675,
-      "learning_rate": 7.481670929241263e-06,
-      "loss": 0.156,
-      "step": 1477
-    },
-    {
-      "epoch": 0.2519966965926952,
-      "grad_norm": 0.8417410254478455,
-      "learning_rate": 7.47996589940324e-06,
-      "loss": 0.1673,
-      "step": 1478
-    },
-    {
-      "epoch": 0.2521671950342329,
-      "grad_norm": 1.1635167598724365,
-      "learning_rate": 7.478260869565218e-06,
-      "loss": 0.1718,
-      "step": 1479
-    },
-    {
-      "epoch": 0.25233769347577056,
-      "grad_norm": 1.0357716083526611,
-      "learning_rate": 7.476555839727196e-06,
-      "loss": 0.2007,
-      "step": 1480
-    },
-    {
-      "epoch": 0.2525081919173083,
-      "grad_norm": 2.6045241355895996,
-      "learning_rate": 7.474850809889174e-06,
-      "loss": 0.3113,
-      "step": 1481
-    },
-    {
-      "epoch": 0.25267869035884594,
-      "grad_norm": 0.8889693021774292,
-      "learning_rate": 7.473145780051152e-06,
-      "loss": 0.1581,
-      "step": 1482
-    },
-    {
-      "epoch": 0.2528491888003836,
-      "grad_norm": 0.6757845878601074,
-      "learning_rate": 7.4714407502131295e-06,
-      "loss": 0.1063,
-      "step": 1483
-    },
-    {
-      "epoch": 0.2530196872419213,
-      "grad_norm": 1.5937942266464233,
-      "learning_rate": 7.4697357203751075e-06,
-      "loss": 0.239,
-      "step": 1484
-    },
-    {
-      "epoch": 0.25319018568345897,
-      "grad_norm": 0.6575447916984558,
-      "learning_rate": 7.468030690537085e-06,
-      "loss": 0.126,
-      "step": 1485
-    },
-    {
-      "epoch": 0.2533606841249967,
-      "grad_norm": 0.8341320157051086,
-      "learning_rate": 7.466325660699063e-06,
-      "loss": 0.1399,
-      "step": 1486
-    },
-    {
-      "epoch": 0.25353118256653434,
-      "grad_norm": 0.6826508641242981,
-      "learning_rate": 7.4646206308610405e-06,
-      "loss": 0.1326,
-      "step": 1487
-    },
-    {
-      "epoch": 0.25370168100807206,
-      "grad_norm": 1.1577479839324951,
-      "learning_rate": 7.4629156010230184e-06,
-      "loss": 0.1091,
-      "step": 1488
-    },
-    {
-      "epoch": 0.2538721794496097,
-      "grad_norm": 0.8800094723701477,
-      "learning_rate": 7.4612105711849955e-06,
-      "loss": 0.1895,
-      "step": 1489
-    },
-    {
-      "epoch": 0.25404267789114743,
-      "grad_norm": 0.8499655723571777,
-      "learning_rate": 7.459505541346974e-06,
-      "loss": 0.1603,
-      "step": 1490
-    },
-    {
-      "epoch": 0.2542131763326851,
-      "grad_norm": 1.1757783889770508,
-      "learning_rate": 7.457800511508952e-06,
-      "loss": 0.1543,
-      "step": 1491
-    },
-    {
-      "epoch": 0.25438367477422275,
-      "grad_norm": 0.9781895279884338,
-      "learning_rate": 7.456095481670929e-06,
-      "loss": 0.1987,
-      "step": 1492
-    },
-    {
-      "epoch": 0.25455417321576046,
-      "grad_norm": 0.866536557674408,
-      "learning_rate": 7.454390451832908e-06,
-      "loss": 0.0631,
-      "step": 1493
-    },
-    {
-      "epoch": 0.2547246716572981,
-      "grad_norm": 0.7580909729003906,
-      "learning_rate": 7.452685421994885e-06,
-      "loss": 0.122,
-      "step": 1494
-    },
-    {
-      "epoch": 0.25489517009883583,
-      "grad_norm": 1.1645458936691284,
-      "learning_rate": 7.450980392156863e-06,
-      "loss": 0.2327,
-      "step": 1495
-    },
-    {
-      "epoch": 0.2550656685403735,
-      "grad_norm": 0.7042590975761414,
-      "learning_rate": 7.44927536231884e-06,
-      "loss": 0.1525,
-      "step": 1496
-    },
-    {
-      "epoch": 0.2552361669819112,
-      "grad_norm": 1.3674838542938232,
-      "learning_rate": 7.447570332480819e-06,
-      "loss": 0.2254,
-      "step": 1497
-    },
-    {
-      "epoch": 0.25540666542344886,
-      "grad_norm": 0.9225447773933411,
-      "learning_rate": 7.445865302642797e-06,
-      "loss": 0.1388,
-      "step": 1498
-    },
-    {
-      "epoch": 0.2555771638649865,
-      "grad_norm": 0.6369597911834717,
-      "learning_rate": 7.444160272804775e-06,
-      "loss": 0.129,
-      "step": 1499
-    },
-    {
-      "epoch": 0.25574766230652424,
-      "grad_norm": 0.8218846917152405,
-      "learning_rate": 7.442455242966753e-06,
-      "loss": 0.1463,
-      "step": 1500
-    },
-    {
-      "epoch": 0.25574766230652424,
-      "eval_f1_score": 0.24148606811145512,
-      "eval_loss": 0.1599266529083252,
-      "eval_runtime": 182.5904,
-      "eval_samples_per_second": 54.767,
-      "eval_steps_per_second": 3.423,
-      "step": 1500
-    },
-    {
-      "epoch": 0.2559181607480619,
-      "grad_norm": 0.6795699596405029,
-      "learning_rate": 7.44075021312873e-06,
-      "loss": 0.1336,
-      "step": 1501
-    },
-    {
-      "epoch": 0.2560886591895996,
-      "grad_norm": 1.0670921802520752,
-      "learning_rate": 7.439045183290709e-06,
-      "loss": 0.2411,
-      "step": 1502
-    },
-    {
-      "epoch": 0.25625915763113727,
-      "grad_norm": 1.0216519832611084,
-      "learning_rate": 7.437340153452686e-06,
-      "loss": 0.1929,
-      "step": 1503
-    },
-    {
-      "epoch": 0.256429656072675,
-      "grad_norm": 0.6687795519828796,
-      "learning_rate": 7.435635123614664e-06,
-      "loss": 0.1713,
-      "step": 1504
-    },
-    {
-      "epoch": 0.25660015451421264,
-      "grad_norm": 0.8873473405838013,
-      "learning_rate": 7.433930093776641e-06,
-      "loss": 0.1443,
-      "step": 1505
-    },
-    {
-      "epoch": 0.2567706529557503,
-      "grad_norm": 0.8436746001243591,
-      "learning_rate": 7.43222506393862e-06,
-      "loss": 0.102,
-      "step": 1506
-    },
-    {
-      "epoch": 0.256941151397288,
-      "grad_norm": 0.6887624263763428,
-      "learning_rate": 7.430520034100598e-06,
-      "loss": 0.1807,
-      "step": 1507
-    },
-    {
-      "epoch": 0.2571116498388257,
-      "grad_norm": 0.7564592361450195,
-      "learning_rate": 7.428815004262575e-06,
-      "loss": 0.0905,
-      "step": 1508
-    },
-    {
-      "epoch": 0.2572821482803634,
-      "grad_norm": 0.6453114151954651,
-      "learning_rate": 7.427109974424554e-06,
-      "loss": 0.0818,
-      "step": 1509
-    },
-    {
-      "epoch": 0.25745264672190105,
-      "grad_norm": 0.5713956356048584,
-      "learning_rate": 7.425404944586531e-06,
-      "loss": 0.1265,
-      "step": 1510
-    },
-    {
-      "epoch": 0.25762314516343876,
-      "grad_norm": 0.7340114712715149,
-      "learning_rate": 7.423699914748509e-06,
-      "loss": 0.1635,
-      "step": 1511
-    },
-    {
-      "epoch": 0.2577936436049764,
-      "grad_norm": 0.7583447694778442,
-      "learning_rate": 7.421994884910486e-06,
-      "loss": 0.1548,
-      "step": 1512
-    },
-    {
-      "epoch": 0.25796414204651413,
-      "grad_norm": 0.750053882598877,
-      "learning_rate": 7.420289855072465e-06,
-      "loss": 0.1527,
-      "step": 1513
-    },
-    {
-      "epoch": 0.2581346404880518,
-      "grad_norm": 1.1867377758026123,
-      "learning_rate": 7.418584825234442e-06,
-      "loss": 0.1923,
-      "step": 1514
-    },
-    {
-      "epoch": 0.25830513892958945,
-      "grad_norm": 1.0661702156066895,
-      "learning_rate": 7.41687979539642e-06,
-      "loss": 0.1561,
-      "step": 1515
-    },
-    {
-      "epoch": 0.25847563737112716,
-      "grad_norm": 0.6860876083374023,
-      "learning_rate": 7.4151747655583984e-06,
-      "loss": 0.0991,
-      "step": 1516
-    },
-    {
-      "epoch": 0.2586461358126648,
-      "grad_norm": 0.6717040538787842,
-      "learning_rate": 7.4134697357203755e-06,
-      "loss": 0.2006,
-      "step": 1517
-    },
-    {
-      "epoch": 0.25881663425420254,
-      "grad_norm": 1.3219722509384155,
-      "learning_rate": 7.4117647058823535e-06,
-      "loss": 0.2566,
-      "step": 1518
-    },
-    {
-      "epoch": 0.2589871326957402,
-      "grad_norm": 1.4991238117218018,
-      "learning_rate": 7.4100596760443314e-06,
-      "loss": 0.2157,
-      "step": 1519
-    },
-    {
-      "epoch": 0.2591576311372779,
-      "grad_norm": 0.6971268653869629,
-      "learning_rate": 7.408354646206309e-06,
-      "loss": 0.1123,
-      "step": 1520
-    },
-    {
-      "epoch": 0.25932812957881557,
-      "grad_norm": 0.9756630659103394,
-      "learning_rate": 7.4066496163682865e-06,
-      "loss": 0.1835,
-      "step": 1521
-    },
-    {
-      "epoch": 0.25949862802035323,
-      "grad_norm": 0.7816482782363892,
-      "learning_rate": 7.404944586530265e-06,
-      "loss": 0.1717,
-      "step": 1522
-    },
-    {
-      "epoch": 0.25966912646189094,
-      "grad_norm": 0.828657865524292,
-      "learning_rate": 7.403239556692243e-06,
-      "loss": 0.1442,
-      "step": 1523
-    },
-    {
-      "epoch": 0.2598396249034286,
-      "grad_norm": 0.9661919474601746,
-      "learning_rate": 7.40153452685422e-06,
-      "loss": 0.1028,
-      "step": 1524
-    },
-    {
-      "epoch": 0.2600101233449663,
-      "grad_norm": 0.9794785976409912,
-      "learning_rate": 7.399829497016199e-06,
-      "loss": 0.1285,
-      "step": 1525
-    },
-    {
-      "epoch": 0.260180621786504,
-      "grad_norm": 1.2873239517211914,
-      "learning_rate": 7.398124467178176e-06,
-      "loss": 0.0812,
-      "step": 1526
-    },
-    {
-      "epoch": 0.2603511202280417,
-      "grad_norm": 0.8300976753234863,
-      "learning_rate": 7.396419437340154e-06,
-      "loss": 0.1411,
-      "step": 1527
-    },
-    {
-      "epoch": 0.26052161866957935,
-      "grad_norm": 0.7613148093223572,
-      "learning_rate": 7.394714407502131e-06,
-      "loss": 0.1102,
-      "step": 1528
-    },
-    {
-      "epoch": 0.260692117111117,
-      "grad_norm": 0.8103572130203247,
-      "learning_rate": 7.39300937766411e-06,
-      "loss": 0.1837,
-      "step": 1529
-    },
-    {
-      "epoch": 0.2608626155526547,
-      "grad_norm": 0.6886932849884033,
-      "learning_rate": 7.391304347826087e-06,
-      "loss": 0.0954,
-      "step": 1530
-    },
-    {
-      "epoch": 0.2610331139941924,
-      "grad_norm": 1.1097701787948608,
-      "learning_rate": 7.389599317988065e-06,
-      "loss": 0.1631,
-      "step": 1531
-    },
-    {
-      "epoch": 0.2612036124357301,
-      "grad_norm": 1.2780195474624634,
-      "learning_rate": 7.387894288150044e-06,
-      "loss": 0.1674,
-      "step": 1532
-    },
-    {
-      "epoch": 0.26137411087726775,
-      "grad_norm": 0.9154688715934753,
-      "learning_rate": 7.386189258312021e-06,
-      "loss": 0.1251,
-      "step": 1533
-    },
-    {
-      "epoch": 0.26154460931880547,
-      "grad_norm": 1.2270097732543945,
-      "learning_rate": 7.384484228473999e-06,
-      "loss": 0.1837,
-      "step": 1534
-    },
-    {
-      "epoch": 0.2617151077603431,
-      "grad_norm": 0.664033055305481,
-      "learning_rate": 7.382779198635977e-06,
-      "loss": 0.0543,
-      "step": 1535
-    },
-    {
-      "epoch": 0.26188560620188084,
-      "grad_norm": 1.0188624858856201,
-      "learning_rate": 7.381074168797955e-06,
-      "loss": 0.1608,
-      "step": 1536
-    },
-    {
-      "epoch": 0.2620561046434185,
-      "grad_norm": 1.2438368797302246,
-      "learning_rate": 7.379369138959932e-06,
-      "loss": 0.1671,
-      "step": 1537
-    },
-    {
-      "epoch": 0.26222660308495616,
-      "grad_norm": 0.8407082557678223,
-      "learning_rate": 7.377664109121911e-06,
-      "loss": 0.0658,
-      "step": 1538
-    },
-    {
-      "epoch": 0.26239710152649387,
-      "grad_norm": 1.376527190208435,
-      "learning_rate": 7.375959079283888e-06,
-      "loss": 0.1673,
-      "step": 1539
-    },
-    {
-      "epoch": 0.26256759996803153,
-      "grad_norm": 0.7966179251670837,
-      "learning_rate": 7.374254049445866e-06,
-      "loss": 0.1,
-      "step": 1540
-    },
-    {
-      "epoch": 0.26273809840956924,
-      "grad_norm": 1.4758623838424683,
-      "learning_rate": 7.372549019607845e-06,
-      "loss": 0.2534,
-      "step": 1541
-    },
-    {
-      "epoch": 0.2629085968511069,
-      "grad_norm": 0.7056615352630615,
-      "learning_rate": 7.370843989769822e-06,
-      "loss": 0.1085,
-      "step": 1542
-    },
-    {
-      "epoch": 0.2630790952926446,
-      "grad_norm": 0.9349361062049866,
-      "learning_rate": 7.3691389599318e-06,
-      "loss": 0.1546,
-      "step": 1543
-    },
-    {
-      "epoch": 0.2632495937341823,
-      "grad_norm": 1.2983108758926392,
-      "learning_rate": 7.367433930093777e-06,
-      "loss": 0.086,
-      "step": 1544
-    },
-    {
-      "epoch": 0.26342009217571993,
-      "grad_norm": 0.9976209998130798,
-      "learning_rate": 7.3657289002557555e-06,
-      "loss": 0.0898,
-      "step": 1545
-    },
-    {
-      "epoch": 0.26359059061725765,
-      "grad_norm": 1.1741667985916138,
-      "learning_rate": 7.364023870417733e-06,
-      "loss": 0.1917,
-      "step": 1546
-    },
-    {
-      "epoch": 0.2637610890587953,
-      "grad_norm": 0.8395033478736877,
-      "learning_rate": 7.3623188405797106e-06,
-      "loss": 0.0972,
-      "step": 1547
-    },
-    {
-      "epoch": 0.263931587500333,
-      "grad_norm": 1.1209566593170166,
-      "learning_rate": 7.360613810741689e-06,
-      "loss": 0.1802,
-      "step": 1548
-    },
-    {
-      "epoch": 0.2641020859418707,
-      "grad_norm": 0.9700178503990173,
-      "learning_rate": 7.3589087809036665e-06,
-      "loss": 0.1734,
-      "step": 1549
-    },
-    {
-      "epoch": 0.2642725843834084,
-      "grad_norm": 0.7460111379623413,
-      "learning_rate": 7.357203751065644e-06,
-      "loss": 0.1348,
-      "step": 1550
-    },
-    {
-      "epoch": 0.26444308282494605,
-      "grad_norm": 0.4582483470439911,
-      "learning_rate": 7.3554987212276215e-06,
-      "loss": 0.0815,
-      "step": 1551
-    },
-    {
-      "epoch": 0.2646135812664837,
-      "grad_norm": 1.1477794647216797,
-      "learning_rate": 7.3537936913896e-06,
-      "loss": 0.1943,
-      "step": 1552
-    },
-    {
-      "epoch": 0.2647840797080214,
-      "grad_norm": 0.8271501660346985,
-      "learning_rate": 7.352088661551577e-06,
-      "loss": 0.1753,
-      "step": 1553
-    },
-    {
-      "epoch": 0.2649545781495591,
-      "grad_norm": 0.8639466762542725,
-      "learning_rate": 7.350383631713555e-06,
-      "loss": 0.1813,
-      "step": 1554
-    },
-    {
-      "epoch": 0.2651250765910968,
-      "grad_norm": 1.1666908264160156,
-      "learning_rate": 7.348678601875533e-06,
-      "loss": 0.1782,
-      "step": 1555
-    },
-    {
-      "epoch": 0.26529557503263446,
-      "grad_norm": 1.2111077308654785,
-      "learning_rate": 7.346973572037511e-06,
-      "loss": 0.2437,
-      "step": 1556
-    },
-    {
-      "epoch": 0.26546607347417217,
-      "grad_norm": 0.5697571635246277,
-      "learning_rate": 7.345268542199489e-06,
-      "loss": 0.1359,
-      "step": 1557
-    },
-    {
-      "epoch": 0.26563657191570983,
-      "grad_norm": 0.6316619515419006,
-      "learning_rate": 7.343563512361467e-06,
-      "loss": 0.1581,
-      "step": 1558
-    },
-    {
-      "epoch": 0.26580707035724754,
-      "grad_norm": 0.9178199768066406,
-      "learning_rate": 7.341858482523445e-06,
-      "loss": 0.1383,
-      "step": 1559
-    },
-    {
-      "epoch": 0.2659775687987852,
-      "grad_norm": 0.9397381544113159,
-      "learning_rate": 7.340153452685422e-06,
-      "loss": 0.1028,
-      "step": 1560
-    },
-    {
-      "epoch": 0.26614806724032286,
-      "grad_norm": 0.6357908248901367,
-      "learning_rate": 7.338448422847401e-06,
-      "loss": 0.1687,
-      "step": 1561
-    },
-    {
-      "epoch": 0.2663185656818606,
-      "grad_norm": 1.449703335762024,
-      "learning_rate": 7.336743393009378e-06,
-      "loss": 0.2689,
-      "step": 1562
-    },
-    {
-      "epoch": 0.26648906412339823,
-      "grad_norm": 1.1114320755004883,
-      "learning_rate": 7.335038363171356e-06,
-      "loss": 0.2566,
-      "step": 1563
-    },
-    {
-      "epoch": 0.26665956256493595,
-      "grad_norm": 0.7258211970329285,
-      "learning_rate": 7.333333333333333e-06,
-      "loss": 0.1485,
-      "step": 1564
-    },
-    {
-      "epoch": 0.2668300610064736,
-      "grad_norm": 0.7410526871681213,
-      "learning_rate": 7.331628303495312e-06,
-      "loss": 0.1826,
-      "step": 1565
-    },
-    {
-      "epoch": 0.2670005594480113,
-      "grad_norm": 0.691279411315918,
-      "learning_rate": 7.32992327365729e-06,
-      "loss": 0.174,
-      "step": 1566
-    },
-    {
-      "epoch": 0.267171057889549,
-      "grad_norm": 0.6794247031211853,
-      "learning_rate": 7.328218243819267e-06,
-      "loss": 0.1715,
-      "step": 1567
-    },
-    {
-      "epoch": 0.26734155633108664,
-      "grad_norm": 0.7400369048118591,
-      "learning_rate": 7.326513213981246e-06,
-      "loss": 0.106,
-      "step": 1568
-    },
-    {
-      "epoch": 0.26751205477262435,
-      "grad_norm": 0.5771182179450989,
-      "learning_rate": 7.324808184143223e-06,
-      "loss": 0.1487,
-      "step": 1569
-    },
-    {
-      "epoch": 0.267682553214162,
-      "grad_norm": 0.8204482197761536,
-      "learning_rate": 7.323103154305201e-06,
-      "loss": 0.1984,
-      "step": 1570
-    },
-    {
-      "epoch": 0.2678530516556997,
-      "grad_norm": 0.8229162693023682,
-      "learning_rate": 7.321398124467178e-06,
-      "loss": 0.2344,
-      "step": 1571
-    },
-    {
-      "epoch": 0.2680235500972374,
-      "grad_norm": 0.6249790191650391,
-      "learning_rate": 7.319693094629157e-06,
-      "loss": 0.1283,
-      "step": 1572
-    },
-    {
-      "epoch": 0.2681940485387751,
-      "grad_norm": 0.9039957523345947,
-      "learning_rate": 7.317988064791135e-06,
-      "loss": 0.1595,
-      "step": 1573
-    },
-    {
-      "epoch": 0.26836454698031276,
-      "grad_norm": 1.0393831729888916,
-      "learning_rate": 7.316283034953113e-06,
-      "loss": 0.2209,
-      "step": 1574
-    },
-    {
-      "epoch": 0.2685350454218504,
-      "grad_norm": 0.6935752034187317,
-      "learning_rate": 7.3145780051150906e-06,
-      "loss": 0.1117,
-      "step": 1575
-    },
-    {
-      "epoch": 0.26870554386338813,
-      "grad_norm": 0.6081529855728149,
-      "learning_rate": 7.312872975277068e-06,
-      "loss": 0.0929,
-      "step": 1576
-    },
-    {
-      "epoch": 0.2688760423049258,
-      "grad_norm": 0.6110521554946899,
-      "learning_rate": 7.3111679454390465e-06,
-      "loss": 0.1149,
-      "step": 1577
-    },
-    {
-      "epoch": 0.2690465407464635,
-      "grad_norm": 0.9071523547172546,
-      "learning_rate": 7.3094629156010236e-06,
-      "loss": 0.1966,
-      "step": 1578
-    },
-    {
-      "epoch": 0.26921703918800116,
-      "grad_norm": 0.7462790608406067,
-      "learning_rate": 7.3077578857630015e-06,
-      "loss": 0.0771,
-      "step": 1579
-    },
-    {
-      "epoch": 0.2693875376295389,
-      "grad_norm": 0.6232742071151733,
-      "learning_rate": 7.306052855924979e-06,
-      "loss": 0.1326,
-      "step": 1580
-    },
-    {
-      "epoch": 0.26955803607107653,
-      "grad_norm": 0.685059130191803,
-      "learning_rate": 7.304347826086957e-06,
-      "loss": 0.1197,
-      "step": 1581
-    },
-    {
-      "epoch": 0.2697285345126142,
-      "grad_norm": 0.9145820140838623,
-      "learning_rate": 7.302642796248935e-06,
-      "loss": 0.1311,
-      "step": 1582
-    },
-    {
-      "epoch": 0.2698990329541519,
-      "grad_norm": 1.031315803527832,
-      "learning_rate": 7.3009377664109125e-06,
-      "loss": 0.2123,
-      "step": 1583
-    },
-    {
-      "epoch": 0.27006953139568957,
-      "grad_norm": 1.0389862060546875,
-      "learning_rate": 7.299232736572891e-06,
-      "loss": 0.2036,
-      "step": 1584
-    },
-    {
-      "epoch": 0.2702400298372273,
-      "grad_norm": 1.0475060939788818,
-      "learning_rate": 7.297527706734868e-06,
-      "loss": 0.1878,
-      "step": 1585
-    },
-    {
-      "epoch": 0.27041052827876494,
-      "grad_norm": 0.5867076516151428,
-      "learning_rate": 7.295822676896846e-06,
-      "loss": 0.096,
-      "step": 1586
-    },
-    {
-      "epoch": 0.27058102672030265,
-      "grad_norm": 0.6274494528770447,
-      "learning_rate": 7.294117647058823e-06,
-      "loss": 0.1282,
-      "step": 1587
-    },
-    {
-      "epoch": 0.2707515251618403,
-      "grad_norm": 0.750312089920044,
-      "learning_rate": 7.292412617220802e-06,
-      "loss": 0.1011,
-      "step": 1588
-    },
-    {
-      "epoch": 0.270922023603378,
-      "grad_norm": 0.9816113710403442,
-      "learning_rate": 7.290707587382779e-06,
-      "loss": 0.0997,
-      "step": 1589
-    },
-    {
-      "epoch": 0.2710925220449157,
-      "grad_norm": 1.259264588356018,
-      "learning_rate": 7.289002557544757e-06,
-      "loss": 0.2381,
-      "step": 1590
-    },
-    {
-      "epoch": 0.27126302048645334,
-      "grad_norm": 0.7334678173065186,
-      "learning_rate": 7.287297527706736e-06,
-      "loss": 0.137,
-      "step": 1591
-    },
-    {
-      "epoch": 0.27143351892799106,
-      "grad_norm": 0.5797063112258911,
-      "learning_rate": 7.285592497868713e-06,
-      "loss": 0.101,
-      "step": 1592
-    },
-    {
-      "epoch": 0.2716040173695287,
-      "grad_norm": 0.707748532295227,
-      "learning_rate": 7.283887468030691e-06,
-      "loss": 0.1033,
-      "step": 1593
-    },
-    {
-      "epoch": 0.27177451581106643,
-      "grad_norm": 0.792129635810852,
-      "learning_rate": 7.282182438192669e-06,
-      "loss": 0.1575,
-      "step": 1594
-    },
-    {
-      "epoch": 0.2719450142526041,
-      "grad_norm": 0.7394453287124634,
-      "learning_rate": 7.280477408354647e-06,
-      "loss": 0.159,
-      "step": 1595
-    },
-    {
-      "epoch": 0.2721155126941418,
-      "grad_norm": 0.6427018642425537,
-      "learning_rate": 7.278772378516624e-06,
-      "loss": 0.114,
-      "step": 1596
-    },
-    {
-      "epoch": 0.27228601113567946,
-      "grad_norm": 0.8114389181137085,
-      "learning_rate": 7.277067348678603e-06,
-      "loss": 0.1574,
-      "step": 1597
-    },
-    {
-      "epoch": 0.2724565095772171,
-      "grad_norm": 0.5855715274810791,
-      "learning_rate": 7.27536231884058e-06,
-      "loss": 0.1004,
-      "step": 1598
-    },
-    {
-      "epoch": 0.27262700801875484,
-      "grad_norm": 1.1192022562026978,
-      "learning_rate": 7.273657289002558e-06,
-      "loss": 0.1619,
-      "step": 1599
-    },
-    {
-      "epoch": 0.2727975064602925,
-      "grad_norm": 0.8948620557785034,
-      "learning_rate": 7.271952259164537e-06,
-      "loss": 0.1634,
-      "step": 1600
-    },
-    {
-      "epoch": 0.2727975064602925,
-      "eval_f1_score": 0.2809917355371901,
-      "eval_loss": 0.16055278480052948,
-      "eval_runtime": 182.5791,
-      "eval_samples_per_second": 54.771,
-      "eval_steps_per_second": 3.423,
-      "step": 1600
-    },
-    {
-      "epoch": 0.2729680049018302,
-      "grad_norm": 0.995793879032135,
-      "learning_rate": 7.270247229326514e-06,
-      "loss": 0.0949,
-      "step": 1601
-    },
-    {
-      "epoch": 0.27313850334336787,
-      "grad_norm": 0.572679877281189,
-      "learning_rate": 7.268542199488492e-06,
-      "loss": 0.1324,
-      "step": 1602
-    },
-    {
-      "epoch": 0.2733090017849056,
-      "grad_norm": 1.449594497680664,
-      "learning_rate": 7.266837169650469e-06,
-      "loss": 0.1787,
-      "step": 1603
-    },
-    {
-      "epoch": 0.27347950022644324,
-      "grad_norm": 0.6399373412132263,
-      "learning_rate": 7.265132139812448e-06,
-      "loss": 0.1394,
-      "step": 1604
-    },
-    {
-      "epoch": 0.2736499986679809,
-      "grad_norm": 1.0330250263214111,
-      "learning_rate": 7.263427109974425e-06,
-      "loss": 0.111,
-      "step": 1605
-    },
-    {
-      "epoch": 0.2738204971095186,
-      "grad_norm": 0.699213981628418,
-      "learning_rate": 7.261722080136403e-06,
-      "loss": 0.1409,
-      "step": 1606
-    },
-    {
-      "epoch": 0.27399099555105627,
-      "grad_norm": 0.8001547455787659,
-      "learning_rate": 7.2600170502983815e-06,
-      "loss": 0.1732,
-      "step": 1607
-    },
-    {
-      "epoch": 0.274161493992594,
-      "grad_norm": 1.5362522602081299,
-      "learning_rate": 7.258312020460359e-06,
-      "loss": 0.1623,
-      "step": 1608
-    },
-    {
-      "epoch": 0.27433199243413164,
-      "grad_norm": 0.6935368776321411,
-      "learning_rate": 7.2566069906223366e-06,
-      "loss": 0.1215,
-      "step": 1609
-    },
-    {
-      "epoch": 0.27450249087566936,
-      "grad_norm": 0.9737477898597717,
-      "learning_rate": 7.2549019607843145e-06,
-      "loss": 0.1726,
-      "step": 1610
-    },
-    {
-      "epoch": 0.274672989317207,
-      "grad_norm": 1.0231482982635498,
-      "learning_rate": 7.2531969309462924e-06,
-      "loss": 0.1836,
-      "step": 1611
-    },
-    {
-      "epoch": 0.27484348775874473,
-      "grad_norm": 0.7062311172485352,
-      "learning_rate": 7.2514919011082696e-06,
-      "loss": 0.1083,
-      "step": 1612
-    },
-    {
-      "epoch": 0.2750139862002824,
-      "grad_norm": 0.964033842086792,
-      "learning_rate": 7.249786871270248e-06,
-      "loss": 0.2435,
-      "step": 1613
-    },
-    {
-      "epoch": 0.27518448464182005,
-      "grad_norm": 0.7868807911872864,
-      "learning_rate": 7.2480818414322254e-06,
-      "loss": 0.181,
-      "step": 1614
-    },
-    {
-      "epoch": 0.27535498308335776,
-      "grad_norm": 0.9056547284126282,
-      "learning_rate": 7.246376811594203e-06,
-      "loss": 0.1895,
-      "step": 1615
-    },
-    {
-      "epoch": 0.2755254815248954,
-      "grad_norm": 1.1099858283996582,
-      "learning_rate": 7.244671781756182e-06,
-      "loss": 0.173,
-      "step": 1616
-    },
-    {
-      "epoch": 0.27569597996643314,
-      "grad_norm": 1.1758391857147217,
-      "learning_rate": 7.242966751918159e-06,
-      "loss": 0.1648,
-      "step": 1617
-    },
-    {
-      "epoch": 0.2758664784079708,
-      "grad_norm": 0.8921638131141663,
-      "learning_rate": 7.241261722080137e-06,
-      "loss": 0.1423,
-      "step": 1618
-    },
-    {
-      "epoch": 0.2760369768495085,
-      "grad_norm": 0.8928386569023132,
-      "learning_rate": 7.239556692242114e-06,
-      "loss": 0.192,
-      "step": 1619
-    },
-    {
-      "epoch": 0.27620747529104617,
-      "grad_norm": 0.9611541628837585,
-      "learning_rate": 7.237851662404093e-06,
-      "loss": 0.1305,
-      "step": 1620
-    },
-    {
-      "epoch": 0.2763779737325838,
-      "grad_norm": 0.8896896839141846,
-      "learning_rate": 7.23614663256607e-06,
-      "loss": 0.1177,
-      "step": 1621
-    },
-    {
-      "epoch": 0.27654847217412154,
-      "grad_norm": 0.7848691940307617,
-      "learning_rate": 7.234441602728048e-06,
-      "loss": 0.1609,
-      "step": 1622
-    },
-    {
-      "epoch": 0.2767189706156592,
-      "grad_norm": 0.5970712304115295,
-      "learning_rate": 7.232736572890025e-06,
-      "loss": 0.0916,
-      "step": 1623
-    },
-    {
-      "epoch": 0.2768894690571969,
-      "grad_norm": 0.9145472645759583,
-      "learning_rate": 7.231031543052004e-06,
-      "loss": 0.1655,
-      "step": 1624
-    },
-    {
-      "epoch": 0.27705996749873457,
-      "grad_norm": 0.7860545516014099,
-      "learning_rate": 7.229326513213982e-06,
-      "loss": 0.1545,
-      "step": 1625
-    },
-    {
-      "epoch": 0.2772304659402723,
-      "grad_norm": 0.5441620349884033,
-      "learning_rate": 7.227621483375959e-06,
-      "loss": 0.0995,
-      "step": 1626
-    },
-    {
-      "epoch": 0.27740096438180994,
-      "grad_norm": 1.0052058696746826,
-      "learning_rate": 7.225916453537938e-06,
-      "loss": 0.2069,
-      "step": 1627
-    },
-    {
-      "epoch": 0.2775714628233476,
-      "grad_norm": 1.0887932777404785,
-      "learning_rate": 7.224211423699915e-06,
-      "loss": 0.1708,
-      "step": 1628
-    },
-    {
-      "epoch": 0.2777419612648853,
-      "grad_norm": 0.9987654685974121,
-      "learning_rate": 7.222506393861893e-06,
-      "loss": 0.1747,
-      "step": 1629
-    },
-    {
-      "epoch": 0.277912459706423,
-      "grad_norm": 1.6356117725372314,
-      "learning_rate": 7.220801364023871e-06,
-      "loss": 0.296,
-      "step": 1630
-    },
-    {
-      "epoch": 0.2780829581479607,
-      "grad_norm": 1.5601258277893066,
-      "learning_rate": 7.219096334185849e-06,
-      "loss": 0.2036,
-      "step": 1631
-    },
-    {
-      "epoch": 0.27825345658949835,
-      "grad_norm": 0.47777241468429565,
-      "learning_rate": 7.217391304347827e-06,
-      "loss": 0.1113,
-      "step": 1632
-    },
-    {
-      "epoch": 0.27842395503103606,
-      "grad_norm": 0.7547958493232727,
-      "learning_rate": 7.215686274509805e-06,
-      "loss": 0.1498,
-      "step": 1633
-    },
-    {
-      "epoch": 0.2785944534725737,
-      "grad_norm": 0.6811045408248901,
-      "learning_rate": 7.213981244671783e-06,
-      "loss": 0.1155,
-      "step": 1634
-    },
-    {
-      "epoch": 0.27876495191411144,
-      "grad_norm": 0.8681180477142334,
-      "learning_rate": 7.21227621483376e-06,
-      "loss": 0.1752,
-      "step": 1635
-    },
-    {
-      "epoch": 0.2789354503556491,
-      "grad_norm": 0.6963757276535034,
-      "learning_rate": 7.210571184995739e-06,
-      "loss": 0.1981,
-      "step": 1636
-    },
-    {
-      "epoch": 0.27910594879718675,
-      "grad_norm": 0.6927238702774048,
-      "learning_rate": 7.208866155157716e-06,
-      "loss": 0.1345,
-      "step": 1637
-    },
-    {
-      "epoch": 0.27927644723872447,
-      "grad_norm": 0.7065186500549316,
-      "learning_rate": 7.207161125319694e-06,
-      "loss": 0.1286,
-      "step": 1638
-    },
-    {
-      "epoch": 0.2794469456802621,
-      "grad_norm": 0.6613034605979919,
-      "learning_rate": 7.205456095481671e-06,
-      "loss": 0.1637,
-      "step": 1639
-    },
-    {
-      "epoch": 0.27961744412179984,
-      "grad_norm": 0.7331064343452454,
-      "learning_rate": 7.2037510656436495e-06,
-      "loss": 0.1949,
-      "step": 1640
-    },
-    {
-      "epoch": 0.2797879425633375,
-      "grad_norm": 0.7222153544425964,
-      "learning_rate": 7.2020460358056275e-06,
-      "loss": 0.0943,
-      "step": 1641
-    },
-    {
-      "epoch": 0.2799584410048752,
-      "grad_norm": 0.8640736937522888,
-      "learning_rate": 7.200341005967605e-06,
-      "loss": 0.1327,
-      "step": 1642
-    },
-    {
-      "epoch": 0.2801289394464129,
-      "grad_norm": 0.7286602854728699,
-      "learning_rate": 7.198635976129583e-06,
-      "loss": 0.1188,
-      "step": 1643
-    },
-    {
-      "epoch": 0.28029943788795053,
-      "grad_norm": 1.4019544124603271,
-      "learning_rate": 7.1969309462915605e-06,
-      "loss": 0.1201,
-      "step": 1644
-    },
-    {
-      "epoch": 0.28046993632948825,
-      "grad_norm": 0.6155040860176086,
-      "learning_rate": 7.1952259164535384e-06,
-      "loss": 0.1034,
-      "step": 1645
-    },
-    {
-      "epoch": 0.2806404347710259,
-      "grad_norm": 0.6862358450889587,
-      "learning_rate": 7.1935208866155155e-06,
-      "loss": 0.0757,
-      "step": 1646
-    },
-    {
-      "epoch": 0.2808109332125636,
-      "grad_norm": 0.4926810562610626,
-      "learning_rate": 7.191815856777494e-06,
-      "loss": 0.0643,
-      "step": 1647
-    },
-    {
-      "epoch": 0.2809814316541013,
-      "grad_norm": 0.5891478061676025,
-      "learning_rate": 7.1901108269394714e-06,
-      "loss": 0.0608,
-      "step": 1648
-    },
-    {
-      "epoch": 0.281151930095639,
-      "grad_norm": 0.7541805505752563,
-      "learning_rate": 7.18840579710145e-06,
-      "loss": 0.1001,
-      "step": 1649
-    },
-    {
-      "epoch": 0.28132242853717665,
-      "grad_norm": 0.7641288638114929,
-      "learning_rate": 7.186700767263428e-06,
-      "loss": 0.156,
-      "step": 1650
-    },
-    {
-      "epoch": 0.2814929269787143,
-      "grad_norm": 1.1662489175796509,
-      "learning_rate": 7.184995737425405e-06,
-      "loss": 0.1361,
-      "step": 1651
-    },
-    {
-      "epoch": 0.281663425420252,
-      "grad_norm": 1.0508551597595215,
-      "learning_rate": 7.183290707587384e-06,
-      "loss": 0.1872,
-      "step": 1652
-    },
-    {
-      "epoch": 0.2818339238617897,
-      "grad_norm": 0.7512367963790894,
-      "learning_rate": 7.181585677749361e-06,
-      "loss": 0.1021,
-      "step": 1653
-    },
-    {
-      "epoch": 0.2820044223033274,
-      "grad_norm": 0.6517552137374878,
-      "learning_rate": 7.179880647911339e-06,
-      "loss": 0.0645,
-      "step": 1654
-    },
-    {
-      "epoch": 0.28217492074486505,
-      "grad_norm": 0.8589548468589783,
-      "learning_rate": 7.178175618073316e-06,
-      "loss": 0.2049,
-      "step": 1655
-    },
-    {
-      "epoch": 0.28234541918640277,
-      "grad_norm": 0.8293384313583374,
-      "learning_rate": 7.176470588235295e-06,
-      "loss": 0.1309,
-      "step": 1656
-    },
-    {
-      "epoch": 0.2825159176279404,
-      "grad_norm": 1.2787649631500244,
-      "learning_rate": 7.174765558397273e-06,
-      "loss": 0.1629,
-      "step": 1657
-    },
-    {
-      "epoch": 0.28268641606947814,
-      "grad_norm": 0.6129804849624634,
-      "learning_rate": 7.17306052855925e-06,
-      "loss": 0.1121,
-      "step": 1658
-    },
-    {
-      "epoch": 0.2828569145110158,
-      "grad_norm": 0.8967618346214294,
-      "learning_rate": 7.171355498721229e-06,
-      "loss": 0.1106,
-      "step": 1659
-    },
-    {
-      "epoch": 0.28302741295255346,
-      "grad_norm": 0.7329991459846497,
-      "learning_rate": 7.169650468883206e-06,
-      "loss": 0.0529,
-      "step": 1660
-    },
-    {
-      "epoch": 0.2831979113940912,
-      "grad_norm": 0.9802336096763611,
-      "learning_rate": 7.167945439045184e-06,
-      "loss": 0.0964,
-      "step": 1661
-    },
-    {
-      "epoch": 0.28336840983562883,
-      "grad_norm": 2.165433168411255,
-      "learning_rate": 7.166240409207161e-06,
-      "loss": 0.1989,
-      "step": 1662
-    },
-    {
-      "epoch": 0.28353890827716655,
-      "grad_norm": 0.9615960717201233,
-      "learning_rate": 7.16453537936914e-06,
-      "loss": 0.1118,
-      "step": 1663
-    },
-    {
-      "epoch": 0.2837094067187042,
-      "grad_norm": 0.830369234085083,
-      "learning_rate": 7.162830349531117e-06,
-      "loss": 0.0781,
-      "step": 1664
-    },
-    {
-      "epoch": 0.2838799051602419,
-      "grad_norm": 1.2290384769439697,
-      "learning_rate": 7.161125319693095e-06,
-      "loss": 0.1,
-      "step": 1665
-    },
-    {
-      "epoch": 0.2840504036017796,
-      "grad_norm": 0.7927495837211609,
-      "learning_rate": 7.159420289855074e-06,
-      "loss": 0.0979,
-      "step": 1666
-    },
-    {
-      "epoch": 0.28422090204331724,
-      "grad_norm": 1.4657909870147705,
-      "learning_rate": 7.157715260017051e-06,
-      "loss": 0.1313,
-      "step": 1667
-    },
-    {
-      "epoch": 0.28439140048485495,
-      "grad_norm": 1.055486798286438,
-      "learning_rate": 7.156010230179029e-06,
-      "loss": 0.0692,
-      "step": 1668
-    },
-    {
-      "epoch": 0.2845618989263926,
-      "grad_norm": 1.2817178964614868,
-      "learning_rate": 7.154305200341007e-06,
-      "loss": 0.1194,
-      "step": 1669
-    },
-    {
-      "epoch": 0.2847323973679303,
-      "grad_norm": 1.1928097009658813,
-      "learning_rate": 7.152600170502985e-06,
-      "loss": 0.1265,
-      "step": 1670
-    },
-    {
-      "epoch": 0.284902895809468,
-      "grad_norm": 1.5519382953643799,
-      "learning_rate": 7.150895140664962e-06,
-      "loss": 0.1602,
-      "step": 1671
-    },
-    {
-      "epoch": 0.2850733942510057,
-      "grad_norm": 1.305989384651184,
-      "learning_rate": 7.1491901108269405e-06,
-      "loss": 0.2005,
-      "step": 1672
-    },
-    {
-      "epoch": 0.28524389269254335,
-      "grad_norm": 1.3667876720428467,
-      "learning_rate": 7.147485080988918e-06,
-      "loss": 0.212,
-      "step": 1673
-    },
-    {
-      "epoch": 0.285414391134081,
-      "grad_norm": 1.4790457487106323,
-      "learning_rate": 7.1457800511508955e-06,
-      "loss": 0.2204,
-      "step": 1674
-    },
-    {
-      "epoch": 0.28558488957561873,
-      "grad_norm": 1.683897852897644,
-      "learning_rate": 7.144075021312874e-06,
-      "loss": 0.1859,
-      "step": 1675
-    },
-    {
-      "epoch": 0.2857553880171564,
-      "grad_norm": 0.9878926873207092,
-      "learning_rate": 7.1423699914748514e-06,
-      "loss": 0.1382,
-      "step": 1676
-    },
-    {
-      "epoch": 0.2859258864586941,
-      "grad_norm": 1.0337659120559692,
-      "learning_rate": 7.140664961636829e-06,
-      "loss": 0.0803,
-      "step": 1677
-    },
-    {
-      "epoch": 0.28609638490023176,
-      "grad_norm": 1.0796700716018677,
-      "learning_rate": 7.1389599317988065e-06,
-      "loss": 0.0952,
-      "step": 1678
-    },
-    {
-      "epoch": 0.2862668833417695,
-      "grad_norm": 1.966192364692688,
-      "learning_rate": 7.137254901960785e-06,
-      "loss": 0.2293,
-      "step": 1679
-    },
-    {
-      "epoch": 0.28643738178330713,
-      "grad_norm": 0.7053760290145874,
-      "learning_rate": 7.135549872122762e-06,
-      "loss": 0.1088,
-      "step": 1680
-    },
-    {
-      "epoch": 0.28660788022484485,
-      "grad_norm": 1.5122227668762207,
-      "learning_rate": 7.13384484228474e-06,
-      "loss": 0.198,
-      "step": 1681
-    },
-    {
-      "epoch": 0.2867783786663825,
-      "grad_norm": 0.47033724188804626,
-      "learning_rate": 7.132139812446719e-06,
-      "loss": 0.0445,
-      "step": 1682
-    },
-    {
-      "epoch": 0.28694887710792016,
-      "grad_norm": 0.885513961315155,
-      "learning_rate": 7.130434782608696e-06,
-      "loss": 0.0858,
-      "step": 1683
-    },
-    {
-      "epoch": 0.2871193755494579,
-      "grad_norm": 1.2802006006240845,
-      "learning_rate": 7.128729752770674e-06,
-      "loss": 0.1765,
-      "step": 1684
-    },
-    {
-      "epoch": 0.28728987399099554,
-      "grad_norm": 0.9488986730575562,
-      "learning_rate": 7.127024722932651e-06,
-      "loss": 0.1583,
-      "step": 1685
-    },
-    {
-      "epoch": 0.28746037243253325,
-      "grad_norm": 1.2158293724060059,
-      "learning_rate": 7.12531969309463e-06,
-      "loss": 0.1831,
-      "step": 1686
-    },
-    {
-      "epoch": 0.2876308708740709,
-      "grad_norm": 1.6682029962539673,
-      "learning_rate": 7.123614663256607e-06,
-      "loss": 0.234,
-      "step": 1687
-    },
-    {
-      "epoch": 0.2878013693156086,
-      "grad_norm": 0.6801013350486755,
-      "learning_rate": 7.121909633418586e-06,
-      "loss": 0.1266,
-      "step": 1688
-    },
-    {
-      "epoch": 0.2879718677571463,
-      "grad_norm": 0.8549294471740723,
-      "learning_rate": 7.120204603580563e-06,
-      "loss": 0.2075,
-      "step": 1689
-    },
-    {
-      "epoch": 0.28814236619868394,
-      "grad_norm": 0.9746972918510437,
-      "learning_rate": 7.118499573742541e-06,
-      "loss": 0.136,
-      "step": 1690
-    },
-    {
-      "epoch": 0.28831286464022166,
-      "grad_norm": 0.7878130674362183,
-      "learning_rate": 7.11679454390452e-06,
-      "loss": 0.1134,
-      "step": 1691
-    },
-    {
-      "epoch": 0.2884833630817593,
-      "grad_norm": 0.9971168041229248,
-      "learning_rate": 7.115089514066497e-06,
-      "loss": 0.1872,
-      "step": 1692
-    },
-    {
-      "epoch": 0.28865386152329703,
-      "grad_norm": 0.8819569945335388,
-      "learning_rate": 7.113384484228475e-06,
-      "loss": 0.107,
-      "step": 1693
-    },
-    {
-      "epoch": 0.2888243599648347,
-      "grad_norm": 1.1124303340911865,
-      "learning_rate": 7.111679454390452e-06,
-      "loss": 0.2033,
-      "step": 1694
-    },
-    {
-      "epoch": 0.2889948584063724,
-      "grad_norm": 0.7073795199394226,
-      "learning_rate": 7.109974424552431e-06,
-      "loss": 0.0919,
-      "step": 1695
-    },
-    {
-      "epoch": 0.28916535684791006,
-      "grad_norm": 1.197323203086853,
-      "learning_rate": 7.108269394714408e-06,
-      "loss": 0.1908,
-      "step": 1696
-    },
-    {
-      "epoch": 0.2893358552894477,
-      "grad_norm": 0.7450862526893616,
-      "learning_rate": 7.106564364876386e-06,
-      "loss": 0.1421,
-      "step": 1697
-    },
-    {
-      "epoch": 0.28950635373098543,
-      "grad_norm": 1.0751428604125977,
-      "learning_rate": 7.104859335038363e-06,
-      "loss": 0.1508,
-      "step": 1698
-    },
-    {
-      "epoch": 0.2896768521725231,
-      "grad_norm": 0.6413664817810059,
-      "learning_rate": 7.103154305200342e-06,
-      "loss": 0.1564,
-      "step": 1699
-    },
-    {
-      "epoch": 0.2898473506140608,
-      "grad_norm": 0.7627037167549133,
-      "learning_rate": 7.10144927536232e-06,
-      "loss": 0.1049,
-      "step": 1700
-    },
-    {
-      "epoch": 0.2898473506140608,
-      "eval_f1_score": 0.2712765957446808,
-      "eval_loss": 0.16056498885154724,
-      "eval_runtime": 182.6358,
-      "eval_samples_per_second": 54.754,
-      "eval_steps_per_second": 3.422,
-      "step": 1700
-    },
-    {
-      "epoch": 0.29001784905559846,
-      "grad_norm": 0.7326638102531433,
-      "learning_rate": 7.099744245524297e-06,
-      "loss": 0.0934,
-      "step": 1701
-    },
-    {
-      "epoch": 0.2901883474971362,
-      "grad_norm": 0.8272231817245483,
-      "learning_rate": 7.0980392156862755e-06,
-      "loss": 0.1346,
-      "step": 1702
-    },
-    {
-      "epoch": 0.29035884593867384,
-      "grad_norm": 0.7498846054077148,
-      "learning_rate": 7.096334185848253e-06,
-      "loss": 0.1581,
-      "step": 1703
-    },
-    {
-      "epoch": 0.29052934438021155,
-      "grad_norm": 0.7257694602012634,
-      "learning_rate": 7.0946291560102306e-06,
-      "loss": 0.1084,
-      "step": 1704
-    },
-    {
-      "epoch": 0.2906998428217492,
-      "grad_norm": 0.9842420220375061,
-      "learning_rate": 7.0929241261722085e-06,
-      "loss": 0.079,
-      "step": 1705
-    },
-    {
-      "epoch": 0.29087034126328687,
-      "grad_norm": 0.729982852935791,
-      "learning_rate": 7.0912190963341865e-06,
-      "loss": 0.1044,
-      "step": 1706
-    },
-    {
-      "epoch": 0.2910408397048246,
-      "grad_norm": 1.0497548580169678,
-      "learning_rate": 7.089514066496164e-06,
-      "loss": 0.2153,
-      "step": 1707
-    },
-    {
-      "epoch": 0.29121133814636224,
-      "grad_norm": 0.7500994801521301,
-      "learning_rate": 7.087809036658142e-06,
-      "loss": 0.0815,
-      "step": 1708
-    },
-    {
-      "epoch": 0.29138183658789996,
-      "grad_norm": 0.6701683402061462,
-      "learning_rate": 7.08610400682012e-06,
-      "loss": 0.0694,
-      "step": 1709
-    },
-    {
-      "epoch": 0.2915523350294376,
-      "grad_norm": 1.0692698955535889,
-      "learning_rate": 7.084398976982097e-06,
-      "loss": 0.1516,
-      "step": 1710
-    },
-    {
-      "epoch": 0.29172283347097533,
-      "grad_norm": 1.1487675905227661,
-      "learning_rate": 7.082693947144076e-06,
-      "loss": 0.1954,
-      "step": 1711
-    },
-    {
-      "epoch": 0.291893331912513,
-      "grad_norm": 1.0295099020004272,
-      "learning_rate": 7.080988917306053e-06,
-      "loss": 0.1305,
-      "step": 1712
-    },
-    {
-      "epoch": 0.29206383035405065,
-      "grad_norm": 1.4119102954864502,
-      "learning_rate": 7.079283887468031e-06,
-      "loss": 0.1735,
-      "step": 1713
-    },
-    {
-      "epoch": 0.29223432879558836,
-      "grad_norm": 1.2772480249404907,
-      "learning_rate": 7.077578857630008e-06,
-      "loss": 0.1499,
-      "step": 1714
-    },
-    {
-      "epoch": 0.292404827237126,
-      "grad_norm": 1.1898237466812134,
-      "learning_rate": 7.075873827791987e-06,
-      "loss": 0.1384,
-      "step": 1715
-    },
-    {
-      "epoch": 0.29257532567866373,
-      "grad_norm": 1.3872730731964111,
-      "learning_rate": 7.074168797953965e-06,
-      "loss": 0.1784,
-      "step": 1716
-    },
-    {
-      "epoch": 0.2927458241202014,
-      "grad_norm": 0.9282299280166626,
-      "learning_rate": 7.072463768115942e-06,
-      "loss": 0.1353,
-      "step": 1717
-    },
-    {
-      "epoch": 0.2929163225617391,
-      "grad_norm": 1.004987359046936,
-      "learning_rate": 7.070758738277921e-06,
-      "loss": 0.0914,
-      "step": 1718
-    },
-    {
-      "epoch": 0.29308682100327677,
-      "grad_norm": 0.8638032674789429,
-      "learning_rate": 7.069053708439898e-06,
-      "loss": 0.1113,
-      "step": 1719
-    },
-    {
-      "epoch": 0.2932573194448144,
-      "grad_norm": 0.935568630695343,
-      "learning_rate": 7.067348678601876e-06,
-      "loss": 0.1775,
-      "step": 1720
-    },
-    {
-      "epoch": 0.29342781788635214,
-      "grad_norm": 0.967147171497345,
-      "learning_rate": 7.065643648763853e-06,
-      "loss": 0.2255,
-      "step": 1721
-    },
-    {
-      "epoch": 0.2935983163278898,
-      "grad_norm": 1.0089436769485474,
-      "learning_rate": 7.063938618925832e-06,
-      "loss": 0.0984,
-      "step": 1722
-    },
-    {
-      "epoch": 0.2937688147694275,
-      "grad_norm": 1.053886890411377,
-      "learning_rate": 7.062233589087809e-06,
-      "loss": 0.1922,
-      "step": 1723
-    },
-    {
-      "epoch": 0.29393931321096517,
-      "grad_norm": 1.6038528680801392,
-      "learning_rate": 7.060528559249788e-06,
-      "loss": 0.1198,
-      "step": 1724
-    },
-    {
-      "epoch": 0.2941098116525029,
-      "grad_norm": 0.8823066353797913,
-      "learning_rate": 7.058823529411766e-06,
-      "loss": 0.1262,
-      "step": 1725
-    },
-    {
-      "epoch": 0.29428031009404054,
-      "grad_norm": 1.1137727499008179,
-      "learning_rate": 7.057118499573743e-06,
-      "loss": 0.1021,
-      "step": 1726
-    },
-    {
-      "epoch": 0.29445080853557826,
-      "grad_norm": 0.8181190490722656,
-      "learning_rate": 7.055413469735722e-06,
-      "loss": 0.1207,
-      "step": 1727
-    },
-    {
-      "epoch": 0.2946213069771159,
-      "grad_norm": 1.24038565158844,
-      "learning_rate": 7.053708439897699e-06,
-      "loss": 0.1683,
-      "step": 1728
-    },
-    {
-      "epoch": 0.2947918054186536,
-      "grad_norm": 0.7535143494606018,
-      "learning_rate": 7.052003410059677e-06,
-      "loss": 0.128,
-      "step": 1729
-    },
-    {
-      "epoch": 0.2949623038601913,
-      "grad_norm": 0.8402575254440308,
-      "learning_rate": 7.050298380221654e-06,
-      "loss": 0.163,
-      "step": 1730
-    },
-    {
-      "epoch": 0.29513280230172895,
-      "grad_norm": 1.4767165184020996,
-      "learning_rate": 7.048593350383633e-06,
-      "loss": 0.2026,
-      "step": 1731
-    },
-    {
-      "epoch": 0.29530330074326666,
-      "grad_norm": 0.7595881223678589,
-      "learning_rate": 7.0468883205456106e-06,
-      "loss": 0.1165,
-      "step": 1732
-    },
-    {
-      "epoch": 0.2954737991848043,
-      "grad_norm": 0.8625977039337158,
-      "learning_rate": 7.045183290707588e-06,
-      "loss": 0.166,
-      "step": 1733
-    },
-    {
-      "epoch": 0.29564429762634203,
-      "grad_norm": 0.5962086915969849,
-      "learning_rate": 7.0434782608695665e-06,
-      "loss": 0.1195,
-      "step": 1734
-    },
-    {
-      "epoch": 0.2958147960678797,
-      "grad_norm": 0.7212527394294739,
-      "learning_rate": 7.0417732310315436e-06,
-      "loss": 0.1048,
-      "step": 1735
-    },
-    {
-      "epoch": 0.29598529450941735,
-      "grad_norm": 1.1094732284545898,
-      "learning_rate": 7.0400682011935215e-06,
-      "loss": 0.1589,
-      "step": 1736
-    },
-    {
-      "epoch": 0.29615579295095507,
-      "grad_norm": 0.724922239780426,
-      "learning_rate": 7.038363171355499e-06,
-      "loss": 0.0816,
-      "step": 1737
-    },
-    {
-      "epoch": 0.2963262913924927,
-      "grad_norm": 0.8069830536842346,
-      "learning_rate": 7.036658141517477e-06,
-      "loss": 0.0891,
-      "step": 1738
-    },
-    {
-      "epoch": 0.29649678983403044,
-      "grad_norm": 0.9070818424224854,
-      "learning_rate": 7.0349531116794545e-06,
-      "loss": 0.0875,
-      "step": 1739
-    },
-    {
-      "epoch": 0.2966672882755681,
-      "grad_norm": 0.9026073813438416,
-      "learning_rate": 7.0332480818414325e-06,
-      "loss": 0.1249,
-      "step": 1740
-    },
-    {
-      "epoch": 0.2968377867171058,
-      "grad_norm": 0.7167872190475464,
-      "learning_rate": 7.031543052003411e-06,
-      "loss": 0.1137,
-      "step": 1741
-    },
-    {
-      "epoch": 0.29700828515864347,
-      "grad_norm": 0.648733913898468,
-      "learning_rate": 7.029838022165388e-06,
-      "loss": 0.0784,
-      "step": 1742
-    },
-    {
-      "epoch": 0.29717878360018113,
-      "grad_norm": 1.9221540689468384,
-      "learning_rate": 7.028132992327366e-06,
-      "loss": 0.2363,
-      "step": 1743
-    },
-    {
-      "epoch": 0.29734928204171884,
-      "grad_norm": 1.255143404006958,
-      "learning_rate": 7.026427962489344e-06,
-      "loss": 0.1992,
-      "step": 1744
-    },
-    {
-      "epoch": 0.2975197804832565,
-      "grad_norm": 0.926605224609375,
-      "learning_rate": 7.024722932651322e-06,
-      "loss": 0.1603,
-      "step": 1745
-    },
-    {
-      "epoch": 0.2976902789247942,
-      "grad_norm": 1.1837265491485596,
-      "learning_rate": 7.023017902813299e-06,
-      "loss": 0.2193,
-      "step": 1746
-    },
-    {
-      "epoch": 0.2978607773663319,
-      "grad_norm": 0.9852341413497925,
-      "learning_rate": 7.021312872975278e-06,
-      "loss": 0.1684,
-      "step": 1747
-    },
-    {
-      "epoch": 0.2980312758078696,
-      "grad_norm": 0.7624542713165283,
-      "learning_rate": 7.019607843137255e-06,
-      "loss": 0.0737,
-      "step": 1748
-    },
-    {
-      "epoch": 0.29820177424940725,
-      "grad_norm": 0.9837128520011902,
-      "learning_rate": 7.017902813299233e-06,
-      "loss": 0.1486,
-      "step": 1749
-    },
-    {
-      "epoch": 0.29837227269094496,
-      "grad_norm": 0.7164487242698669,
-      "learning_rate": 7.016197783461212e-06,
-      "loss": 0.1192,
-      "step": 1750
-    },
-    {
-      "epoch": 0.2985427711324826,
-      "grad_norm": 1.325129747390747,
-      "learning_rate": 7.014492753623189e-06,
-      "loss": 0.1297,
-      "step": 1751
-    },
-    {
-      "epoch": 0.2987132695740203,
-      "grad_norm": 0.8222716450691223,
-      "learning_rate": 7.012787723785167e-06,
-      "loss": 0.1492,
-      "step": 1752
-    },
-    {
-      "epoch": 0.298883768015558,
-      "grad_norm": 1.1504167318344116,
-      "learning_rate": 7.011082693947144e-06,
-      "loss": 0.1816,
-      "step": 1753
-    },
-    {
-      "epoch": 0.29905426645709565,
-      "grad_norm": 0.7564553618431091,
-      "learning_rate": 7.009377664109123e-06,
-      "loss": 0.0947,
-      "step": 1754
-    },
-    {
-      "epoch": 0.29922476489863337,
-      "grad_norm": 1.0340969562530518,
-      "learning_rate": 7.0076726342711e-06,
-      "loss": 0.1252,
-      "step": 1755
-    },
-    {
-      "epoch": 0.299395263340171,
-      "grad_norm": 0.937289297580719,
-      "learning_rate": 7.005967604433078e-06,
-      "loss": 0.1858,
-      "step": 1756
-    },
-    {
-      "epoch": 0.29956576178170874,
-      "grad_norm": 1.3442150354385376,
-      "learning_rate": 7.004262574595057e-06,
-      "loss": 0.2558,
-      "step": 1757
-    },
-    {
-      "epoch": 0.2997362602232464,
-      "grad_norm": 1.5121102333068848,
-      "learning_rate": 7.002557544757034e-06,
-      "loss": 0.183,
-      "step": 1758
-    },
-    {
-      "epoch": 0.29990675866478406,
-      "grad_norm": 0.8529810905456543,
-      "learning_rate": 7.000852514919012e-06,
-      "loss": 0.1801,
-      "step": 1759
-    },
-    {
-      "epoch": 0.30007725710632177,
-      "grad_norm": 1.0886412858963013,
-      "learning_rate": 6.999147485080989e-06,
-      "loss": 0.1705,
-      "step": 1760
-    },
-    {
-      "epoch": 0.30024775554785943,
-      "grad_norm": 1.0613352060317993,
-      "learning_rate": 6.997442455242968e-06,
-      "loss": 0.1219,
-      "step": 1761
-    },
-    {
-      "epoch": 0.30041825398939714,
-      "grad_norm": 0.9400555491447449,
-      "learning_rate": 6.995737425404945e-06,
-      "loss": 0.1675,
-      "step": 1762
-    },
-    {
-      "epoch": 0.3005887524309348,
-      "grad_norm": 1.252254843711853,
-      "learning_rate": 6.9940323955669236e-06,
-      "loss": 0.1409,
-      "step": 1763
-    },
-    {
-      "epoch": 0.3007592508724725,
-      "grad_norm": 0.9437319040298462,
-      "learning_rate": 6.992327365728901e-06,
-      "loss": 0.104,
-      "step": 1764
-    },
-    {
-      "epoch": 0.3009297493140102,
-      "grad_norm": 0.9890301823616028,
-      "learning_rate": 6.990622335890879e-06,
-      "loss": 0.1972,
-      "step": 1765
-    },
-    {
-      "epoch": 0.30110024775554783,
-      "grad_norm": 0.8138511776924133,
-      "learning_rate": 6.988917306052857e-06,
-      "loss": 0.1536,
-      "step": 1766
-    },
-    {
-      "epoch": 0.30127074619708555,
-      "grad_norm": 0.9302570223808289,
-      "learning_rate": 6.9872122762148345e-06,
-      "loss": 0.0714,
-      "step": 1767
-    },
-    {
-      "epoch": 0.3014412446386232,
-      "grad_norm": 0.696306586265564,
-      "learning_rate": 6.9855072463768125e-06,
-      "loss": 0.0719,
-      "step": 1768
-    },
-    {
-      "epoch": 0.3016117430801609,
-      "grad_norm": 0.8247181177139282,
-      "learning_rate": 6.9838022165387896e-06,
-      "loss": 0.1142,
-      "step": 1769
-    },
-    {
-      "epoch": 0.3017822415216986,
-      "grad_norm": 0.5882242918014526,
-      "learning_rate": 6.982097186700768e-06,
-      "loss": 0.1016,
-      "step": 1770
-    },
-    {
-      "epoch": 0.3019527399632363,
-      "grad_norm": 0.8626717925071716,
-      "learning_rate": 6.9803921568627454e-06,
-      "loss": 0.1068,
-      "step": 1771
-    },
-    {
-      "epoch": 0.30212323840477395,
-      "grad_norm": 1.3977041244506836,
-      "learning_rate": 6.978687127024723e-06,
-      "loss": 0.1926,
-      "step": 1772
-    },
-    {
-      "epoch": 0.3022937368463116,
-      "grad_norm": 1.3635149002075195,
-      "learning_rate": 6.9769820971867005e-06,
-      "loss": 0.1599,
-      "step": 1773
-    },
-    {
-      "epoch": 0.3024642352878493,
-      "grad_norm": 0.7197402119636536,
-      "learning_rate": 6.975277067348679e-06,
-      "loss": 0.1477,
-      "step": 1774
-    },
-    {
-      "epoch": 0.302634733729387,
-      "grad_norm": 1.0707645416259766,
-      "learning_rate": 6.973572037510657e-06,
-      "loss": 0.1519,
-      "step": 1775
-    },
-    {
-      "epoch": 0.3028052321709247,
-      "grad_norm": 0.89174884557724,
-      "learning_rate": 6.971867007672634e-06,
-      "loss": 0.1014,
-      "step": 1776
-    },
-    {
-      "epoch": 0.30297573061246236,
-      "grad_norm": 0.7598327994346619,
-      "learning_rate": 6.970161977834613e-06,
-      "loss": 0.1475,
-      "step": 1777
-    },
-    {
-      "epoch": 0.30314622905400007,
-      "grad_norm": 0.6671323180198669,
-      "learning_rate": 6.96845694799659e-06,
-      "loss": 0.0418,
-      "step": 1778
-    },
-    {
-      "epoch": 0.30331672749553773,
-      "grad_norm": 0.7658970355987549,
-      "learning_rate": 6.966751918158568e-06,
-      "loss": 0.1006,
-      "step": 1779
-    },
-    {
-      "epoch": 0.30348722593707544,
-      "grad_norm": 0.9763972759246826,
-      "learning_rate": 6.965046888320546e-06,
-      "loss": 0.1675,
-      "step": 1780
-    },
-    {
-      "epoch": 0.3036577243786131,
-      "grad_norm": 2.200497627258301,
-      "learning_rate": 6.963341858482524e-06,
-      "loss": 0.2854,
-      "step": 1781
-    },
-    {
-      "epoch": 0.30382822282015076,
-      "grad_norm": 1.3654463291168213,
-      "learning_rate": 6.961636828644501e-06,
-      "loss": 0.1426,
-      "step": 1782
-    },
-    {
-      "epoch": 0.3039987212616885,
-      "grad_norm": 1.0287854671478271,
-      "learning_rate": 6.95993179880648e-06,
-      "loss": 0.193,
-      "step": 1783
-    },
-    {
-      "epoch": 0.30416921970322613,
-      "grad_norm": 0.8132631182670593,
-      "learning_rate": 6.958226768968458e-06,
-      "loss": 0.1091,
-      "step": 1784
-    },
-    {
-      "epoch": 0.30433971814476385,
-      "grad_norm": 0.9241107702255249,
-      "learning_rate": 6.956521739130435e-06,
-      "loss": 0.1505,
-      "step": 1785
-    },
-    {
-      "epoch": 0.3045102165863015,
-      "grad_norm": 1.1344634294509888,
-      "learning_rate": 6.954816709292414e-06,
-      "loss": 0.1651,
-      "step": 1786
-    },
-    {
-      "epoch": 0.3046807150278392,
-      "grad_norm": 1.2102185487747192,
-      "learning_rate": 6.953111679454391e-06,
-      "loss": 0.1135,
-      "step": 1787
-    },
-    {
-      "epoch": 0.3048512134693769,
-      "grad_norm": 0.7724695205688477,
-      "learning_rate": 6.951406649616369e-06,
-      "loss": 0.1177,
-      "step": 1788
-    },
-    {
-      "epoch": 0.30502171191091454,
-      "grad_norm": 1.012211561203003,
-      "learning_rate": 6.949701619778346e-06,
-      "loss": 0.1938,
-      "step": 1789
-    },
-    {
-      "epoch": 0.30519221035245225,
-      "grad_norm": 1.2018117904663086,
-      "learning_rate": 6.947996589940325e-06,
-      "loss": 0.1775,
-      "step": 1790
-    },
-    {
-      "epoch": 0.3053627087939899,
-      "grad_norm": 0.951012909412384,
-      "learning_rate": 6.946291560102303e-06,
-      "loss": 0.1552,
-      "step": 1791
-    },
-    {
-      "epoch": 0.3055332072355276,
-      "grad_norm": 0.7436768412590027,
-      "learning_rate": 6.94458653026428e-06,
-      "loss": 0.094,
-      "step": 1792
-    },
-    {
-      "epoch": 0.3057037056770653,
-      "grad_norm": 1.8450157642364502,
-      "learning_rate": 6.942881500426259e-06,
-      "loss": 0.2932,
-      "step": 1793
-    },
-    {
-      "epoch": 0.305874204118603,
-      "grad_norm": 0.6989912390708923,
-      "learning_rate": 6.941176470588236e-06,
-      "loss": 0.1193,
-      "step": 1794
-    },
-    {
-      "epoch": 0.30604470256014066,
-      "grad_norm": 0.9824447631835938,
-      "learning_rate": 6.939471440750214e-06,
-      "loss": 0.1727,
-      "step": 1795
-    },
-    {
-      "epoch": 0.3062152010016783,
-      "grad_norm": 1.169551968574524,
-      "learning_rate": 6.937766410912191e-06,
-      "loss": 0.0808,
-      "step": 1796
-    },
-    {
-      "epoch": 0.30638569944321603,
-      "grad_norm": 0.6870118379592896,
-      "learning_rate": 6.9360613810741695e-06,
-      "loss": 0.1052,
-      "step": 1797
-    },
-    {
-      "epoch": 0.3065561978847537,
-      "grad_norm": 1.383569598197937,
-      "learning_rate": 6.934356351236147e-06,
-      "loss": 0.2392,
-      "step": 1798
-    },
-    {
-      "epoch": 0.3067266963262914,
-      "grad_norm": 0.6766565442085266,
-      "learning_rate": 6.9326513213981254e-06,
-      "loss": 0.097,
-      "step": 1799
-    },
-    {
-      "epoch": 0.30689719476782906,
-      "grad_norm": 0.6242276430130005,
-      "learning_rate": 6.930946291560103e-06,
-      "loss": 0.1012,
-      "step": 1800
-    },
-    {
-      "epoch": 0.30689719476782906,
-      "eval_f1_score": 0.2803234501347709,
-      "eval_loss": 0.15971286594867706,
-      "eval_runtime": 182.6862,
-      "eval_samples_per_second": 54.739,
-      "eval_steps_per_second": 3.421,
-      "step": 1800
-    },
-    {
-      "epoch": 0.3070676932093668,
-      "grad_norm": 1.0924694538116455,
-      "learning_rate": 6.9292412617220805e-06,
-      "loss": 0.1936,
-      "step": 1801
-    },
-    {
-      "epoch": 0.30723819165090444,
-      "grad_norm": 0.8069597482681274,
-      "learning_rate": 6.927536231884059e-06,
-      "loss": 0.1299,
-      "step": 1802
-    },
-    {
-      "epoch": 0.30740869009244215,
-      "grad_norm": 1.007934331893921,
-      "learning_rate": 6.925831202046036e-06,
-      "loss": 0.1719,
-      "step": 1803
-    },
-    {
-      "epoch": 0.3075791885339798,
-      "grad_norm": 1.0395991802215576,
-      "learning_rate": 6.924126172208014e-06,
-      "loss": 0.1675,
-      "step": 1804
-    },
-    {
-      "epoch": 0.30774968697551747,
-      "grad_norm": 0.7160507440567017,
-      "learning_rate": 6.9224211423699914e-06,
-      "loss": 0.1178,
-      "step": 1805
-    },
-    {
-      "epoch": 0.3079201854170552,
-      "grad_norm": 1.3453173637390137,
-      "learning_rate": 6.92071611253197e-06,
-      "loss": 0.197,
-      "step": 1806
-    },
-    {
-      "epoch": 0.30809068385859284,
-      "grad_norm": 0.869848906993866,
-      "learning_rate": 6.919011082693947e-06,
-      "loss": 0.1259,
-      "step": 1807
-    },
-    {
-      "epoch": 0.30826118230013055,
-      "grad_norm": 0.8316013813018799,
-      "learning_rate": 6.917306052855925e-06,
-      "loss": 0.1555,
-      "step": 1808
-    },
-    {
-      "epoch": 0.3084316807416682,
-      "grad_norm": 0.6854284405708313,
-      "learning_rate": 6.915601023017904e-06,
-      "loss": 0.1546,
-      "step": 1809
-    },
-    {
-      "epoch": 0.3086021791832059,
-      "grad_norm": 0.9202719330787659,
-      "learning_rate": 6.913895993179881e-06,
-      "loss": 0.1263,
-      "step": 1810
-    },
-    {
-      "epoch": 0.3087726776247436,
-      "grad_norm": 0.8600966334342957,
-      "learning_rate": 6.912190963341859e-06,
-      "loss": 0.2094,
-      "step": 1811
-    },
-    {
-      "epoch": 0.30894317606628124,
-      "grad_norm": 1.3879214525222778,
-      "learning_rate": 6.910485933503836e-06,
-      "loss": 0.1503,
-      "step": 1812
-    },
-    {
-      "epoch": 0.30911367450781896,
-      "grad_norm": 0.8720242977142334,
-      "learning_rate": 6.908780903665815e-06,
-      "loss": 0.1256,
-      "step": 1813
-    },
-    {
-      "epoch": 0.3092841729493566,
-      "grad_norm": 1.0565900802612305,
-      "learning_rate": 6.907075873827792e-06,
-      "loss": 0.1797,
-      "step": 1814
-    },
-    {
-      "epoch": 0.30945467139089433,
-      "grad_norm": 1.1120818853378296,
-      "learning_rate": 6.90537084398977e-06,
-      "loss": 0.1902,
-      "step": 1815
-    },
-    {
-      "epoch": 0.309625169832432,
-      "grad_norm": 0.8991826772689819,
-      "learning_rate": 6.903665814151749e-06,
-      "loss": 0.1553,
-      "step": 1816
-    },
-    {
-      "epoch": 0.3097956682739697,
-      "grad_norm": 1.149214744567871,
-      "learning_rate": 6.901960784313726e-06,
-      "loss": 0.2201,
-      "step": 1817
-    },
-    {
-      "epoch": 0.30996616671550736,
-      "grad_norm": 0.6388124823570251,
-      "learning_rate": 6.900255754475704e-06,
-      "loss": 0.1307,
-      "step": 1818
-    },
-    {
-      "epoch": 0.310136665157045,
-      "grad_norm": 0.8748459219932556,
-      "learning_rate": 6.898550724637682e-06,
-      "loss": 0.1085,
-      "step": 1819
-    },
-    {
-      "epoch": 0.31030716359858274,
-      "grad_norm": 1.3581770658493042,
-      "learning_rate": 6.89684569479966e-06,
-      "loss": 0.2259,
-      "step": 1820
-    },
-    {
-      "epoch": 0.3104776620401204,
-      "grad_norm": 0.7230364084243774,
-      "learning_rate": 6.895140664961637e-06,
-      "loss": 0.0964,
-      "step": 1821
-    },
-    {
-      "epoch": 0.3106481604816581,
-      "grad_norm": 1.595178246498108,
-      "learning_rate": 6.893435635123616e-06,
-      "loss": 0.242,
-      "step": 1822
-    },
-    {
-      "epoch": 0.31081865892319577,
-      "grad_norm": 0.9053245782852173,
-      "learning_rate": 6.891730605285593e-06,
-      "loss": 0.2075,
-      "step": 1823
-    },
-    {
-      "epoch": 0.3109891573647335,
-      "grad_norm": 0.7452926635742188,
-      "learning_rate": 6.890025575447571e-06,
-      "loss": 0.1048,
-      "step": 1824
-    },
-    {
-      "epoch": 0.31115965580627114,
-      "grad_norm": 0.8311702013015747,
-      "learning_rate": 6.8883205456095495e-06,
-      "loss": 0.1312,
-      "step": 1825
-    },
-    {
-      "epoch": 0.31133015424780885,
-      "grad_norm": 0.9205570220947266,
-      "learning_rate": 6.886615515771527e-06,
-      "loss": 0.1585,
-      "step": 1826
-    },
-    {
-      "epoch": 0.3115006526893465,
-      "grad_norm": 0.9653121829032898,
-      "learning_rate": 6.884910485933505e-06,
-      "loss": 0.199,
-      "step": 1827
-    },
-    {
-      "epoch": 0.31167115113088417,
-      "grad_norm": 0.8807269930839539,
-      "learning_rate": 6.883205456095482e-06,
-      "loss": 0.1618,
-      "step": 1828
-    },
-    {
-      "epoch": 0.3118416495724219,
-      "grad_norm": 0.7480918169021606,
-      "learning_rate": 6.8815004262574605e-06,
-      "loss": 0.1095,
-      "step": 1829
-    },
-    {
-      "epoch": 0.31201214801395954,
-      "grad_norm": 1.1662970781326294,
-      "learning_rate": 6.879795396419438e-06,
-      "loss": 0.2175,
-      "step": 1830
-    },
-    {
-      "epoch": 0.31218264645549726,
-      "grad_norm": 0.8109918832778931,
-      "learning_rate": 6.8780903665814155e-06,
-      "loss": 0.1532,
-      "step": 1831
-    },
-    {
-      "epoch": 0.3123531448970349,
-      "grad_norm": 0.766344428062439,
-      "learning_rate": 6.876385336743393e-06,
-      "loss": 0.1499,
-      "step": 1832
-    },
-    {
-      "epoch": 0.31252364333857263,
-      "grad_norm": 1.1723487377166748,
-      "learning_rate": 6.8746803069053714e-06,
-      "loss": 0.1621,
-      "step": 1833
-    },
-    {
-      "epoch": 0.3126941417801103,
-      "grad_norm": 0.7435732483863831,
-      "learning_rate": 6.872975277067349e-06,
-      "loss": 0.0834,
-      "step": 1834
-    },
-    {
-      "epoch": 0.31286464022164795,
-      "grad_norm": 0.9418601393699646,
-      "learning_rate": 6.8712702472293265e-06,
-      "loss": 0.1322,
-      "step": 1835
-    },
-    {
-      "epoch": 0.31303513866318566,
-      "grad_norm": 0.6800971031188965,
-      "learning_rate": 6.869565217391305e-06,
-      "loss": 0.1264,
-      "step": 1836
-    },
-    {
-      "epoch": 0.3132056371047233,
-      "grad_norm": 0.7308107018470764,
-      "learning_rate": 6.867860187553282e-06,
-      "loss": 0.1478,
-      "step": 1837
-    },
-    {
-      "epoch": 0.31337613554626104,
-      "grad_norm": 0.9065027236938477,
-      "learning_rate": 6.866155157715261e-06,
-      "loss": 0.1384,
-      "step": 1838
-    },
-    {
-      "epoch": 0.3135466339877987,
-      "grad_norm": 1.243708610534668,
-      "learning_rate": 6.864450127877238e-06,
-      "loss": 0.1833,
-      "step": 1839
-    },
-    {
-      "epoch": 0.3137171324293364,
-      "grad_norm": 1.047222375869751,
-      "learning_rate": 6.862745098039216e-06,
-      "loss": 0.1548,
-      "step": 1840
-    },
-    {
-      "epoch": 0.31388763087087407,
-      "grad_norm": 0.6964305639266968,
-      "learning_rate": 6.861040068201195e-06,
-      "loss": 0.1444,
-      "step": 1841
-    },
-    {
-      "epoch": 0.3140581293124117,
-      "grad_norm": 1.2023725509643555,
-      "learning_rate": 6.859335038363172e-06,
-      "loss": 0.1943,
-      "step": 1842
-    },
-    {
-      "epoch": 0.31422862775394944,
-      "grad_norm": 1.2627588510513306,
-      "learning_rate": 6.85763000852515e-06,
-      "loss": 0.1828,
-      "step": 1843
-    },
-    {
-      "epoch": 0.3143991261954871,
-      "grad_norm": 0.8266808986663818,
-      "learning_rate": 6.855924978687127e-06,
-      "loss": 0.1081,
-      "step": 1844
-    },
-    {
-      "epoch": 0.3145696246370248,
-      "grad_norm": 0.79570472240448,
-      "learning_rate": 6.854219948849106e-06,
-      "loss": 0.1366,
-      "step": 1845
-    },
-    {
-      "epoch": 0.3147401230785625,
-      "grad_norm": 1.0187424421310425,
-      "learning_rate": 6.852514919011083e-06,
-      "loss": 0.1525,
-      "step": 1846
-    },
-    {
-      "epoch": 0.3149106215201002,
-      "grad_norm": 0.7972794771194458,
-      "learning_rate": 6.850809889173061e-06,
-      "loss": 0.1098,
-      "step": 1847
-    },
-    {
-      "epoch": 0.31508111996163785,
-      "grad_norm": 1.424127459526062,
-      "learning_rate": 6.849104859335038e-06,
-      "loss": 0.1915,
-      "step": 1848
-    },
-    {
-      "epoch": 0.31525161840317556,
-      "grad_norm": 1.0600439310073853,
-      "learning_rate": 6.847399829497017e-06,
-      "loss": 0.1273,
-      "step": 1849
-    },
-    {
-      "epoch": 0.3154221168447132,
-      "grad_norm": 0.8016909956932068,
-      "learning_rate": 6.845694799658995e-06,
-      "loss": 0.0958,
-      "step": 1850
-    },
-    {
-      "epoch": 0.3155926152862509,
-      "grad_norm": 0.7285593748092651,
-      "learning_rate": 6.843989769820972e-06,
-      "loss": 0.1351,
-      "step": 1851
-    },
-    {
-      "epoch": 0.3157631137277886,
-      "grad_norm": 1.0121415853500366,
-      "learning_rate": 6.842284739982951e-06,
-      "loss": 0.0566,
-      "step": 1852
-    },
-    {
-      "epoch": 0.31593361216932625,
-      "grad_norm": 0.782200038433075,
-      "learning_rate": 6.840579710144928e-06,
-      "loss": 0.1084,
-      "step": 1853
-    },
-    {
-      "epoch": 0.31610411061086396,
-      "grad_norm": 0.8730593323707581,
-      "learning_rate": 6.838874680306906e-06,
-      "loss": 0.1147,
-      "step": 1854
-    },
-    {
-      "epoch": 0.3162746090524016,
-      "grad_norm": 1.0359622240066528,
-      "learning_rate": 6.837169650468884e-06,
-      "loss": 0.1731,
-      "step": 1855
-    },
-    {
-      "epoch": 0.31644510749393934,
-      "grad_norm": 1.381600260734558,
-      "learning_rate": 6.835464620630862e-06,
-      "loss": 0.154,
-      "step": 1856
-    },
-    {
-      "epoch": 0.316615605935477,
-      "grad_norm": 1.3901880979537964,
-      "learning_rate": 6.833759590792839e-06,
-      "loss": 0.1854,
-      "step": 1857
-    },
-    {
-      "epoch": 0.31678610437701465,
-      "grad_norm": 2.308821678161621,
-      "learning_rate": 6.832054560954818e-06,
-      "loss": 0.1926,
-      "step": 1858
-    },
-    {
-      "epoch": 0.31695660281855237,
-      "grad_norm": 2.024123191833496,
-      "learning_rate": 6.8303495311167955e-06,
-      "loss": 0.244,
-      "step": 1859
-    },
-    {
-      "epoch": 0.31712710126009,
-      "grad_norm": 0.6854466795921326,
-      "learning_rate": 6.828644501278773e-06,
-      "loss": 0.0982,
-      "step": 1860
-    },
-    {
-      "epoch": 0.31729759970162774,
-      "grad_norm": 0.8933399319648743,
-      "learning_rate": 6.826939471440751e-06,
-      "loss": 0.0826,
-      "step": 1861
-    },
-    {
-      "epoch": 0.3174680981431654,
-      "grad_norm": 0.8886541128158569,
-      "learning_rate": 6.8252344416027285e-06,
-      "loss": 0.1203,
-      "step": 1862
-    },
-    {
-      "epoch": 0.3176385965847031,
-      "grad_norm": 0.7370439171791077,
-      "learning_rate": 6.8235294117647065e-06,
-      "loss": 0.0999,
-      "step": 1863
-    },
-    {
-      "epoch": 0.3178090950262408,
-      "grad_norm": 1.627724289894104,
-      "learning_rate": 6.8218243819266836e-06,
-      "loss": 0.239,
-      "step": 1864
-    },
-    {
-      "epoch": 0.31797959346777843,
-      "grad_norm": 0.888409435749054,
-      "learning_rate": 6.820119352088662e-06,
-      "loss": 0.1712,
-      "step": 1865
-    },
-    {
-      "epoch": 0.31815009190931615,
-      "grad_norm": 0.9039790034294128,
-      "learning_rate": 6.81841432225064e-06,
-      "loss": 0.1022,
-      "step": 1866
-    },
-    {
-      "epoch": 0.3183205903508538,
-      "grad_norm": 0.6640911102294922,
-      "learning_rate": 6.816709292412617e-06,
-      "loss": 0.1023,
-      "step": 1867
-    },
-    {
-      "epoch": 0.3184910887923915,
-      "grad_norm": 1.0628663301467896,
-      "learning_rate": 6.815004262574596e-06,
-      "loss": 0.161,
-      "step": 1868
-    },
-    {
-      "epoch": 0.3186615872339292,
-      "grad_norm": 0.6437575221061707,
-      "learning_rate": 6.813299232736573e-06,
-      "loss": 0.0882,
-      "step": 1869
-    },
-    {
-      "epoch": 0.3188320856754669,
-      "grad_norm": 0.8309326171875,
-      "learning_rate": 6.811594202898551e-06,
-      "loss": 0.1695,
-      "step": 1870
-    },
-    {
-      "epoch": 0.31900258411700455,
-      "grad_norm": 0.7946399450302124,
-      "learning_rate": 6.809889173060528e-06,
-      "loss": 0.1474,
-      "step": 1871
-    },
-    {
-      "epoch": 0.31917308255854226,
-      "grad_norm": 0.7174097895622253,
-      "learning_rate": 6.808184143222507e-06,
-      "loss": 0.0963,
-      "step": 1872
-    },
-    {
-      "epoch": 0.3193435810000799,
-      "grad_norm": 0.7954621315002441,
-      "learning_rate": 6.806479113384484e-06,
-      "loss": 0.1349,
-      "step": 1873
-    },
-    {
-      "epoch": 0.3195140794416176,
-      "grad_norm": 0.6577284336090088,
-      "learning_rate": 6.804774083546463e-06,
-      "loss": 0.1079,
-      "step": 1874
-    },
-    {
-      "epoch": 0.3196845778831553,
-      "grad_norm": 1.0862046480178833,
-      "learning_rate": 6.803069053708441e-06,
-      "loss": 0.2176,
-      "step": 1875
-    },
-    {
-      "epoch": 0.31985507632469296,
-      "grad_norm": 0.7450264096260071,
-      "learning_rate": 6.801364023870418e-06,
-      "loss": 0.0668,
-      "step": 1876
-    },
-    {
-      "epoch": 0.32002557476623067,
-      "grad_norm": 0.9908052682876587,
-      "learning_rate": 6.799658994032397e-06,
-      "loss": 0.1696,
-      "step": 1877
-    },
-    {
-      "epoch": 0.32019607320776833,
-      "grad_norm": 1.015461802482605,
-      "learning_rate": 6.797953964194374e-06,
-      "loss": 0.1509,
-      "step": 1878
-    },
-    {
-      "epoch": 0.32036657164930604,
-      "grad_norm": 0.8305402398109436,
-      "learning_rate": 6.796248934356352e-06,
-      "loss": 0.0678,
-      "step": 1879
-    },
-    {
-      "epoch": 0.3205370700908437,
-      "grad_norm": 1.0898061990737915,
-      "learning_rate": 6.794543904518329e-06,
-      "loss": 0.1734,
-      "step": 1880
-    },
-    {
-      "epoch": 0.32070756853238136,
-      "grad_norm": 0.6495380997657776,
-      "learning_rate": 6.792838874680308e-06,
-      "loss": 0.1133,
-      "step": 1881
-    },
-    {
-      "epoch": 0.3208780669739191,
-      "grad_norm": 0.9107251167297363,
-      "learning_rate": 6.791133844842285e-06,
-      "loss": 0.1741,
-      "step": 1882
-    },
-    {
-      "epoch": 0.32104856541545673,
-      "grad_norm": 0.8304046392440796,
-      "learning_rate": 6.789428815004263e-06,
-      "loss": 0.0817,
-      "step": 1883
-    },
-    {
-      "epoch": 0.32121906385699445,
-      "grad_norm": 0.6666496396064758,
-      "learning_rate": 6.787723785166242e-06,
-      "loss": 0.0445,
-      "step": 1884
-    },
-    {
-      "epoch": 0.3213895622985321,
-      "grad_norm": 1.2774522304534912,
-      "learning_rate": 6.786018755328219e-06,
-      "loss": 0.2335,
-      "step": 1885
-    },
-    {
-      "epoch": 0.3215600607400698,
-      "grad_norm": 1.08737051486969,
-      "learning_rate": 6.784313725490197e-06,
-      "loss": 0.1862,
-      "step": 1886
-    },
-    {
-      "epoch": 0.3217305591816075,
-      "grad_norm": 1.0256638526916504,
-      "learning_rate": 6.782608695652174e-06,
-      "loss": 0.1821,
-      "step": 1887
-    },
-    {
-      "epoch": 0.32190105762314514,
-      "grad_norm": 1.390364646911621,
-      "learning_rate": 6.780903665814153e-06,
-      "loss": 0.186,
-      "step": 1888
-    },
-    {
-      "epoch": 0.32207155606468285,
-      "grad_norm": 0.5867147445678711,
-      "learning_rate": 6.77919863597613e-06,
-      "loss": 0.0891,
-      "step": 1889
-    },
-    {
-      "epoch": 0.3222420545062205,
-      "grad_norm": 0.7561818957328796,
-      "learning_rate": 6.777493606138108e-06,
-      "loss": 0.1058,
-      "step": 1890
-    },
-    {
-      "epoch": 0.3224125529477582,
-      "grad_norm": 0.6854589581489563,
-      "learning_rate": 6.7757885763000865e-06,
-      "loss": 0.1171,
-      "step": 1891
-    },
-    {
-      "epoch": 0.3225830513892959,
-      "grad_norm": 1.1057910919189453,
-      "learning_rate": 6.7740835464620636e-06,
-      "loss": 0.1567,
-      "step": 1892
-    },
-    {
-      "epoch": 0.3227535498308336,
-      "grad_norm": 1.0578477382659912,
-      "learning_rate": 6.7723785166240415e-06,
-      "loss": 0.1911,
-      "step": 1893
-    },
-    {
-      "epoch": 0.32292404827237126,
-      "grad_norm": 1.0057955980300903,
-      "learning_rate": 6.7706734867860195e-06,
-      "loss": 0.0877,
-      "step": 1894
-    },
-    {
-      "epoch": 0.32309454671390897,
-      "grad_norm": 1.0721625089645386,
-      "learning_rate": 6.768968456947997e-06,
-      "loss": 0.2047,
-      "step": 1895
-    },
-    {
-      "epoch": 0.32326504515544663,
-      "grad_norm": 0.8911730647087097,
-      "learning_rate": 6.7672634271099745e-06,
-      "loss": 0.1215,
-      "step": 1896
-    },
-    {
-      "epoch": 0.3234355435969843,
-      "grad_norm": 0.9344785809516907,
-      "learning_rate": 6.765558397271953e-06,
-      "loss": 0.1038,
-      "step": 1897
-    },
-    {
-      "epoch": 0.323606042038522,
-      "grad_norm": 0.890182375907898,
-      "learning_rate": 6.76385336743393e-06,
-      "loss": 0.1262,
-      "step": 1898
-    },
-    {
-      "epoch": 0.32377654048005966,
-      "grad_norm": 1.6066832542419434,
-      "learning_rate": 6.762148337595908e-06,
-      "loss": 0.2593,
-      "step": 1899
-    },
-    {
-      "epoch": 0.3239470389215974,
-      "grad_norm": 0.9894545078277588,
-      "learning_rate": 6.760443307757887e-06,
-      "loss": 0.1459,
-      "step": 1900
-    },
-    {
-      "epoch": 0.3239470389215974,
-      "eval_f1_score": 0.3062200956937799,
-      "eval_loss": 0.155869260430336,
-      "eval_runtime": 182.6505,
-      "eval_samples_per_second": 54.749,
-      "eval_steps_per_second": 3.422,
-      "step": 1900
-    },
-    {
-      "epoch": 0.32411753736313503,
-      "grad_norm": 0.7483901381492615,
-      "learning_rate": 6.758738277919864e-06,
-      "loss": 0.1084,
-      "step": 1901
-    },
-    {
-      "epoch": 0.32428803580467275,
-      "grad_norm": 0.989708423614502,
-      "learning_rate": 6.757033248081842e-06,
-      "loss": 0.1565,
-      "step": 1902
-    },
-    {
-      "epoch": 0.3244585342462104,
-      "grad_norm": 0.9160560369491577,
-      "learning_rate": 6.755328218243819e-06,
-      "loss": 0.0911,
-      "step": 1903
-    },
-    {
-      "epoch": 0.32462903268774806,
-      "grad_norm": 1.524827241897583,
-      "learning_rate": 6.753623188405798e-06,
-      "loss": 0.137,
-      "step": 1904
-    },
-    {
-      "epoch": 0.3247995311292858,
-      "grad_norm": 0.9177931547164917,
-      "learning_rate": 6.751918158567775e-06,
-      "loss": 0.1413,
-      "step": 1905
-    },
-    {
-      "epoch": 0.32497002957082344,
-      "grad_norm": 0.9163035750389099,
-      "learning_rate": 6.750213128729753e-06,
-      "loss": 0.1107,
-      "step": 1906
-    },
-    {
-      "epoch": 0.32514052801236115,
-      "grad_norm": 1.2288998365402222,
-      "learning_rate": 6.74850809889173e-06,
-      "loss": 0.1198,
-      "step": 1907
-    },
-    {
-      "epoch": 0.3253110264538988,
-      "grad_norm": 1.2662756443023682,
-      "learning_rate": 6.746803069053709e-06,
-      "loss": 0.1756,
-      "step": 1908
-    },
-    {
-      "epoch": 0.3254815248954365,
-      "grad_norm": 1.25589919090271,
-      "learning_rate": 6.745098039215687e-06,
-      "loss": 0.1848,
-      "step": 1909
-    },
-    {
-      "epoch": 0.3256520233369742,
-      "grad_norm": 0.8272511959075928,
-      "learning_rate": 6.743393009377664e-06,
-      "loss": 0.1146,
-      "step": 1910
-    },
-    {
-      "epoch": 0.32582252177851184,
-      "grad_norm": 1.0488500595092773,
-      "learning_rate": 6.741687979539643e-06,
-      "loss": 0.1479,
-      "step": 1911
-    },
-    {
-      "epoch": 0.32599302022004956,
-      "grad_norm": 1.2019003629684448,
-      "learning_rate": 6.73998294970162e-06,
-      "loss": 0.1923,
-      "step": 1912
-    },
-    {
-      "epoch": 0.3261635186615872,
-      "grad_norm": 0.8280180096626282,
-      "learning_rate": 6.738277919863599e-06,
-      "loss": 0.1428,
-      "step": 1913
-    },
-    {
-      "epoch": 0.32633401710312493,
-      "grad_norm": 0.8211914300918579,
-      "learning_rate": 6.736572890025576e-06,
-      "loss": 0.1007,
-      "step": 1914
-    },
-    {
-      "epoch": 0.3265045155446626,
-      "grad_norm": 1.6616082191467285,
-      "learning_rate": 6.734867860187554e-06,
-      "loss": 0.1183,
-      "step": 1915
-    },
-    {
-      "epoch": 0.3266750139862003,
-      "grad_norm": 0.7762541174888611,
-      "learning_rate": 6.733162830349533e-06,
-      "loss": 0.1671,
-      "step": 1916
-    },
-    {
-      "epoch": 0.32684551242773796,
-      "grad_norm": 1.5211981534957886,
-      "learning_rate": 6.73145780051151e-06,
-      "loss": 0.1602,
-      "step": 1917
-    },
-    {
-      "epoch": 0.3270160108692757,
-      "grad_norm": 1.5074344873428345,
-      "learning_rate": 6.729752770673488e-06,
-      "loss": 0.1641,
-      "step": 1918
-    },
-    {
-      "epoch": 0.32718650931081333,
-      "grad_norm": 0.8448349833488464,
-      "learning_rate": 6.728047740835465e-06,
-      "loss": 0.1223,
-      "step": 1919
-    },
-    {
-      "epoch": 0.327357007752351,
-      "grad_norm": 1.0949711799621582,
-      "learning_rate": 6.7263427109974436e-06,
-      "loss": 0.2247,
-      "step": 1920
-    },
-    {
-      "epoch": 0.3275275061938887,
-      "grad_norm": 1.326486587524414,
-      "learning_rate": 6.724637681159421e-06,
-      "loss": 0.2258,
-      "step": 1921
-    },
-    {
-      "epoch": 0.32769800463542637,
-      "grad_norm": 1.2110795974731445,
-      "learning_rate": 6.722932651321399e-06,
-      "loss": 0.1842,
-      "step": 1922
-    },
-    {
-      "epoch": 0.3278685030769641,
-      "grad_norm": 1.0380573272705078,
-      "learning_rate": 6.721227621483376e-06,
-      "loss": 0.1794,
-      "step": 1923
-    },
-    {
-      "epoch": 0.32803900151850174,
-      "grad_norm": 1.1367063522338867,
-      "learning_rate": 6.7195225916453545e-06,
-      "loss": 0.2063,
-      "step": 1924
-    },
-    {
-      "epoch": 0.32820949996003945,
-      "grad_norm": 0.9798280596733093,
-      "learning_rate": 6.7178175618073325e-06,
-      "loss": 0.1611,
-      "step": 1925
-    },
-    {
-      "epoch": 0.3283799984015771,
-      "grad_norm": 0.8064085841178894,
-      "learning_rate": 6.7161125319693096e-06,
-      "loss": 0.1539,
-      "step": 1926
-    },
-    {
-      "epoch": 0.32855049684311477,
-      "grad_norm": 0.9155250191688538,
-      "learning_rate": 6.714407502131288e-06,
-      "loss": 0.1546,
-      "step": 1927
-    },
-    {
-      "epoch": 0.3287209952846525,
-      "grad_norm": 0.9837044477462769,
-      "learning_rate": 6.7127024722932654e-06,
-      "loss": 0.1067,
-      "step": 1928
-    },
-    {
-      "epoch": 0.32889149372619014,
-      "grad_norm": 1.281535029411316,
-      "learning_rate": 6.710997442455243e-06,
-      "loss": 0.09,
-      "step": 1929
-    },
-    {
-      "epoch": 0.32906199216772786,
-      "grad_norm": 0.655259370803833,
-      "learning_rate": 6.709292412617221e-06,
-      "loss": 0.0973,
-      "step": 1930
-    },
-    {
-      "epoch": 0.3292324906092655,
-      "grad_norm": 1.0607177019119263,
-      "learning_rate": 6.707587382779199e-06,
-      "loss": 0.1247,
-      "step": 1931
-    },
-    {
-      "epoch": 0.32940298905080323,
-      "grad_norm": 0.7018961310386658,
-      "learning_rate": 6.705882352941176e-06,
-      "loss": 0.109,
-      "step": 1932
-    },
-    {
-      "epoch": 0.3295734874923409,
-      "grad_norm": 0.699153482913971,
-      "learning_rate": 6.704177323103155e-06,
-      "loss": 0.1227,
-      "step": 1933
-    },
-    {
-      "epoch": 0.32974398593387855,
-      "grad_norm": 1.889542818069458,
-      "learning_rate": 6.702472293265133e-06,
-      "loss": 0.2145,
-      "step": 1934
-    },
-    {
-      "epoch": 0.32991448437541626,
-      "grad_norm": 0.8169059753417969,
-      "learning_rate": 6.70076726342711e-06,
-      "loss": 0.1256,
-      "step": 1935
-    },
-    {
-      "epoch": 0.3300849828169539,
-      "grad_norm": 0.8507729172706604,
-      "learning_rate": 6.699062233589089e-06,
-      "loss": 0.1536,
-      "step": 1936
-    },
-    {
-      "epoch": 0.33025548125849163,
-      "grad_norm": 0.7230017781257629,
-      "learning_rate": 6.697357203751066e-06,
-      "loss": 0.0726,
-      "step": 1937
-    },
-    {
-      "epoch": 0.3304259797000293,
-      "grad_norm": 1.5054045915603638,
-      "learning_rate": 6.695652173913044e-06,
-      "loss": 0.179,
-      "step": 1938
-    },
-    {
-      "epoch": 0.330596478141567,
-      "grad_norm": 0.7017834186553955,
-      "learning_rate": 6.693947144075021e-06,
-      "loss": 0.0953,
-      "step": 1939
-    },
-    {
-      "epoch": 0.33076697658310467,
-      "grad_norm": 0.7926969528198242,
-      "learning_rate": 6.692242114237e-06,
-      "loss": 0.1255,
-      "step": 1940
-    },
-    {
-      "epoch": 0.3309374750246424,
-      "grad_norm": 1.4403599500656128,
-      "learning_rate": 6.690537084398977e-06,
-      "loss": 0.1628,
-      "step": 1941
-    },
-    {
-      "epoch": 0.33110797346618004,
-      "grad_norm": 0.8787623047828674,
-      "learning_rate": 6.688832054560955e-06,
-      "loss": 0.1328,
-      "step": 1942
-    },
-    {
-      "epoch": 0.3312784719077177,
-      "grad_norm": 1.183626651763916,
-      "learning_rate": 6.687127024722934e-06,
-      "loss": 0.1683,
-      "step": 1943
-    },
-    {
-      "epoch": 0.3314489703492554,
-      "grad_norm": 0.9939226508140564,
-      "learning_rate": 6.685421994884911e-06,
-      "loss": 0.1818,
-      "step": 1944
-    },
-    {
-      "epoch": 0.33161946879079307,
-      "grad_norm": 1.0192537307739258,
-      "learning_rate": 6.683716965046889e-06,
-      "loss": 0.1743,
-      "step": 1945
-    },
-    {
-      "epoch": 0.3317899672323308,
-      "grad_norm": 1.2089581489562988,
-      "learning_rate": 6.682011935208866e-06,
-      "loss": 0.237,
-      "step": 1946
-    },
-    {
-      "epoch": 0.33196046567386844,
-      "grad_norm": 0.7481584548950195,
-      "learning_rate": 6.680306905370845e-06,
-      "loss": 0.1147,
-      "step": 1947
-    },
-    {
-      "epoch": 0.33213096411540616,
-      "grad_norm": 1.2319127321243286,
-      "learning_rate": 6.678601875532822e-06,
-      "loss": 0.207,
-      "step": 1948
-    },
-    {
-      "epoch": 0.3323014625569438,
-      "grad_norm": 1.2917490005493164,
-      "learning_rate": 6.676896845694801e-06,
-      "loss": 0.1417,
-      "step": 1949
-    },
-    {
-      "epoch": 0.3324719609984815,
-      "grad_norm": 0.9638323783874512,
-      "learning_rate": 6.675191815856779e-06,
-      "loss": 0.1012,
-      "step": 1950
-    },
-    {
-      "epoch": 0.3326424594400192,
-      "grad_norm": 0.9457072615623474,
-      "learning_rate": 6.673486786018756e-06,
-      "loss": 0.1296,
-      "step": 1951
-    },
-    {
-      "epoch": 0.33281295788155685,
-      "grad_norm": 0.8784216046333313,
-      "learning_rate": 6.6717817561807345e-06,
-      "loss": 0.1163,
-      "step": 1952
-    },
-    {
-      "epoch": 0.33298345632309456,
-      "grad_norm": 0.9807690978050232,
-      "learning_rate": 6.670076726342712e-06,
-      "loss": 0.1292,
-      "step": 1953
-    },
-    {
-      "epoch": 0.3331539547646322,
-      "grad_norm": 0.8995668888092041,
-      "learning_rate": 6.6683716965046895e-06,
-      "loss": 0.0743,
-      "step": 1954
-    },
-    {
-      "epoch": 0.33332445320616994,
-      "grad_norm": 0.9565035104751587,
-      "learning_rate": 6.666666666666667e-06,
-      "loss": 0.163,
-      "step": 1955
-    },
-    {
-      "epoch": 0.3334949516477076,
-      "grad_norm": 0.5363723039627075,
-      "learning_rate": 6.6649616368286454e-06,
-      "loss": 0.0634,
-      "step": 1956
-    },
-    {
-      "epoch": 0.33366545008924525,
-      "grad_norm": 0.5674836039543152,
-      "learning_rate": 6.6632566069906225e-06,
-      "loss": 0.0746,
-      "step": 1957
-    },
-    {
-      "epoch": 0.33383594853078297,
-      "grad_norm": 0.8529496192932129,
-      "learning_rate": 6.6615515771526005e-06,
-      "loss": 0.148,
-      "step": 1958
-    },
-    {
-      "epoch": 0.3340064469723206,
-      "grad_norm": 0.6780075430870056,
-      "learning_rate": 6.659846547314579e-06,
-      "loss": 0.083,
-      "step": 1959
-    },
-    {
-      "epoch": 0.33417694541385834,
-      "grad_norm": 1.7016448974609375,
-      "learning_rate": 6.658141517476556e-06,
-      "loss": 0.2304,
-      "step": 1960
-    },
-    {
-      "epoch": 0.334347443855396,
-      "grad_norm": 0.8263558745384216,
-      "learning_rate": 6.656436487638534e-06,
-      "loss": 0.0988,
-      "step": 1961
-    },
-    {
-      "epoch": 0.3345179422969337,
-      "grad_norm": 0.9661087989807129,
-      "learning_rate": 6.6547314578005114e-06,
-      "loss": 0.1631,
-      "step": 1962
-    },
-    {
-      "epoch": 0.33468844073847137,
-      "grad_norm": 0.7975350022315979,
-      "learning_rate": 6.65302642796249e-06,
-      "loss": 0.1455,
-      "step": 1963
-    },
-    {
-      "epoch": 0.3348589391800091,
-      "grad_norm": 1.0657655000686646,
-      "learning_rate": 6.651321398124467e-06,
-      "loss": 0.1644,
-      "step": 1964
-    },
-    {
-      "epoch": 0.33502943762154674,
-      "grad_norm": 0.6555298566818237,
-      "learning_rate": 6.649616368286445e-06,
-      "loss": 0.0623,
-      "step": 1965
-    },
-    {
-      "epoch": 0.3351999360630844,
-      "grad_norm": 0.9956116080284119,
-      "learning_rate": 6.647911338448423e-06,
-      "loss": 0.1462,
-      "step": 1966
-    },
-    {
-      "epoch": 0.3353704345046221,
-      "grad_norm": 1.17995023727417,
-      "learning_rate": 6.646206308610401e-06,
-      "loss": 0.1679,
-      "step": 1967
-    },
-    {
-      "epoch": 0.3355409329461598,
-      "grad_norm": 1.823115348815918,
-      "learning_rate": 6.644501278772379e-06,
-      "loss": 0.1644,
-      "step": 1968
-    },
-    {
-      "epoch": 0.3357114313876975,
-      "grad_norm": 0.6293953657150269,
-      "learning_rate": 6.642796248934357e-06,
-      "loss": 0.1359,
-      "step": 1969
-    },
-    {
-      "epoch": 0.33588192982923515,
-      "grad_norm": 1.1457743644714355,
-      "learning_rate": 6.641091219096335e-06,
-      "loss": 0.1746,
-      "step": 1970
-    },
-    {
-      "epoch": 0.33605242827077286,
-      "grad_norm": 0.7488301396369934,
-      "learning_rate": 6.639386189258312e-06,
-      "loss": 0.0864,
-      "step": 1971
-    },
-    {
-      "epoch": 0.3362229267123105,
-      "grad_norm": 0.9744401574134827,
-      "learning_rate": 6.637681159420291e-06,
-      "loss": 0.1332,
-      "step": 1972
-    },
-    {
-      "epoch": 0.3363934251538482,
-      "grad_norm": 0.7821024060249329,
-      "learning_rate": 6.635976129582268e-06,
-      "loss": 0.1356,
-      "step": 1973
-    },
-    {
-      "epoch": 0.3365639235953859,
-      "grad_norm": 1.1750373840332031,
-      "learning_rate": 6.634271099744246e-06,
-      "loss": 0.1853,
-      "step": 1974
-    },
-    {
-      "epoch": 0.33673442203692355,
-      "grad_norm": 1.186582326889038,
-      "learning_rate": 6.632566069906225e-06,
-      "loss": 0.1315,
-      "step": 1975
-    },
-    {
-      "epoch": 0.33690492047846127,
-      "grad_norm": 1.009081482887268,
-      "learning_rate": 6.630861040068202e-06,
-      "loss": 0.1265,
-      "step": 1976
-    },
-    {
-      "epoch": 0.3370754189199989,
-      "grad_norm": 1.5751765966415405,
-      "learning_rate": 6.62915601023018e-06,
-      "loss": 0.1152,
-      "step": 1977
-    },
-    {
-      "epoch": 0.33724591736153664,
-      "grad_norm": 0.9581623673439026,
-      "learning_rate": 6.627450980392157e-06,
-      "loss": 0.1351,
-      "step": 1978
-    },
-    {
-      "epoch": 0.3374164158030743,
-      "grad_norm": 0.7846093773841858,
-      "learning_rate": 6.625745950554136e-06,
-      "loss": 0.142,
-      "step": 1979
-    },
-    {
-      "epoch": 0.33758691424461196,
-      "grad_norm": 1.0404683351516724,
-      "learning_rate": 6.624040920716113e-06,
-      "loss": 0.176,
-      "step": 1980
-    },
-    {
-      "epoch": 0.33775741268614967,
-      "grad_norm": 0.6047057509422302,
-      "learning_rate": 6.622335890878091e-06,
-      "loss": 0.1395,
-      "step": 1981
-    },
-    {
-      "epoch": 0.33792791112768733,
-      "grad_norm": 0.870232343673706,
-      "learning_rate": 6.620630861040068e-06,
-      "loss": 0.1167,
-      "step": 1982
-    },
-    {
-      "epoch": 0.33809840956922504,
-      "grad_norm": 0.7972772121429443,
-      "learning_rate": 6.618925831202047e-06,
-      "loss": 0.1582,
-      "step": 1983
-    },
-    {
-      "epoch": 0.3382689080107627,
-      "grad_norm": 0.8232150673866272,
-      "learning_rate": 6.617220801364025e-06,
-      "loss": 0.1336,
-      "step": 1984
-    },
-    {
-      "epoch": 0.3384394064523004,
-      "grad_norm": 0.8988503813743591,
-      "learning_rate": 6.615515771526002e-06,
-      "loss": 0.0448,
-      "step": 1985
-    },
-    {
-      "epoch": 0.3386099048938381,
-      "grad_norm": 0.8263369798660278,
-      "learning_rate": 6.6138107416879805e-06,
-      "loss": 0.0886,
-      "step": 1986
-    },
-    {
-      "epoch": 0.33878040333537573,
-      "grad_norm": 0.8381589651107788,
-      "learning_rate": 6.612105711849958e-06,
-      "loss": 0.1685,
-      "step": 1987
-    },
-    {
-      "epoch": 0.33895090177691345,
-      "grad_norm": 1.221974492073059,
-      "learning_rate": 6.610400682011936e-06,
-      "loss": 0.1797,
-      "step": 1988
-    },
-    {
-      "epoch": 0.3391214002184511,
-      "grad_norm": 1.0716266632080078,
-      "learning_rate": 6.6086956521739135e-06,
-      "loss": 0.1757,
-      "step": 1989
-    },
-    {
-      "epoch": 0.3392918986599888,
-      "grad_norm": 0.771088182926178,
-      "learning_rate": 6.6069906223358914e-06,
-      "loss": 0.136,
-      "step": 1990
-    },
-    {
-      "epoch": 0.3394623971015265,
-      "grad_norm": 0.6183826923370361,
-      "learning_rate": 6.6052855924978685e-06,
-      "loss": 0.1254,
-      "step": 1991
-    },
-    {
-      "epoch": 0.3396328955430642,
-      "grad_norm": 1.1764880418777466,
-      "learning_rate": 6.603580562659847e-06,
-      "loss": 0.1501,
-      "step": 1992
-    },
-    {
-      "epoch": 0.33980339398460185,
-      "grad_norm": 1.0482877492904663,
-      "learning_rate": 6.601875532821825e-06,
-      "loss": 0.2018,
-      "step": 1993
-    },
-    {
-      "epoch": 0.33997389242613957,
-      "grad_norm": 0.6771042943000793,
-      "learning_rate": 6.600170502983802e-06,
-      "loss": 0.0886,
-      "step": 1994
-    },
-    {
-      "epoch": 0.3401443908676772,
-      "grad_norm": 1.0513336658477783,
-      "learning_rate": 6.598465473145781e-06,
-      "loss": 0.158,
-      "step": 1995
-    },
-    {
-      "epoch": 0.3403148893092149,
-      "grad_norm": 0.6636187434196472,
-      "learning_rate": 6.596760443307758e-06,
-      "loss": 0.1076,
-      "step": 1996
-    },
-    {
-      "epoch": 0.3404853877507526,
-      "grad_norm": 0.9513442516326904,
-      "learning_rate": 6.595055413469736e-06,
-      "loss": 0.1058,
-      "step": 1997
-    },
-    {
-      "epoch": 0.34065588619229026,
-      "grad_norm": 0.9921936392784119,
-      "learning_rate": 6.593350383631713e-06,
-      "loss": 0.1267,
-      "step": 1998
-    },
-    {
-      "epoch": 0.340826384633828,
-      "grad_norm": 0.9132459759712219,
-      "learning_rate": 6.591645353793692e-06,
-      "loss": 0.1163,
-      "step": 1999
-    },
-    {
-      "epoch": 0.34099688307536563,
-      "grad_norm": 0.9271344542503357,
-      "learning_rate": 6.58994032395567e-06,
-      "loss": 0.1207,
-      "step": 2000
-    },
-    {
-      "epoch": 0.34099688307536563,
-      "eval_f1_score": 0.297029702970297,
-      "eval_loss": 0.15308597683906555,
-      "eval_runtime": 182.6807,
-      "eval_samples_per_second": 54.74,
-      "eval_steps_per_second": 3.421,
-      "step": 2000
-    },
-    {
-      "epoch": 0.34116738151690335,
-      "grad_norm": 1.0865306854248047,
-      "learning_rate": 6.588235294117647e-06,
-      "loss": 0.1873,
-      "step": 2001
-    },
-    {
-      "epoch": 0.341337879958441,
-      "grad_norm": 0.7954295873641968,
-      "learning_rate": 6.586530264279626e-06,
-      "loss": 0.1125,
-      "step": 2002
-    },
-    {
-      "epoch": 0.34150837839997866,
-      "grad_norm": 0.9557849764823914,
-      "learning_rate": 6.584825234441603e-06,
-      "loss": 0.1104,
-      "step": 2003
-    },
-    {
-      "epoch": 0.3416788768415164,
-      "grad_norm": 0.7364252805709839,
-      "learning_rate": 6.583120204603581e-06,
-      "loss": 0.1215,
-      "step": 2004
-    },
-    {
-      "epoch": 0.34184937528305404,
-      "grad_norm": 1.0843329429626465,
-      "learning_rate": 6.581415174765559e-06,
-      "loss": 0.2042,
-      "step": 2005
-    },
-    {
-      "epoch": 0.34201987372459175,
-      "grad_norm": 1.011513113975525,
-      "learning_rate": 6.579710144927537e-06,
-      "loss": 0.1623,
-      "step": 2006
-    },
-    {
-      "epoch": 0.3421903721661294,
-      "grad_norm": 0.7970641851425171,
-      "learning_rate": 6.578005115089514e-06,
-      "loss": 0.1766,
-      "step": 2007
-    },
-    {
-      "epoch": 0.3423608706076671,
-      "grad_norm": 1.012313961982727,
-      "learning_rate": 6.576300085251493e-06,
-      "loss": 0.1507,
-      "step": 2008
-    },
-    {
-      "epoch": 0.3425313690492048,
-      "grad_norm": 0.9937087893486023,
-      "learning_rate": 6.574595055413471e-06,
-      "loss": 0.1433,
-      "step": 2009
-    },
-    {
-      "epoch": 0.34270186749074244,
-      "grad_norm": 1.0500171184539795,
-      "learning_rate": 6.572890025575448e-06,
-      "loss": 0.1619,
-      "step": 2010
-    },
-    {
-      "epoch": 0.34287236593228015,
-      "grad_norm": 0.9035862684249878,
-      "learning_rate": 6.571184995737427e-06,
-      "loss": 0.1447,
-      "step": 2011
-    },
-    {
-      "epoch": 0.3430428643738178,
-      "grad_norm": 0.9435558319091797,
-      "learning_rate": 6.569479965899404e-06,
-      "loss": 0.1989,
-      "step": 2012
-    },
-    {
-      "epoch": 0.3432133628153555,
-      "grad_norm": 0.7556560039520264,
-      "learning_rate": 6.567774936061382e-06,
-      "loss": 0.1044,
-      "step": 2013
-    },
-    {
-      "epoch": 0.3433838612568932,
-      "grad_norm": 1.0829527378082275,
-      "learning_rate": 6.566069906223359e-06,
-      "loss": 0.1815,
-      "step": 2014
-    },
-    {
-      "epoch": 0.3435543596984309,
-      "grad_norm": 0.760651707649231,
-      "learning_rate": 6.564364876385338e-06,
-      "loss": 0.1291,
-      "step": 2015
-    },
-    {
-      "epoch": 0.34372485813996856,
-      "grad_norm": 0.8749564290046692,
-      "learning_rate": 6.562659846547315e-06,
-      "loss": 0.1251,
-      "step": 2016
-    },
-    {
-      "epoch": 0.3438953565815063,
-      "grad_norm": 0.8915442228317261,
-      "learning_rate": 6.560954816709293e-06,
-      "loss": 0.2182,
-      "step": 2017
-    },
-    {
-      "epoch": 0.34406585502304393,
-      "grad_norm": 1.0313383340835571,
-      "learning_rate": 6.559249786871271e-06,
-      "loss": 0.0874,
-      "step": 2018
-    },
-    {
-      "epoch": 0.3442363534645816,
-      "grad_norm": 1.4438647031784058,
-      "learning_rate": 6.5575447570332485e-06,
-      "loss": 0.2466,
-      "step": 2019
-    },
-    {
-      "epoch": 0.3444068519061193,
-      "grad_norm": 0.759958028793335,
-      "learning_rate": 6.5558397271952265e-06,
-      "loss": 0.1124,
-      "step": 2020
-    },
-    {
-      "epoch": 0.34457735034765696,
-      "grad_norm": 0.8678929805755615,
-      "learning_rate": 6.5541346973572036e-06,
-      "loss": 0.098,
-      "step": 2021
-    },
-    {
-      "epoch": 0.3447478487891947,
-      "grad_norm": 0.7600451111793518,
-      "learning_rate": 6.552429667519182e-06,
-      "loss": 0.0868,
-      "step": 2022
-    },
-    {
-      "epoch": 0.34491834723073234,
-      "grad_norm": 0.9153589606285095,
-      "learning_rate": 6.5507246376811595e-06,
-      "loss": 0.1659,
-      "step": 2023
-    },
-    {
-      "epoch": 0.34508884567227005,
-      "grad_norm": 0.9439913034439087,
-      "learning_rate": 6.549019607843137e-06,
-      "loss": 0.1752,
-      "step": 2024
-    },
-    {
-      "epoch": 0.3452593441138077,
-      "grad_norm": 0.7293012142181396,
-      "learning_rate": 6.547314578005116e-06,
-      "loss": 0.1106,
-      "step": 2025
-    },
-    {
-      "epoch": 0.34542984255534537,
-      "grad_norm": 1.9949567317962646,
-      "learning_rate": 6.545609548167093e-06,
-      "loss": 0.132,
-      "step": 2026
-    },
-    {
-      "epoch": 0.3456003409968831,
-      "grad_norm": 0.8462551832199097,
-      "learning_rate": 6.543904518329072e-06,
-      "loss": 0.1452,
-      "step": 2027
-    },
-    {
-      "epoch": 0.34577083943842074,
-      "grad_norm": 0.6383141279220581,
-      "learning_rate": 6.542199488491049e-06,
-      "loss": 0.1229,
-      "step": 2028
-    },
-    {
-      "epoch": 0.34594133787995845,
-      "grad_norm": 0.6529576778411865,
-      "learning_rate": 6.540494458653027e-06,
-      "loss": 0.0587,
-      "step": 2029
-    },
-    {
-      "epoch": 0.3461118363214961,
-      "grad_norm": 1.0192657709121704,
-      "learning_rate": 6.538789428815004e-06,
-      "loss": 0.1613,
-      "step": 2030
-    },
-    {
-      "epoch": 0.34628233476303383,
-      "grad_norm": 0.8856669068336487,
-      "learning_rate": 6.537084398976983e-06,
-      "loss": 0.1531,
-      "step": 2031
-    },
-    {
-      "epoch": 0.3464528332045715,
-      "grad_norm": 1.6244677305221558,
-      "learning_rate": 6.53537936913896e-06,
-      "loss": 0.1339,
-      "step": 2032
-    },
-    {
-      "epoch": 0.34662333164610915,
-      "grad_norm": 0.8295205235481262,
-      "learning_rate": 6.533674339300938e-06,
-      "loss": 0.0454,
-      "step": 2033
-    },
-    {
-      "epoch": 0.34679383008764686,
-      "grad_norm": 1.043818712234497,
-      "learning_rate": 6.531969309462917e-06,
-      "loss": 0.143,
-      "step": 2034
-    },
-    {
-      "epoch": 0.3469643285291845,
-      "grad_norm": 0.7247282266616821,
-      "learning_rate": 6.530264279624894e-06,
-      "loss": 0.1549,
-      "step": 2035
-    },
-    {
-      "epoch": 0.34713482697072223,
-      "grad_norm": 1.0297534465789795,
-      "learning_rate": 6.528559249786872e-06,
-      "loss": 0.1885,
-      "step": 2036
-    },
-    {
-      "epoch": 0.3473053254122599,
-      "grad_norm": 0.8988025784492493,
-      "learning_rate": 6.526854219948849e-06,
-      "loss": 0.1602,
-      "step": 2037
-    },
-    {
-      "epoch": 0.3474758238537976,
-      "grad_norm": 0.9362964630126953,
-      "learning_rate": 6.525149190110828e-06,
-      "loss": 0.1479,
-      "step": 2038
-    },
-    {
-      "epoch": 0.34764632229533526,
-      "grad_norm": 0.8799692392349243,
-      "learning_rate": 6.523444160272805e-06,
-      "loss": 0.0913,
-      "step": 2039
-    },
-    {
-      "epoch": 0.347816820736873,
-      "grad_norm": 0.5634396076202393,
-      "learning_rate": 6.521739130434783e-06,
-      "loss": 0.0844,
-      "step": 2040
-    },
-    {
-      "epoch": 0.34798731917841064,
-      "grad_norm": 1.2817353010177612,
-      "learning_rate": 6.520034100596761e-06,
-      "loss": 0.1898,
-      "step": 2041
-    },
-    {
-      "epoch": 0.3481578176199483,
-      "grad_norm": 0.9581865072250366,
-      "learning_rate": 6.518329070758739e-06,
-      "loss": 0.1107,
-      "step": 2042
-    },
-    {
-      "epoch": 0.348328316061486,
-      "grad_norm": 0.8063405156135559,
-      "learning_rate": 6.516624040920717e-06,
-      "loss": 0.1494,
-      "step": 2043
-    },
-    {
-      "epoch": 0.34849881450302367,
-      "grad_norm": 0.8312709927558899,
-      "learning_rate": 6.514919011082695e-06,
-      "loss": 0.1413,
-      "step": 2044
-    },
-    {
-      "epoch": 0.3486693129445614,
-      "grad_norm": 0.8263598680496216,
-      "learning_rate": 6.513213981244673e-06,
-      "loss": 0.1417,
-      "step": 2045
-    },
-    {
-      "epoch": 0.34883981138609904,
-      "grad_norm": 0.9109250903129578,
-      "learning_rate": 6.51150895140665e-06,
-      "loss": 0.1575,
-      "step": 2046
-    },
-    {
-      "epoch": 0.34901030982763676,
-      "grad_norm": 0.9140591025352478,
-      "learning_rate": 6.5098039215686285e-06,
-      "loss": 0.1029,
-      "step": 2047
-    },
-    {
-      "epoch": 0.3491808082691744,
-      "grad_norm": 0.8870648145675659,
-      "learning_rate": 6.508098891730606e-06,
-      "loss": 0.0993,
-      "step": 2048
-    },
-    {
-      "epoch": 0.3493513067107121,
-      "grad_norm": 0.7272403240203857,
-      "learning_rate": 6.5063938618925836e-06,
-      "loss": 0.1002,
-      "step": 2049
-    },
-    {
-      "epoch": 0.3495218051522498,
-      "grad_norm": 1.034851312637329,
-      "learning_rate": 6.504688832054562e-06,
-      "loss": 0.1463,
-      "step": 2050
-    },
-    {
-      "epoch": 0.34969230359378745,
-      "grad_norm": 0.9596707224845886,
-      "learning_rate": 6.5029838022165395e-06,
-      "loss": 0.0776,
-      "step": 2051
-    },
-    {
-      "epoch": 0.34986280203532516,
-      "grad_norm": 0.6040682792663574,
-      "learning_rate": 6.501278772378517e-06,
-      "loss": 0.0987,
-      "step": 2052
-    },
-    {
-      "epoch": 0.3500333004768628,
-      "grad_norm": 1.2424696683883667,
-      "learning_rate": 6.4995737425404945e-06,
-      "loss": 0.1698,
-      "step": 2053
-    },
-    {
-      "epoch": 0.35020379891840053,
-      "grad_norm": 0.6647294163703918,
-      "learning_rate": 6.497868712702473e-06,
-      "loss": 0.0825,
-      "step": 2054
-    },
-    {
-      "epoch": 0.3503742973599382,
-      "grad_norm": 1.229026436805725,
-      "learning_rate": 6.49616368286445e-06,
-      "loss": 0.1803,
-      "step": 2055
-    },
-    {
-      "epoch": 0.35054479580147585,
-      "grad_norm": 1.356615662574768,
-      "learning_rate": 6.494458653026428e-06,
-      "loss": 0.2154,
-      "step": 2056
-    },
-    {
-      "epoch": 0.35071529424301356,
-      "grad_norm": 0.7908090353012085,
-      "learning_rate": 6.4927536231884055e-06,
-      "loss": 0.1286,
-      "step": 2057
-    },
-    {
-      "epoch": 0.3508857926845512,
-      "grad_norm": 1.1387441158294678,
-      "learning_rate": 6.491048593350384e-06,
-      "loss": 0.1859,
-      "step": 2058
-    },
-    {
-      "epoch": 0.35105629112608894,
-      "grad_norm": 0.7091562151908875,
-      "learning_rate": 6.489343563512362e-06,
-      "loss": 0.1258,
-      "step": 2059
-    },
-    {
-      "epoch": 0.3512267895676266,
-      "grad_norm": 0.978642463684082,
-      "learning_rate": 6.487638533674339e-06,
-      "loss": 0.1429,
-      "step": 2060
-    },
-    {
-      "epoch": 0.3513972880091643,
-      "grad_norm": 0.9794069528579712,
-      "learning_rate": 6.485933503836318e-06,
-      "loss": 0.1821,
-      "step": 2061
-    },
-    {
-      "epoch": 0.35156778645070197,
-      "grad_norm": 1.1713330745697021,
-      "learning_rate": 6.484228473998295e-06,
-      "loss": 0.1636,
-      "step": 2062
-    },
-    {
-      "epoch": 0.3517382848922397,
-      "grad_norm": 0.9776598811149597,
-      "learning_rate": 6.482523444160274e-06,
-      "loss": 0.1428,
-      "step": 2063
-    },
-    {
-      "epoch": 0.35190878333377734,
-      "grad_norm": 0.7179242372512817,
-      "learning_rate": 6.480818414322251e-06,
-      "loss": 0.0813,
-      "step": 2064
-    },
-    {
-      "epoch": 0.352079281775315,
-      "grad_norm": 0.9282951354980469,
-      "learning_rate": 6.479113384484229e-06,
-      "loss": 0.1871,
-      "step": 2065
-    },
-    {
-      "epoch": 0.3522497802168527,
-      "grad_norm": 0.9170055389404297,
-      "learning_rate": 6.477408354646206e-06,
-      "loss": 0.1127,
-      "step": 2066
-    },
-    {
-      "epoch": 0.3524202786583904,
-      "grad_norm": 1.0189080238342285,
-      "learning_rate": 6.475703324808185e-06,
-      "loss": 0.1624,
-      "step": 2067
-    },
-    {
-      "epoch": 0.3525907770999281,
-      "grad_norm": 0.6744734644889832,
-      "learning_rate": 6.473998294970163e-06,
-      "loss": 0.0822,
-      "step": 2068
-    },
-    {
-      "epoch": 0.35276127554146575,
-      "grad_norm": 1.0721924304962158,
-      "learning_rate": 6.47229326513214e-06,
-      "loss": 0.1305,
-      "step": 2069
-    },
-    {
-      "epoch": 0.35293177398300346,
-      "grad_norm": 0.8503794074058533,
-      "learning_rate": 6.470588235294119e-06,
-      "loss": 0.1042,
-      "step": 2070
-    },
-    {
-      "epoch": 0.3531022724245411,
-      "grad_norm": 0.9037670493125916,
-      "learning_rate": 6.468883205456096e-06,
-      "loss": 0.0831,
-      "step": 2071
-    },
-    {
-      "epoch": 0.3532727708660788,
-      "grad_norm": 0.6255829334259033,
-      "learning_rate": 6.467178175618074e-06,
-      "loss": 0.0766,
-      "step": 2072
-    },
-    {
-      "epoch": 0.3534432693076165,
-      "grad_norm": 0.6307828426361084,
-      "learning_rate": 6.465473145780051e-06,
-      "loss": 0.0733,
-      "step": 2073
-    },
-    {
-      "epoch": 0.35361376774915415,
-      "grad_norm": 0.819678008556366,
-      "learning_rate": 6.46376811594203e-06,
-      "loss": 0.1662,
-      "step": 2074
-    },
-    {
-      "epoch": 0.35378426619069187,
-      "grad_norm": 0.6275694966316223,
-      "learning_rate": 6.462063086104008e-06,
-      "loss": 0.0859,
-      "step": 2075
-    },
-    {
-      "epoch": 0.3539547646322295,
-      "grad_norm": 0.8501904010772705,
-      "learning_rate": 6.460358056265985e-06,
-      "loss": 0.1721,
-      "step": 2076
-    },
-    {
-      "epoch": 0.35412526307376724,
-      "grad_norm": 0.9360889792442322,
-      "learning_rate": 6.4586530264279636e-06,
-      "loss": 0.1261,
-      "step": 2077
-    },
-    {
-      "epoch": 0.3542957615153049,
-      "grad_norm": 1.2045263051986694,
-      "learning_rate": 6.456947996589941e-06,
-      "loss": 0.2182,
-      "step": 2078
-    },
-    {
-      "epoch": 0.35446625995684256,
-      "grad_norm": 0.9660459160804749,
-      "learning_rate": 6.455242966751919e-06,
-      "loss": 0.1774,
-      "step": 2079
-    },
-    {
-      "epoch": 0.35463675839838027,
-      "grad_norm": 1.315540075302124,
-      "learning_rate": 6.4535379369138966e-06,
-      "loss": 0.2046,
-      "step": 2080
-    },
-    {
-      "epoch": 0.35480725683991793,
-      "grad_norm": 0.9870008826255798,
-      "learning_rate": 6.4518329070758745e-06,
-      "loss": 0.1305,
-      "step": 2081
-    },
-    {
-      "epoch": 0.35497775528145564,
-      "grad_norm": 1.0989959239959717,
-      "learning_rate": 6.450127877237852e-06,
-      "loss": 0.172,
-      "step": 2082
-    },
-    {
-      "epoch": 0.3551482537229933,
-      "grad_norm": 0.7137494087219238,
-      "learning_rate": 6.44842284739983e-06,
-      "loss": 0.1374,
-      "step": 2083
-    },
-    {
-      "epoch": 0.355318752164531,
-      "grad_norm": 0.9373939633369446,
-      "learning_rate": 6.446717817561808e-06,
-      "loss": 0.1736,
-      "step": 2084
-    },
-    {
-      "epoch": 0.3554892506060687,
-      "grad_norm": 0.7139600515365601,
-      "learning_rate": 6.4450127877237854e-06,
-      "loss": 0.0822,
-      "step": 2085
-    },
-    {
-      "epoch": 0.3556597490476064,
-      "grad_norm": 1.1834203004837036,
-      "learning_rate": 6.443307757885764e-06,
-      "loss": 0.1904,
-      "step": 2086
-    },
-    {
-      "epoch": 0.35583024748914405,
-      "grad_norm": 1.1191166639328003,
-      "learning_rate": 6.441602728047741e-06,
-      "loss": 0.1758,
-      "step": 2087
-    },
-    {
-      "epoch": 0.3560007459306817,
-      "grad_norm": 0.8284562826156616,
-      "learning_rate": 6.439897698209719e-06,
-      "loss": 0.1564,
-      "step": 2088
-    },
-    {
-      "epoch": 0.3561712443722194,
-      "grad_norm": 1.2975763082504272,
-      "learning_rate": 6.438192668371696e-06,
-      "loss": 0.1769,
-      "step": 2089
-    },
-    {
-      "epoch": 0.3563417428137571,
-      "grad_norm": 0.823640763759613,
-      "learning_rate": 6.436487638533675e-06,
-      "loss": 0.1212,
-      "step": 2090
-    },
-    {
-      "epoch": 0.3565122412552948,
-      "grad_norm": 0.904367208480835,
-      "learning_rate": 6.434782608695652e-06,
-      "loss": 0.1113,
-      "step": 2091
-    },
-    {
-      "epoch": 0.35668273969683245,
-      "grad_norm": 0.9860475659370422,
-      "learning_rate": 6.43307757885763e-06,
-      "loss": 0.1047,
-      "step": 2092
-    },
-    {
-      "epoch": 0.35685323813837017,
-      "grad_norm": 1.125466227531433,
-      "learning_rate": 6.431372549019609e-06,
-      "loss": 0.1366,
-      "step": 2093
-    },
-    {
-      "epoch": 0.3570237365799078,
-      "grad_norm": 0.9330949783325195,
-      "learning_rate": 6.429667519181586e-06,
-      "loss": 0.1179,
-      "step": 2094
-    },
-    {
-      "epoch": 0.3571942350214455,
-      "grad_norm": 0.9403000473976135,
-      "learning_rate": 6.427962489343564e-06,
-      "loss": 0.1651,
-      "step": 2095
-    },
-    {
-      "epoch": 0.3573647334629832,
-      "grad_norm": 0.839396059513092,
-      "learning_rate": 6.426257459505541e-06,
-      "loss": 0.132,
-      "step": 2096
-    },
-    {
-      "epoch": 0.35753523190452086,
-      "grad_norm": 0.8294098973274231,
-      "learning_rate": 6.42455242966752e-06,
-      "loss": 0.1451,
-      "step": 2097
-    },
-    {
-      "epoch": 0.35770573034605857,
-      "grad_norm": 0.7742642164230347,
-      "learning_rate": 6.422847399829497e-06,
-      "loss": 0.1309,
-      "step": 2098
-    },
-    {
-      "epoch": 0.35787622878759623,
-      "grad_norm": 0.7605495452880859,
-      "learning_rate": 6.421142369991475e-06,
-      "loss": 0.109,
-      "step": 2099
-    },
-    {
-      "epoch": 0.35804672722913394,
-      "grad_norm": 1.56952965259552,
-      "learning_rate": 6.419437340153453e-06,
-      "loss": 0.1774,
-      "step": 2100
-    },
-    {
-      "epoch": 0.35804672722913394,
-      "eval_f1_score": 0.27472527472527475,
-      "eval_loss": 0.1539647877216339,
-      "eval_runtime": 182.5857,
-      "eval_samples_per_second": 54.769,
-      "eval_steps_per_second": 3.423,
-      "step": 2100
-    },
-    {
-      "epoch": 0.3582172256706716,
-      "grad_norm": 0.6405848264694214,
-      "learning_rate": 6.417732310315431e-06,
-      "loss": 0.1012,
-      "step": 2101
-    },
-    {
-      "epoch": 0.35838772411220926,
-      "grad_norm": 1.0083062648773193,
-      "learning_rate": 6.41602728047741e-06,
-      "loss": 0.1576,
-      "step": 2102
-    },
-    {
-      "epoch": 0.358558222553747,
-      "grad_norm": 1.0331164598464966,
-      "learning_rate": 6.414322250639387e-06,
-      "loss": 0.0868,
-      "step": 2103
-    },
-    {
-      "epoch": 0.35872872099528463,
-      "grad_norm": 1.3799405097961426,
-      "learning_rate": 6.412617220801365e-06,
-      "loss": 0.167,
-      "step": 2104
-    },
-    {
-      "epoch": 0.35889921943682235,
-      "grad_norm": 1.3380049467086792,
-      "learning_rate": 6.410912190963342e-06,
-      "loss": 0.1638,
-      "step": 2105
-    },
-    {
-      "epoch": 0.35906971787836,
-      "grad_norm": 1.0186902284622192,
-      "learning_rate": 6.409207161125321e-06,
-      "loss": 0.1676,
-      "step": 2106
-    },
-    {
-      "epoch": 0.3592402163198977,
-      "grad_norm": 0.5658654570579529,
-      "learning_rate": 6.407502131287298e-06,
-      "loss": 0.0894,
-      "step": 2107
-    },
-    {
-      "epoch": 0.3594107147614354,
-      "grad_norm": 1.1313732862472534,
-      "learning_rate": 6.405797101449276e-06,
-      "loss": 0.1236,
-      "step": 2108
-    },
-    {
-      "epoch": 0.3595812132029731,
-      "grad_norm": 1.4047788381576538,
-      "learning_rate": 6.4040920716112545e-06,
-      "loss": 0.1463,
-      "step": 2109
-    },
-    {
-      "epoch": 0.35975171164451075,
-      "grad_norm": 1.1757776737213135,
-      "learning_rate": 6.402387041773232e-06,
-      "loss": 0.2258,
-      "step": 2110
-    },
-    {
-      "epoch": 0.3599222100860484,
-      "grad_norm": 1.0296345949172974,
-      "learning_rate": 6.4006820119352095e-06,
-      "loss": 0.1408,
-      "step": 2111
-    },
-    {
-      "epoch": 0.3600927085275861,
-      "grad_norm": 0.879396378993988,
-      "learning_rate": 6.398976982097187e-06,
-      "loss": 0.1294,
-      "step": 2112
-    },
-    {
-      "epoch": 0.3602632069691238,
-      "grad_norm": 1.3096729516983032,
-      "learning_rate": 6.3972719522591654e-06,
-      "loss": 0.2237,
-      "step": 2113
-    },
-    {
-      "epoch": 0.3604337054106615,
-      "grad_norm": 0.7490139007568359,
-      "learning_rate": 6.3955669224211425e-06,
-      "loss": 0.0969,
-      "step": 2114
-    },
-    {
-      "epoch": 0.36060420385219916,
-      "grad_norm": 0.7394745349884033,
-      "learning_rate": 6.3938618925831205e-06,
-      "loss": 0.1222,
-      "step": 2115
-    },
-    {
-      "epoch": 0.36077470229373687,
-      "grad_norm": 0.8975431323051453,
-      "learning_rate": 6.3921568627450984e-06,
-      "loss": 0.1205,
-      "step": 2116
-    },
-    {
-      "epoch": 0.36094520073527453,
-      "grad_norm": 1.3461922407150269,
-      "learning_rate": 6.390451832907076e-06,
-      "loss": 0.163,
-      "step": 2117
-    },
-    {
-      "epoch": 0.3611156991768122,
-      "grad_norm": 0.9350199103355408,
-      "learning_rate": 6.388746803069054e-06,
-      "loss": 0.177,
-      "step": 2118
-    },
-    {
-      "epoch": 0.3612861976183499,
-      "grad_norm": 1.1496632099151611,
-      "learning_rate": 6.387041773231032e-06,
-      "loss": 0.1248,
-      "step": 2119
-    },
-    {
-      "epoch": 0.36145669605988756,
-      "grad_norm": 1.1352415084838867,
-      "learning_rate": 6.38533674339301e-06,
-      "loss": 0.1746,
-      "step": 2120
-    },
-    {
-      "epoch": 0.3616271945014253,
-      "grad_norm": 0.9327718019485474,
-      "learning_rate": 6.383631713554987e-06,
-      "loss": 0.1904,
-      "step": 2121
-    },
-    {
-      "epoch": 0.36179769294296293,
-      "grad_norm": 0.7479307055473328,
-      "learning_rate": 6.381926683716966e-06,
-      "loss": 0.0699,
-      "step": 2122
-    },
-    {
-      "epoch": 0.36196819138450065,
-      "grad_norm": 0.8722934126853943,
-      "learning_rate": 6.380221653878943e-06,
-      "loss": 0.0925,
-      "step": 2123
-    },
-    {
-      "epoch": 0.3621386898260383,
-      "grad_norm": 1.530072569847107,
-      "learning_rate": 6.378516624040921e-06,
-      "loss": 0.1793,
-      "step": 2124
-    },
-    {
-      "epoch": 0.36230918826757597,
-      "grad_norm": 0.6114324927330017,
-      "learning_rate": 6.376811594202898e-06,
-      "loss": 0.0849,
-      "step": 2125
-    },
-    {
-      "epoch": 0.3624796867091137,
-      "grad_norm": 0.9590505957603455,
-      "learning_rate": 6.375106564364877e-06,
-      "loss": 0.121,
-      "step": 2126
-    },
-    {
-      "epoch": 0.36265018515065134,
-      "grad_norm": 0.6645707488059998,
-      "learning_rate": 6.373401534526855e-06,
-      "loss": 0.0806,
-      "step": 2127
-    },
-    {
-      "epoch": 0.36282068359218905,
-      "grad_norm": 0.6528565287590027,
-      "learning_rate": 6.371696504688832e-06,
-      "loss": 0.0899,
-      "step": 2128
-    },
-    {
-      "epoch": 0.3629911820337267,
-      "grad_norm": 1.1450191736221313,
-      "learning_rate": 6.369991474850811e-06,
-      "loss": 0.1427,
-      "step": 2129
-    },
-    {
-      "epoch": 0.3631616804752644,
-      "grad_norm": 1.0308457612991333,
-      "learning_rate": 6.368286445012788e-06,
-      "loss": 0.1379,
-      "step": 2130
-    },
-    {
-      "epoch": 0.3633321789168021,
-      "grad_norm": 1.4925493001937866,
-      "learning_rate": 6.366581415174766e-06,
-      "loss": 0.1987,
-      "step": 2131
-    },
-    {
-      "epoch": 0.3635026773583398,
-      "grad_norm": 0.7048131227493286,
-      "learning_rate": 6.364876385336743e-06,
-      "loss": 0.1062,
-      "step": 2132
-    },
-    {
-      "epoch": 0.36367317579987746,
-      "grad_norm": 0.8379212617874146,
-      "learning_rate": 6.363171355498722e-06,
-      "loss": 0.1552,
-      "step": 2133
-    },
-    {
-      "epoch": 0.3638436742414151,
-      "grad_norm": 1.1558237075805664,
-      "learning_rate": 6.3614663256607e-06,
-      "loss": 0.1038,
-      "step": 2134
-    },
-    {
-      "epoch": 0.36401417268295283,
-      "grad_norm": 0.670208752155304,
-      "learning_rate": 6.359761295822677e-06,
-      "loss": 0.0811,
-      "step": 2135
-    },
-    {
-      "epoch": 0.3641846711244905,
-      "grad_norm": 1.4409935474395752,
-      "learning_rate": 6.358056265984656e-06,
-      "loss": 0.1946,
-      "step": 2136
-    },
-    {
-      "epoch": 0.3643551695660282,
-      "grad_norm": 0.7791793346405029,
-      "learning_rate": 6.356351236146633e-06,
-      "loss": 0.1485,
-      "step": 2137
-    },
-    {
-      "epoch": 0.36452566800756586,
-      "grad_norm": 1.1148107051849365,
-      "learning_rate": 6.354646206308612e-06,
-      "loss": 0.2146,
-      "step": 2138
-    },
-    {
-      "epoch": 0.3646961664491036,
-      "grad_norm": 0.9492986798286438,
-      "learning_rate": 6.352941176470589e-06,
-      "loss": 0.1227,
-      "step": 2139
-    },
-    {
-      "epoch": 0.36486666489064123,
-      "grad_norm": 0.7307665348052979,
-      "learning_rate": 6.351236146632567e-06,
-      "loss": 0.0764,
-      "step": 2140
-    },
-    {
-      "epoch": 0.3650371633321789,
-      "grad_norm": 1.2648804187774658,
-      "learning_rate": 6.349531116794544e-06,
-      "loss": 0.1314,
-      "step": 2141
-    },
-    {
-      "epoch": 0.3652076617737166,
-      "grad_norm": 0.7799243927001953,
-      "learning_rate": 6.3478260869565225e-06,
-      "loss": 0.0747,
-      "step": 2142
-    },
-    {
-      "epoch": 0.36537816021525427,
-      "grad_norm": 0.7744673490524292,
-      "learning_rate": 6.3461210571185005e-06,
-      "loss": 0.1224,
-      "step": 2143
-    },
-    {
-      "epoch": 0.365548658656792,
-      "grad_norm": 0.9403333067893982,
-      "learning_rate": 6.344416027280478e-06,
-      "loss": 0.0992,
-      "step": 2144
-    },
-    {
-      "epoch": 0.36571915709832964,
-      "grad_norm": 0.9702940583229065,
-      "learning_rate": 6.342710997442456e-06,
-      "loss": 0.1358,
-      "step": 2145
-    },
-    {
-      "epoch": 0.36588965553986735,
-      "grad_norm": 1.0209053754806519,
-      "learning_rate": 6.3410059676044335e-06,
-      "loss": 0.116,
-      "step": 2146
-    },
-    {
-      "epoch": 0.366060153981405,
-      "grad_norm": 0.7160688042640686,
-      "learning_rate": 6.3393009377664114e-06,
-      "loss": 0.1244,
-      "step": 2147
-    },
-    {
-      "epoch": 0.36623065242294267,
-      "grad_norm": 1.0120749473571777,
-      "learning_rate": 6.3375959079283885e-06,
-      "loss": 0.1397,
-      "step": 2148
-    },
-    {
-      "epoch": 0.3664011508644804,
-      "grad_norm": 0.6767970323562622,
-      "learning_rate": 6.335890878090367e-06,
-      "loss": 0.1211,
-      "step": 2149
-    },
-    {
-      "epoch": 0.36657164930601804,
-      "grad_norm": 0.5627476572990417,
-      "learning_rate": 6.3341858482523444e-06,
-      "loss": 0.0939,
-      "step": 2150
-    },
-    {
-      "epoch": 0.36674214774755576,
-      "grad_norm": 0.6991590261459351,
-      "learning_rate": 6.332480818414322e-06,
-      "loss": 0.1212,
-      "step": 2151
-    },
-    {
-      "epoch": 0.3669126461890934,
-      "grad_norm": 1.034399151802063,
-      "learning_rate": 6.330775788576301e-06,
-      "loss": 0.1489,
-      "step": 2152
-    },
-    {
-      "epoch": 0.36708314463063113,
-      "grad_norm": 0.8625597953796387,
-      "learning_rate": 6.329070758738278e-06,
-      "loss": 0.0497,
-      "step": 2153
-    },
-    {
-      "epoch": 0.3672536430721688,
-      "grad_norm": 1.123618721961975,
-      "learning_rate": 6.327365728900256e-06,
-      "loss": 0.207,
-      "step": 2154
-    },
-    {
-      "epoch": 0.3674241415137065,
-      "grad_norm": 2.084270715713501,
-      "learning_rate": 6.325660699062234e-06,
-      "loss": 0.1573,
-      "step": 2155
-    },
-    {
-      "epoch": 0.36759463995524416,
-      "grad_norm": 0.6721711754798889,
-      "learning_rate": 6.323955669224212e-06,
-      "loss": 0.1054,
-      "step": 2156
-    },
-    {
-      "epoch": 0.3677651383967818,
-      "grad_norm": 1.4094241857528687,
-      "learning_rate": 6.322250639386189e-06,
-      "loss": 0.1926,
-      "step": 2157
-    },
-    {
-      "epoch": 0.36793563683831954,
-      "grad_norm": 1.0103540420532227,
-      "learning_rate": 6.320545609548168e-06,
-      "loss": 0.1321,
-      "step": 2158
-    },
-    {
-      "epoch": 0.3681061352798572,
-      "grad_norm": 0.9904351234436035,
-      "learning_rate": 6.318840579710146e-06,
-      "loss": 0.096,
-      "step": 2159
-    },
-    {
-      "epoch": 0.3682766337213949,
-      "grad_norm": 1.041150689125061,
-      "learning_rate": 6.317135549872123e-06,
-      "loss": 0.126,
-      "step": 2160
-    },
-    {
-      "epoch": 0.36844713216293257,
-      "grad_norm": 1.380527138710022,
-      "learning_rate": 6.315430520034102e-06,
-      "loss": 0.2002,
-      "step": 2161
-    },
-    {
-      "epoch": 0.3686176306044703,
-      "grad_norm": 0.9068813323974609,
-      "learning_rate": 6.313725490196079e-06,
-      "loss": 0.1357,
-      "step": 2162
-    },
-    {
-      "epoch": 0.36878812904600794,
-      "grad_norm": 0.8607088923454285,
-      "learning_rate": 6.312020460358057e-06,
-      "loss": 0.0703,
-      "step": 2163
-    },
-    {
-      "epoch": 0.3689586274875456,
-      "grad_norm": 0.7945460677146912,
-      "learning_rate": 6.310315430520034e-06,
-      "loss": 0.1227,
-      "step": 2164
-    },
-    {
-      "epoch": 0.3691291259290833,
-      "grad_norm": 1.271232008934021,
-      "learning_rate": 6.308610400682013e-06,
-      "loss": 0.1381,
-      "step": 2165
-    },
-    {
-      "epoch": 0.36929962437062097,
-      "grad_norm": 0.909999668598175,
-      "learning_rate": 6.30690537084399e-06,
-      "loss": 0.1043,
-      "step": 2166
-    },
-    {
-      "epoch": 0.3694701228121587,
-      "grad_norm": 1.6058598756790161,
-      "learning_rate": 6.305200341005968e-06,
-      "loss": 0.2397,
-      "step": 2167
-    },
-    {
-      "epoch": 0.36964062125369634,
-      "grad_norm": 0.7543137669563293,
-      "learning_rate": 6.303495311167947e-06,
-      "loss": 0.13,
-      "step": 2168
-    },
-    {
-      "epoch": 0.36981111969523406,
-      "grad_norm": 1.472127914428711,
-      "learning_rate": 6.301790281329924e-06,
-      "loss": 0.1905,
-      "step": 2169
-    },
-    {
-      "epoch": 0.3699816181367717,
-      "grad_norm": 1.0304391384124756,
-      "learning_rate": 6.300085251491902e-06,
-      "loss": 0.1102,
-      "step": 2170
-    },
-    {
-      "epoch": 0.3701521165783094,
-      "grad_norm": 0.9876975417137146,
-      "learning_rate": 6.298380221653879e-06,
-      "loss": 0.2271,
-      "step": 2171
-    },
-    {
-      "epoch": 0.3703226150198471,
-      "grad_norm": 1.176422357559204,
-      "learning_rate": 6.296675191815858e-06,
-      "loss": 0.1687,
-      "step": 2172
-    },
-    {
-      "epoch": 0.37049311346138475,
-      "grad_norm": 1.720004916191101,
-      "learning_rate": 6.294970161977835e-06,
-      "loss": 0.2063,
-      "step": 2173
-    },
-    {
-      "epoch": 0.37066361190292246,
-      "grad_norm": 0.9195306897163391,
-      "learning_rate": 6.293265132139813e-06,
-      "loss": 0.1601,
-      "step": 2174
-    },
-    {
-      "epoch": 0.3708341103444601,
-      "grad_norm": 1.4446338415145874,
-      "learning_rate": 6.291560102301791e-06,
-      "loss": 0.1781,
-      "step": 2175
-    },
-    {
-      "epoch": 0.37100460878599784,
-      "grad_norm": 1.0467724800109863,
-      "learning_rate": 6.2898550724637685e-06,
-      "loss": 0.1786,
-      "step": 2176
-    },
-    {
-      "epoch": 0.3711751072275355,
-      "grad_norm": 0.8518681526184082,
-      "learning_rate": 6.288150042625747e-06,
-      "loss": 0.137,
-      "step": 2177
-    },
-    {
-      "epoch": 0.3713456056690732,
-      "grad_norm": 1.0049577951431274,
-      "learning_rate": 6.286445012787724e-06,
-      "loss": 0.1792,
-      "step": 2178
-    },
-    {
-      "epoch": 0.37151610411061087,
-      "grad_norm": 1.1468156576156616,
-      "learning_rate": 6.284739982949702e-06,
-      "loss": 0.0627,
-      "step": 2179
-    },
-    {
-      "epoch": 0.3716866025521485,
-      "grad_norm": 1.0543559789657593,
-      "learning_rate": 6.2830349531116795e-06,
-      "loss": 0.1402,
-      "step": 2180
-    },
-    {
-      "epoch": 0.37185710099368624,
-      "grad_norm": 1.046242594718933,
-      "learning_rate": 6.281329923273658e-06,
-      "loss": 0.1341,
-      "step": 2181
-    },
-    {
-      "epoch": 0.3720275994352239,
-      "grad_norm": 0.9223925471305847,
-      "learning_rate": 6.279624893435635e-06,
-      "loss": 0.0667,
-      "step": 2182
-    },
-    {
-      "epoch": 0.3721980978767616,
-      "grad_norm": 0.9626695513725281,
-      "learning_rate": 6.277919863597613e-06,
-      "loss": 0.1337,
-      "step": 2183
-    },
-    {
-      "epoch": 0.37236859631829927,
-      "grad_norm": 0.7089354991912842,
-      "learning_rate": 6.276214833759592e-06,
-      "loss": 0.1158,
-      "step": 2184
-    },
-    {
-      "epoch": 0.372539094759837,
-      "grad_norm": 0.5326353311538696,
-      "learning_rate": 6.274509803921569e-06,
-      "loss": 0.0786,
-      "step": 2185
-    },
-    {
-      "epoch": 0.37270959320137464,
-      "grad_norm": 0.8547235727310181,
-      "learning_rate": 6.272804774083547e-06,
-      "loss": 0.1664,
-      "step": 2186
-    },
-    {
-      "epoch": 0.3728800916429123,
-      "grad_norm": 0.9501067996025085,
-      "learning_rate": 6.271099744245524e-06,
-      "loss": 0.1644,
-      "step": 2187
-    },
-    {
-      "epoch": 0.37305059008445,
-      "grad_norm": 0.8592075109481812,
-      "learning_rate": 6.269394714407503e-06,
-      "loss": 0.1002,
-      "step": 2188
-    },
-    {
-      "epoch": 0.3732210885259877,
-      "grad_norm": 1.842674732208252,
-      "learning_rate": 6.26768968456948e-06,
-      "loss": 0.2311,
-      "step": 2189
-    },
-    {
-      "epoch": 0.3733915869675254,
-      "grad_norm": 1.5288238525390625,
-      "learning_rate": 6.265984654731458e-06,
-      "loss": 0.1733,
-      "step": 2190
-    },
-    {
-      "epoch": 0.37356208540906305,
-      "grad_norm": 0.6214331388473511,
-      "learning_rate": 6.264279624893436e-06,
-      "loss": 0.1119,
-      "step": 2191
-    },
-    {
-      "epoch": 0.37373258385060076,
-      "grad_norm": 0.7459906935691833,
-      "learning_rate": 6.262574595055414e-06,
-      "loss": 0.0975,
-      "step": 2192
-    },
-    {
-      "epoch": 0.3739030822921384,
-      "grad_norm": 1.2965167760849,
-      "learning_rate": 6.260869565217392e-06,
-      "loss": 0.1533,
-      "step": 2193
-    },
-    {
-      "epoch": 0.3740735807336761,
-      "grad_norm": 0.9013891220092773,
-      "learning_rate": 6.25916453537937e-06,
-      "loss": 0.1074,
-      "step": 2194
-    },
-    {
-      "epoch": 0.3742440791752138,
-      "grad_norm": 1.2202616930007935,
-      "learning_rate": 6.257459505541348e-06,
-      "loss": 0.1525,
-      "step": 2195
-    },
-    {
-      "epoch": 0.37441457761675145,
-      "grad_norm": 1.3237905502319336,
-      "learning_rate": 6.255754475703325e-06,
-      "loss": 0.1626,
-      "step": 2196
-    },
-    {
-      "epoch": 0.37458507605828917,
-      "grad_norm": 0.794424831867218,
-      "learning_rate": 6.254049445865304e-06,
-      "loss": 0.1528,
-      "step": 2197
-    },
-    {
-      "epoch": 0.3747555744998268,
-      "grad_norm": 0.7876736521720886,
-      "learning_rate": 6.252344416027281e-06,
-      "loss": 0.0942,
-      "step": 2198
-    },
-    {
-      "epoch": 0.37492607294136454,
-      "grad_norm": 1.0070785284042358,
-      "learning_rate": 6.250639386189259e-06,
-      "loss": 0.1803,
-      "step": 2199
-    },
-    {
-      "epoch": 0.3750965713829022,
-      "grad_norm": 1.2898237705230713,
-      "learning_rate": 6.248934356351236e-06,
-      "loss": 0.1796,
-      "step": 2200
-    },
-    {
-      "epoch": 0.3750965713829022,
-      "eval_f1_score": 0.2619047619047619,
-      "eval_loss": 0.1529710590839386,
-      "eval_runtime": 182.6841,
-      "eval_samples_per_second": 54.739,
-      "eval_steps_per_second": 3.421,
-      "step": 2200
-    },
-    {
-      "epoch": 0.37526706982443986,
-      "grad_norm": 1.1882941722869873,
-      "learning_rate": 6.247229326513215e-06,
-      "loss": 0.1849,
-      "step": 2201
-    },
-    {
-      "epoch": 0.3754375682659776,
-      "grad_norm": 0.7369084358215332,
-      "learning_rate": 6.245524296675193e-06,
-      "loss": 0.1282,
-      "step": 2202
-    },
-    {
-      "epoch": 0.37560806670751523,
-      "grad_norm": 0.88145911693573,
-      "learning_rate": 6.24381926683717e-06,
-      "loss": 0.0625,
-      "step": 2203
-    },
-    {
-      "epoch": 0.37577856514905295,
-      "grad_norm": 0.8131815195083618,
-      "learning_rate": 6.2421142369991485e-06,
-      "loss": 0.1424,
-      "step": 2204
-    },
-    {
-      "epoch": 0.3759490635905906,
-      "grad_norm": 0.9672574996948242,
-      "learning_rate": 6.240409207161126e-06,
-      "loss": 0.1381,
-      "step": 2205
-    },
-    {
-      "epoch": 0.3761195620321283,
-      "grad_norm": 0.8028636574745178,
-      "learning_rate": 6.2387041773231036e-06,
-      "loss": 0.1439,
-      "step": 2206
-    },
-    {
-      "epoch": 0.376290060473666,
-      "grad_norm": 1.0327125787734985,
-      "learning_rate": 6.236999147485081e-06,
-      "loss": 0.154,
-      "step": 2207
-    },
-    {
-      "epoch": 0.3764605589152037,
-      "grad_norm": 0.8577430248260498,
-      "learning_rate": 6.2352941176470595e-06,
-      "loss": 0.1516,
-      "step": 2208
-    },
-    {
-      "epoch": 0.37663105735674135,
-      "grad_norm": 0.625697135925293,
-      "learning_rate": 6.233589087809037e-06,
-      "loss": 0.1135,
-      "step": 2209
-    },
-    {
-      "epoch": 0.376801555798279,
-      "grad_norm": 0.8129452466964722,
-      "learning_rate": 6.2318840579710145e-06,
-      "loss": 0.1557,
-      "step": 2210
-    },
-    {
-      "epoch": 0.3769720542398167,
-      "grad_norm": 0.8950178623199463,
-      "learning_rate": 6.230179028132993e-06,
-      "loss": 0.1792,
-      "step": 2211
-    },
-    {
-      "epoch": 0.3771425526813544,
-      "grad_norm": 0.8850390911102295,
-      "learning_rate": 6.22847399829497e-06,
-      "loss": 0.161,
-      "step": 2212
-    },
-    {
-      "epoch": 0.3773130511228921,
-      "grad_norm": 0.9739559888839722,
-      "learning_rate": 6.226768968456949e-06,
-      "loss": 0.1295,
-      "step": 2213
-    },
-    {
-      "epoch": 0.37748354956442975,
-      "grad_norm": 1.1448336839675903,
-      "learning_rate": 6.225063938618926e-06,
-      "loss": 0.1796,
-      "step": 2214
-    },
-    {
-      "epoch": 0.37765404800596747,
-      "grad_norm": 0.9003750085830688,
-      "learning_rate": 6.223358908780904e-06,
-      "loss": 0.1529,
-      "step": 2215
-    },
-    {
-      "epoch": 0.3778245464475051,
-      "grad_norm": 0.7714499235153198,
-      "learning_rate": 6.221653878942881e-06,
-      "loss": 0.1255,
-      "step": 2216
-    },
-    {
-      "epoch": 0.3779950448890428,
-      "grad_norm": 0.6482207179069519,
-      "learning_rate": 6.21994884910486e-06,
-      "loss": 0.1168,
-      "step": 2217
-    },
-    {
-      "epoch": 0.3781655433305805,
-      "grad_norm": 0.67027747631073,
-      "learning_rate": 6.218243819266838e-06,
-      "loss": 0.096,
-      "step": 2218
-    },
-    {
-      "epoch": 0.37833604177211816,
-      "grad_norm": 0.8746976256370544,
-      "learning_rate": 6.216538789428815e-06,
-      "loss": 0.1316,
-      "step": 2219
-    },
-    {
-      "epoch": 0.3785065402136559,
-      "grad_norm": 0.8811997175216675,
-      "learning_rate": 6.214833759590794e-06,
-      "loss": 0.1429,
-      "step": 2220
-    },
-    {
-      "epoch": 0.37867703865519353,
-      "grad_norm": 0.872652530670166,
-      "learning_rate": 6.213128729752771e-06,
-      "loss": 0.0681,
-      "step": 2221
-    },
-    {
-      "epoch": 0.37884753709673125,
-      "grad_norm": 0.7932902574539185,
-      "learning_rate": 6.211423699914749e-06,
-      "loss": 0.1206,
-      "step": 2222
-    },
-    {
-      "epoch": 0.3790180355382689,
-      "grad_norm": 1.5943965911865234,
-      "learning_rate": 6.209718670076726e-06,
-      "loss": 0.1709,
-      "step": 2223
-    },
-    {
-      "epoch": 0.37918853397980656,
-      "grad_norm": 0.9252127408981323,
-      "learning_rate": 6.208013640238705e-06,
-      "loss": 0.1187,
-      "step": 2224
-    },
-    {
-      "epoch": 0.3793590324213443,
-      "grad_norm": 0.6615201830863953,
-      "learning_rate": 6.206308610400682e-06,
-      "loss": 0.1144,
-      "step": 2225
-    },
-    {
-      "epoch": 0.37952953086288194,
-      "grad_norm": 0.9420225620269775,
-      "learning_rate": 6.20460358056266e-06,
-      "loss": 0.1841,
-      "step": 2226
-    },
-    {
-      "epoch": 0.37970002930441965,
-      "grad_norm": 0.7444522380828857,
-      "learning_rate": 6.202898550724639e-06,
-      "loss": 0.1269,
-      "step": 2227
-    },
-    {
-      "epoch": 0.3798705277459573,
-      "grad_norm": 1.0659302473068237,
-      "learning_rate": 6.201193520886616e-06,
-      "loss": 0.1517,
-      "step": 2228
-    },
-    {
-      "epoch": 0.380041026187495,
-      "grad_norm": 0.8699449300765991,
-      "learning_rate": 6.199488491048594e-06,
-      "loss": 0.1347,
-      "step": 2229
-    },
-    {
-      "epoch": 0.3802115246290327,
-      "grad_norm": 0.9697796106338501,
-      "learning_rate": 6.197783461210572e-06,
-      "loss": 0.1445,
-      "step": 2230
-    },
-    {
-      "epoch": 0.3803820230705704,
-      "grad_norm": 1.4614863395690918,
-      "learning_rate": 6.19607843137255e-06,
-      "loss": 0.2126,
-      "step": 2231
-    },
-    {
-      "epoch": 0.38055252151210806,
-      "grad_norm": 0.6998443007469177,
-      "learning_rate": 6.194373401534527e-06,
-      "loss": 0.106,
-      "step": 2232
-    },
-    {
-      "epoch": 0.3807230199536457,
-      "grad_norm": 1.6563403606414795,
-      "learning_rate": 6.192668371696506e-06,
-      "loss": 0.2081,
-      "step": 2233
-    },
-    {
-      "epoch": 0.38089351839518343,
-      "grad_norm": 0.9660869240760803,
-      "learning_rate": 6.1909633418584836e-06,
-      "loss": 0.2294,
-      "step": 2234
-    },
-    {
-      "epoch": 0.3810640168367211,
-      "grad_norm": 1.1604372262954712,
-      "learning_rate": 6.189258312020461e-06,
-      "loss": 0.093,
-      "step": 2235
-    },
-    {
-      "epoch": 0.3812345152782588,
-      "grad_norm": 1.12831449508667,
-      "learning_rate": 6.1875532821824395e-06,
-      "loss": 0.1999,
-      "step": 2236
-    },
-    {
-      "epoch": 0.38140501371979646,
-      "grad_norm": 0.8325589299201965,
-      "learning_rate": 6.1858482523444166e-06,
-      "loss": 0.1504,
-      "step": 2237
-    },
-    {
-      "epoch": 0.3815755121613342,
-      "grad_norm": 1.3243125677108765,
-      "learning_rate": 6.1841432225063945e-06,
-      "loss": 0.1809,
-      "step": 2238
-    },
-    {
-      "epoch": 0.38174601060287183,
-      "grad_norm": 0.9601839184761047,
-      "learning_rate": 6.182438192668372e-06,
-      "loss": 0.125,
-      "step": 2239
-    },
-    {
-      "epoch": 0.3819165090444095,
-      "grad_norm": 1.0139927864074707,
-      "learning_rate": 6.18073316283035e-06,
-      "loss": 0.128,
-      "step": 2240
-    },
-    {
-      "epoch": 0.3820870074859472,
-      "grad_norm": 0.9789314866065979,
-      "learning_rate": 6.1790281329923275e-06,
-      "loss": 0.1124,
-      "step": 2241
-    },
-    {
-      "epoch": 0.38225750592748486,
-      "grad_norm": 0.8316740989685059,
-      "learning_rate": 6.1773231031543054e-06,
-      "loss": 0.1605,
-      "step": 2242
-    },
-    {
-      "epoch": 0.3824280043690226,
-      "grad_norm": 0.8386850357055664,
-      "learning_rate": 6.175618073316284e-06,
-      "loss": 0.1018,
-      "step": 2243
-    },
-    {
-      "epoch": 0.38259850281056024,
-      "grad_norm": 0.849038302898407,
-      "learning_rate": 6.173913043478261e-06,
-      "loss": 0.1214,
-      "step": 2244
-    },
-    {
-      "epoch": 0.38276900125209795,
-      "grad_norm": 0.8374232053756714,
-      "learning_rate": 6.172208013640239e-06,
-      "loss": 0.1207,
-      "step": 2245
-    },
-    {
-      "epoch": 0.3829394996936356,
-      "grad_norm": 0.8584263324737549,
-      "learning_rate": 6.170502983802216e-06,
-      "loss": 0.101,
-      "step": 2246
-    },
-    {
-      "epoch": 0.38310999813517327,
-      "grad_norm": 0.8575809597969055,
-      "learning_rate": 6.168797953964195e-06,
-      "loss": 0.1521,
-      "step": 2247
-    },
-    {
-      "epoch": 0.383280496576711,
-      "grad_norm": 1.1430909633636475,
-      "learning_rate": 6.167092924126172e-06,
-      "loss": 0.1273,
-      "step": 2248
-    },
-    {
-      "epoch": 0.38345099501824864,
-      "grad_norm": 1.7613104581832886,
-      "learning_rate": 6.16538789428815e-06,
-      "loss": 0.093,
-      "step": 2249
-    },
-    {
-      "epoch": 0.38362149345978636,
-      "grad_norm": 1.461144208908081,
-      "learning_rate": 6.163682864450128e-06,
-      "loss": 0.1602,
-      "step": 2250
-    },
-    {
-      "epoch": 0.383791991901324,
-      "grad_norm": 0.9410567283630371,
-      "learning_rate": 6.161977834612106e-06,
-      "loss": 0.1331,
-      "step": 2251
-    },
-    {
-      "epoch": 0.38396249034286173,
-      "grad_norm": 0.6462012529373169,
-      "learning_rate": 6.160272804774085e-06,
-      "loss": 0.0823,
-      "step": 2252
-    },
-    {
-      "epoch": 0.3841329887843994,
-      "grad_norm": 1.0113216638565063,
-      "learning_rate": 6.158567774936062e-06,
-      "loss": 0.1388,
-      "step": 2253
-    },
-    {
-      "epoch": 0.3843034872259371,
-      "grad_norm": 0.8928611278533936,
-      "learning_rate": 6.15686274509804e-06,
-      "loss": 0.17,
-      "step": 2254
-    },
-    {
-      "epoch": 0.38447398566747476,
-      "grad_norm": 1.2479135990142822,
-      "learning_rate": 6.155157715260017e-06,
-      "loss": 0.1164,
-      "step": 2255
-    },
-    {
-      "epoch": 0.3846444841090124,
-      "grad_norm": 1.9075334072113037,
-      "learning_rate": 6.153452685421996e-06,
-      "loss": 0.1827,
-      "step": 2256
-    },
-    {
-      "epoch": 0.38481498255055013,
-      "grad_norm": 0.836900532245636,
-      "learning_rate": 6.151747655583973e-06,
-      "loss": 0.0707,
-      "step": 2257
-    },
-    {
-      "epoch": 0.3849854809920878,
-      "grad_norm": 0.961868941783905,
-      "learning_rate": 6.150042625745951e-06,
-      "loss": 0.0876,
-      "step": 2258
-    },
-    {
-      "epoch": 0.3851559794336255,
-      "grad_norm": 1.439834713935852,
-      "learning_rate": 6.14833759590793e-06,
-      "loss": 0.177,
-      "step": 2259
-    },
-    {
-      "epoch": 0.38532647787516316,
-      "grad_norm": 0.7403258681297302,
-      "learning_rate": 6.146632566069907e-06,
-      "loss": 0.0915,
-      "step": 2260
-    },
-    {
-      "epoch": 0.3854969763167009,
-      "grad_norm": 0.9234430193901062,
-      "learning_rate": 6.144927536231885e-06,
-      "loss": 0.129,
-      "step": 2261
-    },
-    {
-      "epoch": 0.38566747475823854,
-      "grad_norm": 1.1485732793807983,
-      "learning_rate": 6.143222506393862e-06,
-      "loss": 0.1597,
-      "step": 2262
-    },
-    {
-      "epoch": 0.3858379731997762,
-      "grad_norm": 1.0878947973251343,
-      "learning_rate": 6.141517476555841e-06,
-      "loss": 0.1158,
-      "step": 2263
-    },
-    {
-      "epoch": 0.3860084716413139,
-      "grad_norm": 1.095202088356018,
-      "learning_rate": 6.139812446717818e-06,
-      "loss": 0.1728,
-      "step": 2264
-    },
-    {
-      "epoch": 0.38617897008285157,
-      "grad_norm": 0.9918746948242188,
-      "learning_rate": 6.138107416879796e-06,
-      "loss": 0.0813,
-      "step": 2265
-    },
-    {
-      "epoch": 0.3863494685243893,
-      "grad_norm": 1.457942247390747,
-      "learning_rate": 6.136402387041774e-06,
-      "loss": 0.1729,
-      "step": 2266
-    },
-    {
-      "epoch": 0.38651996696592694,
-      "grad_norm": 0.9414834976196289,
-      "learning_rate": 6.134697357203752e-06,
-      "loss": 0.0493,
-      "step": 2267
-    },
-    {
-      "epoch": 0.38669046540746466,
-      "grad_norm": 0.8592334389686584,
-      "learning_rate": 6.1329923273657295e-06,
-      "loss": 0.115,
-      "step": 2268
-    },
-    {
-      "epoch": 0.3868609638490023,
-      "grad_norm": 0.8972296118736267,
-      "learning_rate": 6.1312872975277075e-06,
-      "loss": 0.1338,
-      "step": 2269
-    },
-    {
-      "epoch": 0.38703146229054,
-      "grad_norm": 0.7704724073410034,
-      "learning_rate": 6.1295822676896854e-06,
-      "loss": 0.0741,
-      "step": 2270
-    },
-    {
-      "epoch": 0.3872019607320777,
-      "grad_norm": 0.8177960515022278,
-      "learning_rate": 6.1278772378516625e-06,
-      "loss": 0.0844,
-      "step": 2271
-    },
-    {
-      "epoch": 0.38737245917361535,
-      "grad_norm": 0.719103217124939,
-      "learning_rate": 6.126172208013641e-06,
-      "loss": 0.089,
-      "step": 2272
-    },
-    {
-      "epoch": 0.38754295761515306,
-      "grad_norm": 1.591321349143982,
-      "learning_rate": 6.1244671781756184e-06,
-      "loss": 0.2422,
-      "step": 2273
-    },
-    {
-      "epoch": 0.3877134560566907,
-      "grad_norm": 0.5507732629776001,
-      "learning_rate": 6.122762148337596e-06,
-      "loss": 0.0382,
-      "step": 2274
-    },
-    {
-      "epoch": 0.38788395449822843,
-      "grad_norm": 0.6792088747024536,
-      "learning_rate": 6.1210571184995735e-06,
-      "loss": 0.1013,
-      "step": 2275
-    },
-    {
-      "epoch": 0.3880544529397661,
-      "grad_norm": 0.6091058850288391,
-      "learning_rate": 6.119352088661552e-06,
-      "loss": 0.0477,
-      "step": 2276
-    },
-    {
-      "epoch": 0.3882249513813038,
-      "grad_norm": 1.6229771375656128,
-      "learning_rate": 6.11764705882353e-06,
-      "loss": 0.1975,
-      "step": 2277
-    },
-    {
-      "epoch": 0.38839544982284147,
-      "grad_norm": 1.1221989393234253,
-      "learning_rate": 6.115942028985507e-06,
-      "loss": 0.1165,
-      "step": 2278
-    },
-    {
-      "epoch": 0.3885659482643791,
-      "grad_norm": 1.2790297269821167,
-      "learning_rate": 6.114236999147486e-06,
-      "loss": 0.1424,
-      "step": 2279
-    },
-    {
-      "epoch": 0.38873644670591684,
-      "grad_norm": 0.8660223484039307,
-      "learning_rate": 6.112531969309463e-06,
-      "loss": 0.1013,
-      "step": 2280
-    },
-    {
-      "epoch": 0.3889069451474545,
-      "grad_norm": 1.3250168561935425,
-      "learning_rate": 6.110826939471441e-06,
-      "loss": 0.1911,
-      "step": 2281
-    },
-    {
-      "epoch": 0.3890774435889922,
-      "grad_norm": 0.8523133397102356,
-      "learning_rate": 6.109121909633418e-06,
-      "loss": 0.1004,
-      "step": 2282
-    },
-    {
-      "epoch": 0.38924794203052987,
-      "grad_norm": 1.2505394220352173,
-      "learning_rate": 6.107416879795397e-06,
-      "loss": 0.1516,
-      "step": 2283
-    },
-    {
-      "epoch": 0.3894184404720676,
-      "grad_norm": 1.5190775394439697,
-      "learning_rate": 6.105711849957374e-06,
-      "loss": 0.1719,
-      "step": 2284
-    },
-    {
-      "epoch": 0.38958893891360524,
-      "grad_norm": 0.8516865968704224,
-      "learning_rate": 6.104006820119352e-06,
-      "loss": 0.0865,
-      "step": 2285
-    },
-    {
-      "epoch": 0.3897594373551429,
-      "grad_norm": 0.7336435317993164,
-      "learning_rate": 6.102301790281331e-06,
-      "loss": 0.0977,
-      "step": 2286
-    },
-    {
-      "epoch": 0.3899299357966806,
-      "grad_norm": 1.225153923034668,
-      "learning_rate": 6.100596760443308e-06,
-      "loss": 0.1039,
-      "step": 2287
-    },
-    {
-      "epoch": 0.3901004342382183,
-      "grad_norm": 0.9977211952209473,
-      "learning_rate": 6.098891730605287e-06,
-      "loss": 0.1931,
-      "step": 2288
-    },
-    {
-      "epoch": 0.390270932679756,
-      "grad_norm": 0.9641508460044861,
-      "learning_rate": 6.097186700767264e-06,
-      "loss": 0.1688,
-      "step": 2289
-    },
-    {
-      "epoch": 0.39044143112129365,
-      "grad_norm": 1.0043286085128784,
-      "learning_rate": 6.095481670929242e-06,
-      "loss": 0.1296,
-      "step": 2290
-    },
-    {
-      "epoch": 0.39061192956283136,
-      "grad_norm": 0.9000137448310852,
-      "learning_rate": 6.093776641091219e-06,
-      "loss": 0.1551,
-      "step": 2291
-    },
-    {
-      "epoch": 0.390782428004369,
-      "grad_norm": 1.0267075300216675,
-      "learning_rate": 6.092071611253198e-06,
-      "loss": 0.1337,
-      "step": 2292
-    },
-    {
-      "epoch": 0.3909529264459067,
-      "grad_norm": 1.451011061668396,
-      "learning_rate": 6.090366581415176e-06,
-      "loss": 0.218,
-      "step": 2293
-    },
-    {
-      "epoch": 0.3911234248874444,
-      "grad_norm": 1.2185776233673096,
-      "learning_rate": 6.088661551577153e-06,
-      "loss": 0.2079,
-      "step": 2294
-    },
-    {
-      "epoch": 0.39129392332898205,
-      "grad_norm": 1.1939146518707275,
-      "learning_rate": 6.086956521739132e-06,
-      "loss": 0.1223,
-      "step": 2295
-    },
-    {
-      "epoch": 0.39146442177051977,
-      "grad_norm": 0.7564530372619629,
-      "learning_rate": 6.085251491901109e-06,
-      "loss": 0.1188,
-      "step": 2296
-    },
-    {
-      "epoch": 0.3916349202120574,
-      "grad_norm": 1.1471365690231323,
-      "learning_rate": 6.083546462063087e-06,
-      "loss": 0.1619,
-      "step": 2297
-    },
-    {
-      "epoch": 0.39180541865359514,
-      "grad_norm": 1.486033320426941,
-      "learning_rate": 6.081841432225064e-06,
-      "loss": 0.1774,
-      "step": 2298
-    },
-    {
-      "epoch": 0.3919759170951328,
-      "grad_norm": 0.9266932010650635,
-      "learning_rate": 6.0801364023870425e-06,
-      "loss": 0.1548,
-      "step": 2299
-    },
-    {
-      "epoch": 0.3921464155366705,
-      "grad_norm": 0.9137939214706421,
-      "learning_rate": 6.07843137254902e-06,
-      "loss": 0.1048,
-      "step": 2300
-    },
-    {
-      "epoch": 0.3921464155366705,
-      "eval_f1_score": 0.29896907216494845,
-      "eval_loss": 0.14963874220848083,
-      "eval_runtime": 182.6197,
-      "eval_samples_per_second": 54.759,
-      "eval_steps_per_second": 3.422,
-      "step": 2300
-    },
-    {
-      "epoch": 0.39231691397820817,
-      "grad_norm": 0.8086863160133362,
-      "learning_rate": 6.076726342710998e-06,
-      "loss": 0.117,
-      "step": 2301
-    },
-    {
-      "epoch": 0.39248741241974583,
-      "grad_norm": 0.8732712268829346,
-      "learning_rate": 6.075021312872976e-06,
-      "loss": 0.0763,
-      "step": 2302
-    },
-    {
-      "epoch": 0.39265791086128354,
-      "grad_norm": 0.8038264513015747,
-      "learning_rate": 6.0733162830349535e-06,
-      "loss": 0.1339,
-      "step": 2303
-    },
-    {
-      "epoch": 0.3928284093028212,
-      "grad_norm": 0.7566570043563843,
-      "learning_rate": 6.0716112531969314e-06,
-      "loss": 0.0899,
-      "step": 2304
-    },
-    {
-      "epoch": 0.3929989077443589,
-      "grad_norm": 0.9701877236366272,
-      "learning_rate": 6.069906223358909e-06,
-      "loss": 0.114,
-      "step": 2305
-    },
-    {
-      "epoch": 0.3931694061858966,
-      "grad_norm": 1.1687184572219849,
-      "learning_rate": 6.068201193520887e-06,
-      "loss": 0.1838,
-      "step": 2306
-    },
-    {
-      "epoch": 0.3933399046274343,
-      "grad_norm": 1.289287805557251,
-      "learning_rate": 6.0664961636828644e-06,
-      "loss": 0.1721,
-      "step": 2307
-    },
-    {
-      "epoch": 0.39351040306897195,
-      "grad_norm": 0.985457718372345,
-      "learning_rate": 6.064791133844843e-06,
-      "loss": 0.1179,
-      "step": 2308
-    },
-    {
-      "epoch": 0.3936809015105096,
-      "grad_norm": 1.1207270622253418,
-      "learning_rate": 6.06308610400682e-06,
-      "loss": 0.1381,
-      "step": 2309
-    },
-    {
-      "epoch": 0.3938513999520473,
-      "grad_norm": 1.2061148881912231,
-      "learning_rate": 6.061381074168798e-06,
-      "loss": 0.1639,
-      "step": 2310
-    },
-    {
-      "epoch": 0.394021898393585,
-      "grad_norm": 1.4372804164886475,
-      "learning_rate": 6.059676044330777e-06,
-      "loss": 0.2017,
-      "step": 2311
-    },
-    {
-      "epoch": 0.3941923968351227,
-      "grad_norm": 1.2852275371551514,
-      "learning_rate": 6.057971014492754e-06,
-      "loss": 0.1528,
-      "step": 2312
-    },
-    {
-      "epoch": 0.39436289527666035,
-      "grad_norm": 1.009509801864624,
-      "learning_rate": 6.056265984654732e-06,
-      "loss": 0.1401,
-      "step": 2313
-    },
-    {
-      "epoch": 0.39453339371819807,
-      "grad_norm": 0.8609738349914551,
-      "learning_rate": 6.054560954816709e-06,
-      "loss": 0.1215,
-      "step": 2314
-    },
-    {
-      "epoch": 0.3947038921597357,
-      "grad_norm": 0.8159873485565186,
-      "learning_rate": 6.052855924978688e-06,
-      "loss": 0.117,
-      "step": 2315
-    },
-    {
-      "epoch": 0.3948743906012734,
-      "grad_norm": 0.9817201495170593,
-      "learning_rate": 6.051150895140665e-06,
-      "loss": 0.1891,
-      "step": 2316
-    },
-    {
-      "epoch": 0.3950448890428111,
-      "grad_norm": 0.7218739986419678,
-      "learning_rate": 6.049445865302643e-06,
-      "loss": 0.0861,
-      "step": 2317
-    },
-    {
-      "epoch": 0.39521538748434876,
-      "grad_norm": 0.8500337600708008,
-      "learning_rate": 6.047740835464622e-06,
-      "loss": 0.1335,
-      "step": 2318
-    },
-    {
-      "epoch": 0.39538588592588647,
-      "grad_norm": 0.9201745986938477,
-      "learning_rate": 6.046035805626599e-06,
-      "loss": 0.1352,
-      "step": 2319
-    },
-    {
-      "epoch": 0.39555638436742413,
-      "grad_norm": 1.1116868257522583,
-      "learning_rate": 6.044330775788577e-06,
-      "loss": 0.1093,
-      "step": 2320
-    },
-    {
-      "epoch": 0.39572688280896184,
-      "grad_norm": 1.2167072296142578,
-      "learning_rate": 6.042625745950554e-06,
-      "loss": 0.072,
-      "step": 2321
-    },
-    {
-      "epoch": 0.3958973812504995,
-      "grad_norm": 0.785332202911377,
-      "learning_rate": 6.040920716112533e-06,
-      "loss": 0.1055,
-      "step": 2322
-    },
-    {
-      "epoch": 0.3960678796920372,
-      "grad_norm": 0.8508793115615845,
-      "learning_rate": 6.03921568627451e-06,
-      "loss": 0.139,
-      "step": 2323
-    },
-    {
-      "epoch": 0.3962383781335749,
-      "grad_norm": 1.0231958627700806,
-      "learning_rate": 6.037510656436488e-06,
-      "loss": 0.1833,
-      "step": 2324
-    },
-    {
-      "epoch": 0.39640887657511253,
-      "grad_norm": 0.5762442350387573,
-      "learning_rate": 6.035805626598466e-06,
-      "loss": 0.0719,
-      "step": 2325
-    },
-    {
-      "epoch": 0.39657937501665025,
-      "grad_norm": 1.129563808441162,
-      "learning_rate": 6.034100596760444e-06,
-      "loss": 0.1799,
-      "step": 2326
-    },
-    {
-      "epoch": 0.3967498734581879,
-      "grad_norm": 1.4465104341506958,
-      "learning_rate": 6.0323955669224225e-06,
-      "loss": 0.1404,
-      "step": 2327
-    },
-    {
-      "epoch": 0.3969203718997256,
-      "grad_norm": 1.0937806367874146,
-      "learning_rate": 6.0306905370844e-06,
-      "loss": 0.1169,
-      "step": 2328
-    },
-    {
-      "epoch": 0.3970908703412633,
-      "grad_norm": 1.3986912965774536,
-      "learning_rate": 6.028985507246378e-06,
-      "loss": 0.1871,
-      "step": 2329
-    },
-    {
-      "epoch": 0.397261368782801,
-      "grad_norm": 1.3107599020004272,
-      "learning_rate": 6.027280477408355e-06,
-      "loss": 0.224,
-      "step": 2330
-    },
-    {
-      "epoch": 0.39743186722433865,
-      "grad_norm": 1.5524184703826904,
-      "learning_rate": 6.0255754475703335e-06,
-      "loss": 0.242,
-      "step": 2331
-    },
-    {
-      "epoch": 0.3976023656658763,
-      "grad_norm": 0.8739752173423767,
-      "learning_rate": 6.023870417732311e-06,
-      "loss": 0.1174,
-      "step": 2332
-    },
-    {
-      "epoch": 0.397772864107414,
-      "grad_norm": 0.711715042591095,
-      "learning_rate": 6.0221653878942885e-06,
-      "loss": 0.0916,
-      "step": 2333
-    },
-    {
-      "epoch": 0.3979433625489517,
-      "grad_norm": 0.7139453291893005,
-      "learning_rate": 6.020460358056266e-06,
-      "loss": 0.0759,
-      "step": 2334
-    },
-    {
-      "epoch": 0.3981138609904894,
-      "grad_norm": 0.8483171463012695,
-      "learning_rate": 6.018755328218244e-06,
-      "loss": 0.136,
-      "step": 2335
-    },
-    {
-      "epoch": 0.39828435943202706,
-      "grad_norm": 0.8804246187210083,
-      "learning_rate": 6.017050298380222e-06,
-      "loss": 0.1069,
-      "step": 2336
-    },
-    {
-      "epoch": 0.39845485787356477,
-      "grad_norm": 1.0845495462417603,
-      "learning_rate": 6.0153452685421995e-06,
-      "loss": 0.148,
-      "step": 2337
-    },
-    {
-      "epoch": 0.39862535631510243,
-      "grad_norm": 0.8838271498680115,
-      "learning_rate": 6.013640238704178e-06,
-      "loss": 0.0673,
-      "step": 2338
-    },
-    {
-      "epoch": 0.3987958547566401,
-      "grad_norm": 0.9725067019462585,
-      "learning_rate": 6.011935208866155e-06,
-      "loss": 0.177,
-      "step": 2339
-    },
-    {
-      "epoch": 0.3989663531981778,
-      "grad_norm": 1.3371416330337524,
-      "learning_rate": 6.010230179028133e-06,
-      "loss": 0.2014,
-      "step": 2340
-    },
-    {
-      "epoch": 0.39913685163971546,
-      "grad_norm": 0.7962145209312439,
-      "learning_rate": 6.008525149190111e-06,
-      "loss": 0.1034,
-      "step": 2341
-    },
-    {
-      "epoch": 0.3993073500812532,
-      "grad_norm": 0.7338995933532715,
-      "learning_rate": 6.006820119352089e-06,
-      "loss": 0.1091,
-      "step": 2342
-    },
-    {
-      "epoch": 0.39947784852279083,
-      "grad_norm": 0.82045578956604,
-      "learning_rate": 6.005115089514067e-06,
-      "loss": 0.1205,
-      "step": 2343
-    },
-    {
-      "epoch": 0.39964834696432855,
-      "grad_norm": 0.7830996513366699,
-      "learning_rate": 6.003410059676045e-06,
-      "loss": 0.1117,
-      "step": 2344
-    },
-    {
-      "epoch": 0.3998188454058662,
-      "grad_norm": 1.263693928718567,
-      "learning_rate": 6.001705029838023e-06,
-      "loss": 0.243,
-      "step": 2345
-    },
-    {
-      "epoch": 0.3999893438474039,
-      "grad_norm": 0.866906464099884,
-      "learning_rate": 6e-06,
-      "loss": 0.1502,
-      "step": 2346
-    },
-    {
-      "epoch": 0.4001598422889416,
-      "grad_norm": 0.856949508190155,
-      "learning_rate": 5.998294970161979e-06,
-      "loss": 0.149,
-      "step": 2347
-    },
-    {
-      "epoch": 0.40033034073047924,
-      "grad_norm": 0.8175544738769531,
-      "learning_rate": 5.996589940323956e-06,
-      "loss": 0.1468,
-      "step": 2348
-    },
-    {
-      "epoch": 0.40050083917201695,
-      "grad_norm": 0.986709713935852,
-      "learning_rate": 5.994884910485934e-06,
-      "loss": 0.144,
-      "step": 2349
-    },
-    {
-      "epoch": 0.4006713376135546,
-      "grad_norm": 1.01996910572052,
-      "learning_rate": 5.993179880647911e-06,
-      "loss": 0.1834,
-      "step": 2350
-    },
-    {
-      "epoch": 0.4008418360550923,
-      "grad_norm": 1.0860642194747925,
-      "learning_rate": 5.99147485080989e-06,
-      "loss": 0.1928,
-      "step": 2351
-    },
-    {
-      "epoch": 0.40101233449663,
-      "grad_norm": 0.6442664861679077,
-      "learning_rate": 5.989769820971868e-06,
-      "loss": 0.0916,
-      "step": 2352
-    },
-    {
-      "epoch": 0.4011828329381677,
-      "grad_norm": 1.1461782455444336,
-      "learning_rate": 5.988064791133845e-06,
-      "loss": 0.1605,
-      "step": 2353
-    },
-    {
-      "epoch": 0.40135333137970536,
-      "grad_norm": 0.7978892922401428,
-      "learning_rate": 5.986359761295824e-06,
-      "loss": 0.129,
-      "step": 2354
-    },
-    {
-      "epoch": 0.401523829821243,
-      "grad_norm": 0.7703652381896973,
-      "learning_rate": 5.984654731457801e-06,
-      "loss": 0.1015,
-      "step": 2355
-    },
-    {
-      "epoch": 0.40169432826278073,
-      "grad_norm": 0.939119815826416,
-      "learning_rate": 5.982949701619779e-06,
-      "loss": 0.1502,
-      "step": 2356
-    },
-    {
-      "epoch": 0.4018648267043184,
-      "grad_norm": 0.6256553530693054,
-      "learning_rate": 5.981244671781756e-06,
-      "loss": 0.1149,
-      "step": 2357
-    },
-    {
-      "epoch": 0.4020353251458561,
-      "grad_norm": 0.7845792174339294,
-      "learning_rate": 5.979539641943735e-06,
-      "loss": 0.1002,
-      "step": 2358
-    },
-    {
-      "epoch": 0.40220582358739376,
-      "grad_norm": 1.0356690883636475,
-      "learning_rate": 5.977834612105712e-06,
-      "loss": 0.1763,
-      "step": 2359
-    },
-    {
-      "epoch": 0.4023763220289315,
-      "grad_norm": 0.631354808807373,
-      "learning_rate": 5.97612958226769e-06,
-      "loss": 0.1229,
-      "step": 2360
-    },
-    {
-      "epoch": 0.40254682047046914,
-      "grad_norm": 0.8355875015258789,
-      "learning_rate": 5.9744245524296685e-06,
-      "loss": 0.1467,
-      "step": 2361
-    },
-    {
-      "epoch": 0.4027173189120068,
-      "grad_norm": 0.8429080843925476,
-      "learning_rate": 5.972719522591646e-06,
-      "loss": 0.1641,
-      "step": 2362
-    },
-    {
-      "epoch": 0.4028878173535445,
-      "grad_norm": 1.1086691617965698,
-      "learning_rate": 5.9710144927536236e-06,
-      "loss": 0.2006,
-      "step": 2363
-    },
-    {
-      "epoch": 0.40305831579508217,
-      "grad_norm": 1.2838102579116821,
-      "learning_rate": 5.9693094629156015e-06,
-      "loss": 0.1822,
-      "step": 2364
-    },
-    {
-      "epoch": 0.4032288142366199,
-      "grad_norm": 1.2585407495498657,
-      "learning_rate": 5.9676044330775795e-06,
-      "loss": 0.2134,
-      "step": 2365
-    },
-    {
-      "epoch": 0.40339931267815754,
-      "grad_norm": 1.0826866626739502,
-      "learning_rate": 5.9658994032395566e-06,
-      "loss": 0.1734,
-      "step": 2366
-    },
-    {
-      "epoch": 0.40356981111969525,
-      "grad_norm": 0.88109290599823,
-      "learning_rate": 5.964194373401535e-06,
-      "loss": 0.1703,
-      "step": 2367
-    },
-    {
-      "epoch": 0.4037403095612329,
-      "grad_norm": 0.7502427697181702,
-      "learning_rate": 5.962489343563513e-06,
-      "loss": 0.1001,
-      "step": 2368
-    },
-    {
-      "epoch": 0.4039108080027706,
-      "grad_norm": 1.4130350351333618,
-      "learning_rate": 5.96078431372549e-06,
-      "loss": 0.2545,
-      "step": 2369
-    },
-    {
-      "epoch": 0.4040813064443083,
-      "grad_norm": 0.9190012216567993,
-      "learning_rate": 5.959079283887469e-06,
-      "loss": 0.1548,
-      "step": 2370
-    },
-    {
-      "epoch": 0.40425180488584594,
-      "grad_norm": 0.8156841397285461,
-      "learning_rate": 5.957374254049446e-06,
-      "loss": 0.0857,
-      "step": 2371
-    },
-    {
-      "epoch": 0.40442230332738366,
-      "grad_norm": 1.1620196104049683,
-      "learning_rate": 5.955669224211424e-06,
-      "loss": 0.1501,
-      "step": 2372
-    },
-    {
-      "epoch": 0.4045928017689213,
-      "grad_norm": 0.8737968802452087,
-      "learning_rate": 5.953964194373401e-06,
-      "loss": 0.1385,
-      "step": 2373
-    },
-    {
-      "epoch": 0.40476330021045903,
-      "grad_norm": 0.7628181576728821,
-      "learning_rate": 5.95225916453538e-06,
-      "loss": 0.1445,
-      "step": 2374
-    },
-    {
-      "epoch": 0.4049337986519967,
-      "grad_norm": 0.750722348690033,
-      "learning_rate": 5.950554134697357e-06,
-      "loss": 0.1131,
-      "step": 2375
-    },
-    {
-      "epoch": 0.4051042970935344,
-      "grad_norm": 1.0749493837356567,
-      "learning_rate": 5.948849104859335e-06,
-      "loss": 0.1112,
-      "step": 2376
-    },
-    {
-      "epoch": 0.40527479553507206,
-      "grad_norm": 0.653072714805603,
-      "learning_rate": 5.947144075021314e-06,
-      "loss": 0.094,
-      "step": 2377
-    },
-    {
-      "epoch": 0.4054452939766097,
-      "grad_norm": 0.8208218812942505,
-      "learning_rate": 5.945439045183291e-06,
-      "loss": 0.1553,
-      "step": 2378
-    },
-    {
-      "epoch": 0.40561579241814744,
-      "grad_norm": 0.857717752456665,
-      "learning_rate": 5.943734015345269e-06,
-      "loss": 0.1312,
-      "step": 2379
-    },
-    {
-      "epoch": 0.4057862908596851,
-      "grad_norm": 0.6189414858818054,
-      "learning_rate": 5.942028985507247e-06,
-      "loss": 0.1063,
-      "step": 2380
-    },
-    {
-      "epoch": 0.4059567893012228,
-      "grad_norm": 1.1543437242507935,
-      "learning_rate": 5.940323955669225e-06,
-      "loss": 0.194,
-      "step": 2381
-    },
-    {
-      "epoch": 0.40612728774276047,
-      "grad_norm": 1.019155502319336,
-      "learning_rate": 5.938618925831202e-06,
-      "loss": 0.1542,
-      "step": 2382
-    },
-    {
-      "epoch": 0.4062977861842982,
-      "grad_norm": 0.5917924642562866,
-      "learning_rate": 5.936913895993181e-06,
-      "loss": 0.0949,
-      "step": 2383
-    },
-    {
-      "epoch": 0.40646828462583584,
-      "grad_norm": 1.1552553176879883,
-      "learning_rate": 5.935208866155158e-06,
-      "loss": 0.132,
-      "step": 2384
-    },
-    {
-      "epoch": 0.4066387830673735,
-      "grad_norm": 1.129095196723938,
-      "learning_rate": 5.933503836317136e-06,
-      "loss": 0.1757,
-      "step": 2385
-    },
-    {
-      "epoch": 0.4068092815089112,
-      "grad_norm": 0.8844802975654602,
-      "learning_rate": 5.931798806479115e-06,
-      "loss": 0.1515,
-      "step": 2386
-    },
-    {
-      "epoch": 0.40697977995044887,
-      "grad_norm": 0.7525590062141418,
-      "learning_rate": 5.930093776641092e-06,
-      "loss": 0.121,
-      "step": 2387
-    },
-    {
-      "epoch": 0.4071502783919866,
-      "grad_norm": 0.6554398536682129,
-      "learning_rate": 5.92838874680307e-06,
-      "loss": 0.0379,
-      "step": 2388
-    },
-    {
-      "epoch": 0.40732077683352425,
-      "grad_norm": 0.8680796027183533,
-      "learning_rate": 5.926683716965047e-06,
-      "loss": 0.1016,
-      "step": 2389
-    },
-    {
-      "epoch": 0.40749127527506196,
-      "grad_norm": 0.7904180288314819,
-      "learning_rate": 5.924978687127026e-06,
-      "loss": 0.0739,
-      "step": 2390
-    },
-    {
-      "epoch": 0.4076617737165996,
-      "grad_norm": 0.5275554060935974,
-      "learning_rate": 5.923273657289003e-06,
-      "loss": 0.0848,
-      "step": 2391
-    },
-    {
-      "epoch": 0.40783227215813733,
-      "grad_norm": 0.6864534020423889,
-      "learning_rate": 5.921568627450981e-06,
-      "loss": 0.0928,
-      "step": 2392
-    },
-    {
-      "epoch": 0.408002770599675,
-      "grad_norm": 0.8688926696777344,
-      "learning_rate": 5.9198635976129595e-06,
-      "loss": 0.1337,
-      "step": 2393
-    },
-    {
-      "epoch": 0.40817326904121265,
-      "grad_norm": 0.8549246191978455,
-      "learning_rate": 5.9181585677749366e-06,
-      "loss": 0.1339,
-      "step": 2394
-    },
-    {
-      "epoch": 0.40834376748275036,
-      "grad_norm": 1.4684349298477173,
-      "learning_rate": 5.9164535379369145e-06,
-      "loss": 0.2219,
-      "step": 2395
-    },
-    {
-      "epoch": 0.408514265924288,
-      "grad_norm": 1.7323837280273438,
-      "learning_rate": 5.914748508098892e-06,
-      "loss": 0.324,
-      "step": 2396
-    },
-    {
-      "epoch": 0.40868476436582574,
-      "grad_norm": 0.6567860245704651,
-      "learning_rate": 5.91304347826087e-06,
-      "loss": 0.0704,
-      "step": 2397
-    },
-    {
-      "epoch": 0.4088552628073634,
-      "grad_norm": 0.7284254431724548,
-      "learning_rate": 5.9113384484228475e-06,
-      "loss": 0.1179,
-      "step": 2398
-    },
-    {
-      "epoch": 0.4090257612489011,
-      "grad_norm": 1.1074248552322388,
-      "learning_rate": 5.9096334185848255e-06,
-      "loss": 0.1273,
-      "step": 2399
-    },
-    {
-      "epoch": 0.40919625969043877,
-      "grad_norm": 0.9865325093269348,
-      "learning_rate": 5.907928388746803e-06,
-      "loss": 0.1429,
-      "step": 2400
-    },
-    {
-      "epoch": 0.40919625969043877,
-      "eval_f1_score": 0.31876606683804626,
-      "eval_loss": 0.14625637233257294,
-      "eval_runtime": 182.619,
-      "eval_samples_per_second": 54.759,
-      "eval_steps_per_second": 3.422,
-      "step": 2400
-    },
-    {
-      "epoch": 0.4093667581319764,
-      "grad_norm": 0.9457055926322937,
-      "learning_rate": 5.906223358908781e-06,
-      "loss": 0.1687,
-      "step": 2401
-    },
-    {
-      "epoch": 0.40953725657351414,
-      "grad_norm": 1.100854754447937,
-      "learning_rate": 5.90451832907076e-06,
-      "loss": 0.1754,
-      "step": 2402
-    },
-    {
-      "epoch": 0.4097077550150518,
-      "grad_norm": 0.8571153879165649,
-      "learning_rate": 5.902813299232737e-06,
-      "loss": 0.1214,
-      "step": 2403
-    },
-    {
-      "epoch": 0.4098782534565895,
-      "grad_norm": 0.955055832862854,
-      "learning_rate": 5.901108269394715e-06,
-      "loss": 0.1678,
-      "step": 2404
-    },
-    {
-      "epoch": 0.4100487518981272,
-      "grad_norm": 0.8582931160926819,
-      "learning_rate": 5.899403239556692e-06,
-      "loss": 0.1298,
-      "step": 2405
-    },
-    {
-      "epoch": 0.4102192503396649,
-      "grad_norm": 1.1340924501419067,
-      "learning_rate": 5.897698209718671e-06,
-      "loss": 0.1669,
-      "step": 2406
-    },
-    {
-      "epoch": 0.41038974878120255,
-      "grad_norm": 1.0416170358657837,
-      "learning_rate": 5.895993179880648e-06,
-      "loss": 0.1432,
-      "step": 2407
-    },
-    {
-      "epoch": 0.4105602472227402,
-      "grad_norm": 1.0653588771820068,
-      "learning_rate": 5.894288150042626e-06,
-      "loss": 0.1826,
-      "step": 2408
-    },
-    {
-      "epoch": 0.4107307456642779,
-      "grad_norm": 1.1291371583938599,
-      "learning_rate": 5.892583120204603e-06,
-      "loss": 0.2454,
-      "step": 2409
-    },
-    {
-      "epoch": 0.4109012441058156,
-      "grad_norm": 1.0526398420333862,
-      "learning_rate": 5.890878090366582e-06,
-      "loss": 0.1268,
-      "step": 2410
-    },
-    {
-      "epoch": 0.4110717425473533,
-      "grad_norm": 1.0905719995498657,
-      "learning_rate": 5.88917306052856e-06,
-      "loss": 0.174,
-      "step": 2411
-    },
-    {
-      "epoch": 0.41124224098889095,
-      "grad_norm": 0.8474220037460327,
-      "learning_rate": 5.887468030690537e-06,
-      "loss": 0.1421,
-      "step": 2412
-    },
-    {
-      "epoch": 0.41141273943042866,
-      "grad_norm": 1.1635980606079102,
-      "learning_rate": 5.885763000852516e-06,
-      "loss": 0.1261,
-      "step": 2413
-    },
-    {
-      "epoch": 0.4115832378719663,
-      "grad_norm": 0.8044309616088867,
-      "learning_rate": 5.884057971014493e-06,
-      "loss": 0.1569,
-      "step": 2414
-    },
-    {
-      "epoch": 0.411753736313504,
-      "grad_norm": 0.7889132499694824,
-      "learning_rate": 5.882352941176471e-06,
-      "loss": 0.1622,
-      "step": 2415
-    },
-    {
-      "epoch": 0.4119242347550417,
-      "grad_norm": 0.9566287398338318,
-      "learning_rate": 5.880647911338449e-06,
-      "loss": 0.0778,
-      "step": 2416
-    },
-    {
-      "epoch": 0.41209473319657935,
-      "grad_norm": 0.743043839931488,
-      "learning_rate": 5.878942881500427e-06,
-      "loss": 0.0944,
-      "step": 2417
-    },
-    {
-      "epoch": 0.41226523163811707,
-      "grad_norm": 1.1795198917388916,
-      "learning_rate": 5.877237851662405e-06,
-      "loss": 0.213,
-      "step": 2418
-    },
-    {
-      "epoch": 0.4124357300796547,
-      "grad_norm": 0.9205105304718018,
-      "learning_rate": 5.875532821824383e-06,
-      "loss": 0.0509,
-      "step": 2419
-    },
-    {
-      "epoch": 0.41260622852119244,
-      "grad_norm": 0.9393637180328369,
-      "learning_rate": 5.873827791986361e-06,
-      "loss": 0.1202,
-      "step": 2420
-    },
-    {
-      "epoch": 0.4127767269627301,
-      "grad_norm": 0.8949941396713257,
-      "learning_rate": 5.872122762148338e-06,
-      "loss": 0.1036,
-      "step": 2421
-    },
-    {
-      "epoch": 0.4129472254042678,
-      "grad_norm": 0.9795252680778503,
-      "learning_rate": 5.8704177323103166e-06,
-      "loss": 0.1586,
-      "step": 2422
-    },
-    {
-      "epoch": 0.4131177238458055,
-      "grad_norm": 0.8107895255088806,
-      "learning_rate": 5.868712702472294e-06,
-      "loss": 0.1404,
-      "step": 2423
-    },
-    {
-      "epoch": 0.41328822228734313,
-      "grad_norm": 0.9952834248542786,
-      "learning_rate": 5.867007672634272e-06,
-      "loss": 0.1253,
-      "step": 2424
-    },
-    {
-      "epoch": 0.41345872072888085,
-      "grad_norm": 1.0126919746398926,
-      "learning_rate": 5.865302642796249e-06,
-      "loss": 0.1819,
-      "step": 2425
-    },
-    {
-      "epoch": 0.4136292191704185,
-      "grad_norm": 1.6776386499404907,
-      "learning_rate": 5.8635976129582275e-06,
-      "loss": 0.1804,
-      "step": 2426
-    },
-    {
-      "epoch": 0.4137997176119562,
-      "grad_norm": 0.8046997785568237,
-      "learning_rate": 5.8618925831202054e-06,
-      "loss": 0.0819,
-      "step": 2427
-    },
-    {
-      "epoch": 0.4139702160534939,
-      "grad_norm": 1.4850029945373535,
-      "learning_rate": 5.8601875532821825e-06,
-      "loss": 0.1769,
-      "step": 2428
-    },
-    {
-      "epoch": 0.4141407144950316,
-      "grad_norm": 1.0972051620483398,
-      "learning_rate": 5.858482523444161e-06,
-      "loss": 0.1405,
-      "step": 2429
-    },
-    {
-      "epoch": 0.41431121293656925,
-      "grad_norm": 1.0775092840194702,
-      "learning_rate": 5.8567774936061384e-06,
-      "loss": 0.1195,
-      "step": 2430
-    },
-    {
-      "epoch": 0.4144817113781069,
-      "grad_norm": 1.1166963577270508,
-      "learning_rate": 5.855072463768116e-06,
-      "loss": 0.2072,
-      "step": 2431
-    },
-    {
-      "epoch": 0.4146522098196446,
-      "grad_norm": 0.7372758388519287,
-      "learning_rate": 5.8533674339300935e-06,
-      "loss": 0.0925,
-      "step": 2432
-    },
-    {
-      "epoch": 0.4148227082611823,
-      "grad_norm": 1.672086477279663,
-      "learning_rate": 5.851662404092072e-06,
-      "loss": 0.1642,
-      "step": 2433
-    },
-    {
-      "epoch": 0.41499320670272,
-      "grad_norm": 1.4401135444641113,
-      "learning_rate": 5.849957374254049e-06,
-      "loss": 0.1533,
-      "step": 2434
-    },
-    {
-      "epoch": 0.41516370514425766,
-      "grad_norm": 1.2260637283325195,
-      "learning_rate": 5.848252344416027e-06,
-      "loss": 0.2409,
-      "step": 2435
-    },
-    {
-      "epoch": 0.41533420358579537,
-      "grad_norm": 0.8733996748924255,
-      "learning_rate": 5.846547314578006e-06,
-      "loss": 0.1268,
-      "step": 2436
-    },
-    {
-      "epoch": 0.41550470202733303,
-      "grad_norm": 1.0725549459457397,
-      "learning_rate": 5.844842284739983e-06,
-      "loss": 0.1409,
-      "step": 2437
-    },
-    {
-      "epoch": 0.4156752004688707,
-      "grad_norm": 0.8036240339279175,
-      "learning_rate": 5.843137254901961e-06,
-      "loss": 0.1157,
-      "step": 2438
-    },
-    {
-      "epoch": 0.4158456989104084,
-      "grad_norm": 0.9177947044372559,
-      "learning_rate": 5.841432225063939e-06,
-      "loss": 0.1678,
-      "step": 2439
-    },
-    {
-      "epoch": 0.41601619735194606,
-      "grad_norm": 1.0422391891479492,
-      "learning_rate": 5.839727195225917e-06,
-      "loss": 0.1663,
-      "step": 2440
-    },
-    {
-      "epoch": 0.4161866957934838,
-      "grad_norm": 0.8394445776939392,
-      "learning_rate": 5.838022165387894e-06,
-      "loss": 0.1335,
-      "step": 2441
-    },
-    {
-      "epoch": 0.41635719423502143,
-      "grad_norm": 0.8703233599662781,
-      "learning_rate": 5.836317135549873e-06,
-      "loss": 0.1439,
-      "step": 2442
-    },
-    {
-      "epoch": 0.41652769267655915,
-      "grad_norm": 0.8049616813659668,
-      "learning_rate": 5.83461210571185e-06,
-      "loss": 0.1275,
-      "step": 2443
-    },
-    {
-      "epoch": 0.4166981911180968,
-      "grad_norm": 1.0591437816619873,
-      "learning_rate": 5.832907075873828e-06,
-      "loss": 0.1545,
-      "step": 2444
-    },
-    {
-      "epoch": 0.4168686895596345,
-      "grad_norm": 0.7541891932487488,
-      "learning_rate": 5.831202046035807e-06,
-      "loss": 0.0799,
-      "step": 2445
-    },
-    {
-      "epoch": 0.4170391880011722,
-      "grad_norm": 1.1001821756362915,
-      "learning_rate": 5.829497016197784e-06,
-      "loss": 0.0832,
-      "step": 2446
-    },
-    {
-      "epoch": 0.41720968644270984,
-      "grad_norm": 0.8933374881744385,
-      "learning_rate": 5.827791986359762e-06,
-      "loss": 0.1245,
-      "step": 2447
-    },
-    {
-      "epoch": 0.41738018488424755,
-      "grad_norm": 0.9707329273223877,
-      "learning_rate": 5.826086956521739e-06,
-      "loss": 0.1819,
-      "step": 2448
-    },
-    {
-      "epoch": 0.4175506833257852,
-      "grad_norm": 0.82521653175354,
-      "learning_rate": 5.824381926683718e-06,
-      "loss": 0.1236,
-      "step": 2449
-    },
-    {
-      "epoch": 0.4177211817673229,
-      "grad_norm": 1.5881226062774658,
-      "learning_rate": 5.822676896845695e-06,
-      "loss": 0.2405,
-      "step": 2450
-    },
-    {
-      "epoch": 0.4178916802088606,
-      "grad_norm": 0.9384194016456604,
-      "learning_rate": 5.820971867007673e-06,
-      "loss": 0.1518,
-      "step": 2451
-    },
-    {
-      "epoch": 0.4180621786503983,
-      "grad_norm": 1.1350349187850952,
-      "learning_rate": 5.819266837169652e-06,
-      "loss": 0.1233,
-      "step": 2452
-    },
-    {
-      "epoch": 0.41823267709193596,
-      "grad_norm": 0.788532555103302,
-      "learning_rate": 5.817561807331629e-06,
-      "loss": 0.0644,
-      "step": 2453
-    },
-    {
-      "epoch": 0.4184031755334736,
-      "grad_norm": 1.639487862586975,
-      "learning_rate": 5.815856777493607e-06,
-      "loss": 0.2061,
-      "step": 2454
-    },
-    {
-      "epoch": 0.41857367397501133,
-      "grad_norm": 0.8484280705451965,
-      "learning_rate": 5.814151747655585e-06,
-      "loss": 0.1293,
-      "step": 2455
-    },
-    {
-      "epoch": 0.418744172416549,
-      "grad_norm": 0.7462180852890015,
-      "learning_rate": 5.8124467178175625e-06,
-      "loss": 0.0625,
-      "step": 2456
-    },
-    {
-      "epoch": 0.4189146708580867,
-      "grad_norm": 0.6771153211593628,
-      "learning_rate": 5.81074168797954e-06,
-      "loss": 0.0534,
-      "step": 2457
-    },
-    {
-      "epoch": 0.41908516929962436,
-      "grad_norm": 1.4879751205444336,
-      "learning_rate": 5.8090366581415184e-06,
-      "loss": 0.2232,
-      "step": 2458
-    },
-    {
-      "epoch": 0.4192556677411621,
-      "grad_norm": 1.8108857870101929,
-      "learning_rate": 5.8073316283034955e-06,
-      "loss": 0.2234,
-      "step": 2459
-    },
-    {
-      "epoch": 0.41942616618269973,
-      "grad_norm": 0.8870851993560791,
-      "learning_rate": 5.8056265984654735e-06,
-      "loss": 0.1446,
-      "step": 2460
-    },
-    {
-      "epoch": 0.4195966646242374,
-      "grad_norm": 1.2096353769302368,
-      "learning_rate": 5.803921568627452e-06,
-      "loss": 0.1226,
-      "step": 2461
-    },
-    {
-      "epoch": 0.4197671630657751,
-      "grad_norm": 0.8411592245101929,
-      "learning_rate": 5.802216538789429e-06,
-      "loss": 0.0914,
-      "step": 2462
-    },
-    {
-      "epoch": 0.41993766150731276,
-      "grad_norm": 0.7253265380859375,
-      "learning_rate": 5.800511508951407e-06,
-      "loss": 0.1,
-      "step": 2463
-    },
-    {
-      "epoch": 0.4201081599488505,
-      "grad_norm": 1.141318440437317,
-      "learning_rate": 5.7988064791133844e-06,
-      "loss": 0.1903,
-      "step": 2464
-    },
-    {
-      "epoch": 0.42027865839038814,
-      "grad_norm": 1.2091392278671265,
-      "learning_rate": 5.797101449275363e-06,
-      "loss": 0.1197,
-      "step": 2465
-    },
-    {
-      "epoch": 0.42044915683192585,
-      "grad_norm": 1.1575111150741577,
-      "learning_rate": 5.79539641943734e-06,
-      "loss": 0.1327,
-      "step": 2466
-    },
-    {
-      "epoch": 0.4206196552734635,
-      "grad_norm": 0.7505564093589783,
-      "learning_rate": 5.793691389599318e-06,
-      "loss": 0.0827,
-      "step": 2467
-    },
-    {
-      "epoch": 0.4207901537150012,
-      "grad_norm": 0.7998400330543518,
-      "learning_rate": 5.791986359761295e-06,
-      "loss": 0.0942,
-      "step": 2468
-    },
-    {
-      "epoch": 0.4209606521565389,
-      "grad_norm": 1.3579339981079102,
-      "learning_rate": 5.790281329923274e-06,
-      "loss": 0.2063,
-      "step": 2469
-    },
-    {
-      "epoch": 0.42113115059807654,
-      "grad_norm": 0.9726285934448242,
-      "learning_rate": 5.788576300085252e-06,
-      "loss": 0.1824,
-      "step": 2470
-    },
-    {
-      "epoch": 0.42130164903961426,
-      "grad_norm": 0.8264228105545044,
-      "learning_rate": 5.786871270247229e-06,
-      "loss": 0.1467,
-      "step": 2471
-    },
-    {
-      "epoch": 0.4214721474811519,
-      "grad_norm": 0.8461482524871826,
-      "learning_rate": 5.785166240409208e-06,
-      "loss": 0.1412,
-      "step": 2472
-    },
-    {
-      "epoch": 0.42164264592268963,
-      "grad_norm": 0.8347538709640503,
-      "learning_rate": 5.783461210571185e-06,
-      "loss": 0.0636,
-      "step": 2473
-    },
-    {
-      "epoch": 0.4218131443642273,
-      "grad_norm": 0.5894155502319336,
-      "learning_rate": 5.781756180733163e-06,
-      "loss": 0.0559,
-      "step": 2474
-    },
-    {
-      "epoch": 0.421983642805765,
-      "grad_norm": 1.2116296291351318,
-      "learning_rate": 5.780051150895141e-06,
-      "loss": 0.1289,
-      "step": 2475
-    },
-    {
-      "epoch": 0.42215414124730266,
-      "grad_norm": 0.7940623164176941,
-      "learning_rate": 5.778346121057119e-06,
-      "loss": 0.0536,
-      "step": 2476
-    },
-    {
-      "epoch": 0.4223246396888403,
-      "grad_norm": 0.9565810561180115,
-      "learning_rate": 5.776641091219098e-06,
-      "loss": 0.1271,
-      "step": 2477
-    },
-    {
-      "epoch": 0.42249513813037803,
-      "grad_norm": 0.9501217007637024,
-      "learning_rate": 5.774936061381075e-06,
-      "loss": 0.1019,
-      "step": 2478
-    },
-    {
-      "epoch": 0.4226656365719157,
-      "grad_norm": 1.0174156427383423,
-      "learning_rate": 5.773231031543053e-06,
-      "loss": 0.1307,
-      "step": 2479
-    },
-    {
-      "epoch": 0.4228361350134534,
-      "grad_norm": 1.1585232019424438,
-      "learning_rate": 5.77152600170503e-06,
-      "loss": 0.1416,
-      "step": 2480
-    },
-    {
-      "epoch": 0.42300663345499107,
-      "grad_norm": 1.277368426322937,
-      "learning_rate": 5.769820971867009e-06,
-      "loss": 0.1621,
-      "step": 2481
-    },
-    {
-      "epoch": 0.4231771318965288,
-      "grad_norm": 0.6454185843467712,
-      "learning_rate": 5.768115942028986e-06,
-      "loss": 0.0797,
-      "step": 2482
-    },
-    {
-      "epoch": 0.42334763033806644,
-      "grad_norm": 0.8542653918266296,
-      "learning_rate": 5.766410912190964e-06,
-      "loss": 0.0422,
-      "step": 2483
-    },
-    {
-      "epoch": 0.4235181287796041,
-      "grad_norm": 1.4021434783935547,
-      "learning_rate": 5.764705882352941e-06,
-      "loss": 0.1871,
-      "step": 2484
-    },
-    {
-      "epoch": 0.4236886272211418,
-      "grad_norm": 1.434823751449585,
-      "learning_rate": 5.76300085251492e-06,
-      "loss": 0.1693,
-      "step": 2485
-    },
-    {
-      "epoch": 0.42385912566267947,
-      "grad_norm": 1.4753504991531372,
-      "learning_rate": 5.761295822676898e-06,
-      "loss": 0.1383,
-      "step": 2486
-    },
-    {
-      "epoch": 0.4240296241042172,
-      "grad_norm": 1.8268640041351318,
-      "learning_rate": 5.759590792838875e-06,
-      "loss": 0.213,
-      "step": 2487
-    },
-    {
-      "epoch": 0.42420012254575484,
-      "grad_norm": 1.1246039867401123,
-      "learning_rate": 5.7578857630008535e-06,
-      "loss": 0.1891,
-      "step": 2488
-    },
-    {
-      "epoch": 0.42437062098729256,
-      "grad_norm": 1.0577718019485474,
-      "learning_rate": 5.756180733162831e-06,
-      "loss": 0.1782,
-      "step": 2489
-    },
-    {
-      "epoch": 0.4245411194288302,
-      "grad_norm": 1.2953256368637085,
-      "learning_rate": 5.7544757033248085e-06,
-      "loss": 0.1958,
-      "step": 2490
-    },
-    {
-      "epoch": 0.42471161787036793,
-      "grad_norm": 1.3340262174606323,
-      "learning_rate": 5.7527706734867865e-06,
-      "loss": 0.2098,
-      "step": 2491
-    },
-    {
-      "epoch": 0.4248821163119056,
-      "grad_norm": 1.4178156852722168,
-      "learning_rate": 5.751065643648764e-06,
-      "loss": 0.1211,
-      "step": 2492
-    },
-    {
-      "epoch": 0.42505261475344325,
-      "grad_norm": 1.2180986404418945,
-      "learning_rate": 5.7493606138107415e-06,
-      "loss": 0.17,
-      "step": 2493
-    },
-    {
-      "epoch": 0.42522311319498096,
-      "grad_norm": 1.0153529644012451,
-      "learning_rate": 5.74765558397272e-06,
-      "loss": 0.1066,
-      "step": 2494
-    },
-    {
-      "epoch": 0.4253936116365186,
-      "grad_norm": 1.0207457542419434,
-      "learning_rate": 5.745950554134698e-06,
-      "loss": 0.1531,
-      "step": 2495
-    },
-    {
-      "epoch": 0.42556411007805633,
-      "grad_norm": 1.096200942993164,
-      "learning_rate": 5.744245524296675e-06,
-      "loss": 0.0736,
-      "step": 2496
-    },
-    {
-      "epoch": 0.425734608519594,
-      "grad_norm": 1.0620962381362915,
-      "learning_rate": 5.742540494458654e-06,
-      "loss": 0.1664,
-      "step": 2497
-    },
-    {
-      "epoch": 0.4259051069611317,
-      "grad_norm": 1.4350156784057617,
-      "learning_rate": 5.740835464620631e-06,
-      "loss": 0.1166,
-      "step": 2498
-    },
-    {
-      "epoch": 0.42607560540266937,
-      "grad_norm": 1.700987458229065,
-      "learning_rate": 5.739130434782609e-06,
-      "loss": 0.2208,
-      "step": 2499
-    },
-    {
-      "epoch": 0.426246103844207,
-      "grad_norm": 0.8968698978424072,
-      "learning_rate": 5.737425404944586e-06,
-      "loss": 0.1277,
-      "step": 2500
-    },
-    {
-      "epoch": 0.426246103844207,
-      "eval_f1_score": 0.3444730077120823,
-      "eval_loss": 0.14721998572349548,
-      "eval_runtime": 182.6201,
-      "eval_samples_per_second": 54.758,
-      "eval_steps_per_second": 3.422,
-      "step": 2500
-    },
-    {
-      "epoch": 0.42641660228574474,
-      "grad_norm": 1.3592954874038696,
-      "learning_rate": 5.735720375106565e-06,
-      "loss": 0.1583,
-      "step": 2501
-    },
-    {
-      "epoch": 0.4265871007272824,
-      "grad_norm": 0.8200425505638123,
-      "learning_rate": 5.734015345268543e-06,
-      "loss": 0.1133,
-      "step": 2502
-    },
-    {
-      "epoch": 0.4267575991688201,
-      "grad_norm": 0.929311990737915,
-      "learning_rate": 5.73231031543052e-06,
-      "loss": 0.1461,
-      "step": 2503
-    },
-    {
-      "epoch": 0.42692809761035777,
-      "grad_norm": 0.7825623154640198,
-      "learning_rate": 5.730605285592499e-06,
-      "loss": 0.1384,
-      "step": 2504
-    },
-    {
-      "epoch": 0.4270985960518955,
-      "grad_norm": 0.8742008805274963,
-      "learning_rate": 5.728900255754476e-06,
-      "loss": 0.1035,
-      "step": 2505
-    },
-    {
-      "epoch": 0.42726909449343314,
-      "grad_norm": 0.7557575106620789,
-      "learning_rate": 5.727195225916454e-06,
-      "loss": 0.1381,
-      "step": 2506
-    },
-    {
-      "epoch": 0.4274395929349708,
-      "grad_norm": 1.0146187543869019,
-      "learning_rate": 5.725490196078431e-06,
-      "loss": 0.1468,
-      "step": 2507
-    },
-    {
-      "epoch": 0.4276100913765085,
-      "grad_norm": 1.1496630907058716,
-      "learning_rate": 5.72378516624041e-06,
-      "loss": 0.1925,
-      "step": 2508
-    },
-    {
-      "epoch": 0.4277805898180462,
-      "grad_norm": 1.1652343273162842,
-      "learning_rate": 5.722080136402387e-06,
-      "loss": 0.1873,
-      "step": 2509
-    },
-    {
-      "epoch": 0.4279510882595839,
-      "grad_norm": 1.3467806577682495,
-      "learning_rate": 5.720375106564365e-06,
-      "loss": 0.1659,
-      "step": 2510
-    },
-    {
-      "epoch": 0.42812158670112155,
-      "grad_norm": 1.2672559022903442,
-      "learning_rate": 5.718670076726344e-06,
-      "loss": 0.18,
-      "step": 2511
-    },
-    {
-      "epoch": 0.42829208514265926,
-      "grad_norm": 0.8156704306602478,
-      "learning_rate": 5.716965046888321e-06,
-      "loss": 0.1384,
-      "step": 2512
-    },
-    {
-      "epoch": 0.4284625835841969,
-      "grad_norm": 0.9403845071792603,
-      "learning_rate": 5.715260017050299e-06,
-      "loss": 0.1376,
-      "step": 2513
-    },
-    {
-      "epoch": 0.42863308202573464,
-      "grad_norm": 0.8247608542442322,
-      "learning_rate": 5.713554987212277e-06,
-      "loss": 0.1188,
-      "step": 2514
-    },
-    {
-      "epoch": 0.4288035804672723,
-      "grad_norm": 1.153664231300354,
-      "learning_rate": 5.711849957374255e-06,
-      "loss": 0.0713,
-      "step": 2515
-    },
-    {
-      "epoch": 0.42897407890880995,
-      "grad_norm": 1.1019221544265747,
-      "learning_rate": 5.710144927536232e-06,
-      "loss": 0.1233,
-      "step": 2516
-    },
-    {
-      "epoch": 0.42914457735034767,
-      "grad_norm": 1.1238685846328735,
-      "learning_rate": 5.7084398976982106e-06,
-      "loss": 0.1802,
-      "step": 2517
-    },
-    {
-      "epoch": 0.4293150757918853,
-      "grad_norm": 0.8956630229949951,
-      "learning_rate": 5.706734867860188e-06,
-      "loss": 0.111,
-      "step": 2518
-    },
-    {
-      "epoch": 0.42948557423342304,
-      "grad_norm": 0.7892765402793884,
-      "learning_rate": 5.705029838022166e-06,
-      "loss": 0.0833,
-      "step": 2519
-    },
-    {
-      "epoch": 0.4296560726749607,
-      "grad_norm": 0.8521721363067627,
-      "learning_rate": 5.703324808184144e-06,
-      "loss": 0.0968,
-      "step": 2520
-    },
-    {
-      "epoch": 0.4298265711164984,
-      "grad_norm": 0.9164546132087708,
-      "learning_rate": 5.7016197783461215e-06,
-      "loss": 0.1135,
-      "step": 2521
-    },
-    {
-      "epoch": 0.42999706955803607,
-      "grad_norm": 0.948046088218689,
-      "learning_rate": 5.6999147485080995e-06,
-      "loss": 0.0859,
-      "step": 2522
-    },
-    {
-      "epoch": 0.43016756799957373,
-      "grad_norm": 0.5832633376121521,
-      "learning_rate": 5.6982097186700766e-06,
-      "loss": 0.0928,
-      "step": 2523
-    },
-    {
-      "epoch": 0.43033806644111144,
-      "grad_norm": 0.8962041735649109,
-      "learning_rate": 5.696504688832055e-06,
-      "loss": 0.1255,
-      "step": 2524
-    },
-    {
-      "epoch": 0.4305085648826491,
-      "grad_norm": 0.7559568881988525,
-      "learning_rate": 5.6947996589940325e-06,
-      "loss": 0.0877,
-      "step": 2525
-    },
-    {
-      "epoch": 0.4306790633241868,
-      "grad_norm": 1.1643157005310059,
-      "learning_rate": 5.69309462915601e-06,
-      "loss": 0.1177,
-      "step": 2526
-    },
-    {
-      "epoch": 0.4308495617657245,
-      "grad_norm": 1.4461086988449097,
-      "learning_rate": 5.691389599317989e-06,
-      "loss": 0.1757,
-      "step": 2527
-    },
-    {
-      "epoch": 0.4310200602072622,
-      "grad_norm": 1.2662277221679688,
-      "learning_rate": 5.689684569479966e-06,
-      "loss": 0.1868,
-      "step": 2528
-    },
-    {
-      "epoch": 0.43119055864879985,
-      "grad_norm": 0.7486140727996826,
-      "learning_rate": 5.687979539641944e-06,
-      "loss": 0.0875,
-      "step": 2529
-    },
-    {
-      "epoch": 0.4313610570903375,
-      "grad_norm": 0.9595978260040283,
-      "learning_rate": 5.686274509803922e-06,
-      "loss": 0.1163,
-      "step": 2530
-    },
-    {
-      "epoch": 0.4315315555318752,
-      "grad_norm": 0.8530360460281372,
-      "learning_rate": 5.6845694799659e-06,
-      "loss": 0.1208,
-      "step": 2531
-    },
-    {
-      "epoch": 0.4317020539734129,
-      "grad_norm": 1.6544944047927856,
-      "learning_rate": 5.682864450127877e-06,
-      "loss": 0.1726,
-      "step": 2532
-    },
-    {
-      "epoch": 0.4318725524149506,
-      "grad_norm": 1.4921165704727173,
-      "learning_rate": 5.681159420289856e-06,
-      "loss": 0.1588,
-      "step": 2533
-    },
-    {
-      "epoch": 0.43204305085648825,
-      "grad_norm": 0.7876043915748596,
-      "learning_rate": 5.679454390451833e-06,
-      "loss": 0.1184,
-      "step": 2534
-    },
-    {
-      "epoch": 0.43221354929802597,
-      "grad_norm": 1.3764668703079224,
-      "learning_rate": 5.677749360613811e-06,
-      "loss": 0.1859,
-      "step": 2535
-    },
-    {
-      "epoch": 0.4323840477395636,
-      "grad_norm": 0.6521998047828674,
-      "learning_rate": 5.67604433077579e-06,
-      "loss": 0.0985,
-      "step": 2536
-    },
-    {
-      "epoch": 0.43255454618110134,
-      "grad_norm": 0.9827353358268738,
-      "learning_rate": 5.674339300937767e-06,
-      "loss": 0.0902,
-      "step": 2537
-    },
-    {
-      "epoch": 0.432725044622639,
-      "grad_norm": 0.9378973245620728,
-      "learning_rate": 5.672634271099745e-06,
-      "loss": 0.1793,
-      "step": 2538
-    },
-    {
-      "epoch": 0.43289554306417666,
-      "grad_norm": 0.8462679982185364,
-      "learning_rate": 5.670929241261722e-06,
-      "loss": 0.1012,
-      "step": 2539
-    },
-    {
-      "epoch": 0.43306604150571437,
-      "grad_norm": 0.865755021572113,
-      "learning_rate": 5.669224211423701e-06,
-      "loss": 0.1187,
-      "step": 2540
-    },
-    {
-      "epoch": 0.43323653994725203,
-      "grad_norm": 0.9696370363235474,
-      "learning_rate": 5.667519181585678e-06,
-      "loss": 0.1371,
-      "step": 2541
-    },
-    {
-      "epoch": 0.43340703838878974,
-      "grad_norm": 0.8784907460212708,
-      "learning_rate": 5.665814151747656e-06,
-      "loss": 0.0928,
-      "step": 2542
-    },
-    {
-      "epoch": 0.4335775368303274,
-      "grad_norm": 0.9317367076873779,
-      "learning_rate": 5.664109121909633e-06,
-      "loss": 0.1148,
-      "step": 2543
-    },
-    {
-      "epoch": 0.4337480352718651,
-      "grad_norm": 0.8701735734939575,
-      "learning_rate": 5.662404092071612e-06,
-      "loss": 0.1485,
-      "step": 2544
-    },
-    {
-      "epoch": 0.4339185337134028,
-      "grad_norm": 0.946918249130249,
-      "learning_rate": 5.66069906223359e-06,
-      "loss": 0.1782,
-      "step": 2545
-    },
-    {
-      "epoch": 0.43408903215494044,
-      "grad_norm": 1.190557599067688,
-      "learning_rate": 5.658994032395567e-06,
-      "loss": 0.1412,
-      "step": 2546
-    },
-    {
-      "epoch": 0.43425953059647815,
-      "grad_norm": 0.9963315725326538,
-      "learning_rate": 5.657289002557546e-06,
-      "loss": 0.1271,
-      "step": 2547
-    },
-    {
-      "epoch": 0.4344300290380158,
-      "grad_norm": 0.9686456918716431,
-      "learning_rate": 5.655583972719523e-06,
-      "loss": 0.1744,
-      "step": 2548
-    },
-    {
-      "epoch": 0.4346005274795535,
-      "grad_norm": 1.0159608125686646,
-      "learning_rate": 5.653878942881501e-06,
-      "loss": 0.1458,
-      "step": 2549
-    },
-    {
-      "epoch": 0.4347710259210912,
-      "grad_norm": 0.9071111679077148,
-      "learning_rate": 5.652173913043479e-06,
-      "loss": 0.1453,
-      "step": 2550
-    },
-    {
-      "epoch": 0.4349415243626289,
-      "grad_norm": 1.0276676416397095,
-      "learning_rate": 5.6504688832054566e-06,
-      "loss": 0.2083,
-      "step": 2551
-    },
-    {
-      "epoch": 0.43511202280416655,
-      "grad_norm": 0.961588978767395,
-      "learning_rate": 5.648763853367435e-06,
-      "loss": 0.113,
-      "step": 2552
-    },
-    {
-      "epoch": 0.4352825212457042,
-      "grad_norm": 1.0574619770050049,
-      "learning_rate": 5.6470588235294125e-06,
-      "loss": 0.07,
-      "step": 2553
-    },
-    {
-      "epoch": 0.4354530196872419,
-      "grad_norm": 1.0037537813186646,
-      "learning_rate": 5.64535379369139e-06,
-      "loss": 0.098,
-      "step": 2554
-    },
-    {
-      "epoch": 0.4356235181287796,
-      "grad_norm": 1.0496973991394043,
-      "learning_rate": 5.6436487638533675e-06,
-      "loss": 0.0868,
-      "step": 2555
-    },
-    {
-      "epoch": 0.4357940165703173,
-      "grad_norm": 0.7770395874977112,
-      "learning_rate": 5.641943734015346e-06,
-      "loss": 0.1122,
-      "step": 2556
-    },
-    {
-      "epoch": 0.43596451501185496,
-      "grad_norm": 1.2403090000152588,
-      "learning_rate": 5.640238704177323e-06,
-      "loss": 0.151,
-      "step": 2557
-    },
-    {
-      "epoch": 0.4361350134533927,
-      "grad_norm": 0.9569463133811951,
-      "learning_rate": 5.638533674339301e-06,
-      "loss": 0.1395,
-      "step": 2558
-    },
-    {
-      "epoch": 0.43630551189493033,
-      "grad_norm": 1.369671106338501,
-      "learning_rate": 5.6368286445012784e-06,
-      "loss": 0.1845,
-      "step": 2559
-    },
-    {
-      "epoch": 0.43647601033646805,
-      "grad_norm": 1.5948243141174316,
-      "learning_rate": 5.635123614663257e-06,
-      "loss": 0.1768,
-      "step": 2560
-    },
-    {
-      "epoch": 0.4366465087780057,
-      "grad_norm": 0.9856839179992676,
-      "learning_rate": 5.633418584825235e-06,
-      "loss": 0.1632,
-      "step": 2561
-    },
-    {
-      "epoch": 0.43681700721954336,
-      "grad_norm": 1.1320617198944092,
-      "learning_rate": 5.631713554987212e-06,
-      "loss": 0.1526,
-      "step": 2562
-    },
-    {
-      "epoch": 0.4369875056610811,
-      "grad_norm": 0.9488789439201355,
-      "learning_rate": 5.630008525149191e-06,
-      "loss": 0.1419,
-      "step": 2563
-    },
-    {
-      "epoch": 0.43715800410261874,
-      "grad_norm": 1.1834876537322998,
-      "learning_rate": 5.628303495311168e-06,
-      "loss": 0.1438,
-      "step": 2564
-    },
-    {
-      "epoch": 0.43732850254415645,
-      "grad_norm": 0.8948205709457397,
-      "learning_rate": 5.626598465473146e-06,
-      "loss": 0.1362,
-      "step": 2565
-    },
-    {
-      "epoch": 0.4374990009856941,
-      "grad_norm": 0.7960487008094788,
-      "learning_rate": 5.624893435635124e-06,
-      "loss": 0.0527,
-      "step": 2566
-    },
-    {
-      "epoch": 0.4376694994272318,
-      "grad_norm": 1.006632685661316,
-      "learning_rate": 5.623188405797102e-06,
-      "loss": 0.0975,
-      "step": 2567
-    },
-    {
-      "epoch": 0.4378399978687695,
-      "grad_norm": 0.7732202410697937,
-      "learning_rate": 5.621483375959079e-06,
-      "loss": 0.0871,
-      "step": 2568
-    },
-    {
-      "epoch": 0.43801049631030714,
-      "grad_norm": 1.3993794918060303,
-      "learning_rate": 5.619778346121058e-06,
-      "loss": 0.12,
-      "step": 2569
-    },
-    {
-      "epoch": 0.43818099475184485,
-      "grad_norm": 0.6594243049621582,
-      "learning_rate": 5.618073316283036e-06,
-      "loss": 0.0496,
-      "step": 2570
-    },
-    {
-      "epoch": 0.4383514931933825,
-      "grad_norm": 1.2651886940002441,
-      "learning_rate": 5.616368286445013e-06,
-      "loss": 0.207,
-      "step": 2571
-    },
-    {
-      "epoch": 0.4385219916349202,
-      "grad_norm": 1.066919207572937,
-      "learning_rate": 5.614663256606992e-06,
-      "loss": 0.1872,
-      "step": 2572
-    },
-    {
-      "epoch": 0.4386924900764579,
-      "grad_norm": 0.9206414222717285,
-      "learning_rate": 5.612958226768969e-06,
-      "loss": 0.1029,
-      "step": 2573
-    },
-    {
-      "epoch": 0.4388629885179956,
-      "grad_norm": 0.7342972159385681,
-      "learning_rate": 5.611253196930947e-06,
-      "loss": 0.0985,
-      "step": 2574
-    },
-    {
-      "epoch": 0.43903348695953326,
-      "grad_norm": 1.378420352935791,
-      "learning_rate": 5.609548167092924e-06,
-      "loss": 0.1871,
-      "step": 2575
-    },
-    {
-      "epoch": 0.4392039854010709,
-      "grad_norm": 0.7292861938476562,
-      "learning_rate": 5.607843137254903e-06,
-      "loss": 0.0928,
-      "step": 2576
-    },
-    {
-      "epoch": 0.43937448384260863,
-      "grad_norm": 0.9209851026535034,
-      "learning_rate": 5.606138107416881e-06,
-      "loss": 0.1098,
-      "step": 2577
-    },
-    {
-      "epoch": 0.4395449822841463,
-      "grad_norm": 1.275430679321289,
-      "learning_rate": 5.604433077578858e-06,
-      "loss": 0.1728,
-      "step": 2578
-    },
-    {
-      "epoch": 0.439715480725684,
-      "grad_norm": 1.1448216438293457,
-      "learning_rate": 5.6027280477408366e-06,
-      "loss": 0.155,
-      "step": 2579
-    },
-    {
-      "epoch": 0.43988597916722166,
-      "grad_norm": 1.0969963073730469,
-      "learning_rate": 5.601023017902814e-06,
-      "loss": 0.1484,
-      "step": 2580
-    },
-    {
-      "epoch": 0.4400564776087594,
-      "grad_norm": 1.186651349067688,
-      "learning_rate": 5.599317988064792e-06,
-      "loss": 0.1065,
-      "step": 2581
-    },
-    {
-      "epoch": 0.44022697605029704,
-      "grad_norm": 1.65304696559906,
-      "learning_rate": 5.597612958226769e-06,
-      "loss": 0.2714,
-      "step": 2582
-    },
-    {
-      "epoch": 0.44039747449183475,
-      "grad_norm": 0.9879024624824524,
-      "learning_rate": 5.5959079283887475e-06,
-      "loss": 0.1222,
-      "step": 2583
-    },
-    {
-      "epoch": 0.4405679729333724,
-      "grad_norm": 1.8686827421188354,
-      "learning_rate": 5.594202898550725e-06,
-      "loss": 0.2354,
-      "step": 2584
-    },
-    {
-      "epoch": 0.44073847137491007,
-      "grad_norm": 0.6579173803329468,
-      "learning_rate": 5.5924978687127025e-06,
-      "loss": 0.1011,
-      "step": 2585
-    },
-    {
-      "epoch": 0.4409089698164478,
-      "grad_norm": 0.9362009167671204,
-      "learning_rate": 5.590792838874681e-06,
-      "loss": 0.0956,
-      "step": 2586
-    },
-    {
-      "epoch": 0.44107946825798544,
-      "grad_norm": 0.7479416728019714,
-      "learning_rate": 5.5890878090366584e-06,
-      "loss": 0.1188,
-      "step": 2587
-    },
-    {
-      "epoch": 0.44124996669952316,
-      "grad_norm": 1.0462018251419067,
-      "learning_rate": 5.587382779198636e-06,
-      "loss": 0.104,
-      "step": 2588
-    },
-    {
-      "epoch": 0.4414204651410608,
-      "grad_norm": 1.1154561042785645,
-      "learning_rate": 5.585677749360614e-06,
-      "loss": 0.1217,
-      "step": 2589
-    },
-    {
-      "epoch": 0.44159096358259853,
-      "grad_norm": 0.7707915306091309,
-      "learning_rate": 5.583972719522592e-06,
-      "loss": 0.1014,
-      "step": 2590
-    },
-    {
-      "epoch": 0.4417614620241362,
-      "grad_norm": 0.7629173994064331,
-      "learning_rate": 5.582267689684569e-06,
-      "loss": 0.11,
-      "step": 2591
-    },
-    {
-      "epoch": 0.44193196046567385,
-      "grad_norm": 1.3871240615844727,
-      "learning_rate": 5.580562659846548e-06,
-      "loss": 0.1768,
-      "step": 2592
-    },
-    {
-      "epoch": 0.44210245890721156,
-      "grad_norm": 0.8774785399436951,
-      "learning_rate": 5.578857630008525e-06,
-      "loss": 0.0447,
-      "step": 2593
-    },
-    {
-      "epoch": 0.4422729573487492,
-      "grad_norm": 0.794389545917511,
-      "learning_rate": 5.577152600170503e-06,
-      "loss": 0.1039,
-      "step": 2594
-    },
-    {
-      "epoch": 0.44244345579028693,
-      "grad_norm": 0.7697991728782654,
-      "learning_rate": 5.575447570332482e-06,
-      "loss": 0.1103,
-      "step": 2595
-    },
-    {
-      "epoch": 0.4426139542318246,
-      "grad_norm": 0.800696849822998,
-      "learning_rate": 5.573742540494459e-06,
-      "loss": 0.0875,
-      "step": 2596
-    },
-    {
-      "epoch": 0.4427844526733623,
-      "grad_norm": 2.0255141258239746,
-      "learning_rate": 5.572037510656437e-06,
-      "loss": 0.2019,
-      "step": 2597
-    },
-    {
-      "epoch": 0.44295495111489996,
-      "grad_norm": 1.5932087898254395,
-      "learning_rate": 5.570332480818414e-06,
-      "loss": 0.1325,
-      "step": 2598
-    },
-    {
-      "epoch": 0.4431254495564376,
-      "grad_norm": 1.2170913219451904,
-      "learning_rate": 5.568627450980393e-06,
-      "loss": 0.1787,
-      "step": 2599
-    },
-    {
-      "epoch": 0.44329594799797534,
-      "grad_norm": 1.0791867971420288,
-      "learning_rate": 5.56692242114237e-06,
-      "loss": 0.1481,
-      "step": 2600
-    },
-    {
-      "epoch": 0.44329594799797534,
-      "eval_f1_score": 0.30978260869565216,
-      "eval_loss": 0.14772719144821167,
-      "eval_runtime": 182.7464,
-      "eval_samples_per_second": 54.721,
-      "eval_steps_per_second": 3.42,
-      "step": 2600
-    },
-    {
-      "epoch": 0.443466446439513,
-      "grad_norm": 1.393898367881775,
-      "learning_rate": 5.565217391304348e-06,
-      "loss": 0.1351,
-      "step": 2601
-    },
-    {
-      "epoch": 0.4436369448810507,
-      "grad_norm": 1.1062179803848267,
-      "learning_rate": 5.563512361466326e-06,
-      "loss": 0.098,
-      "step": 2602
-    },
-    {
-      "epoch": 0.44380744332258837,
-      "grad_norm": 0.7602462768554688,
-      "learning_rate": 5.561807331628304e-06,
-      "loss": 0.0767,
-      "step": 2603
-    },
-    {
-      "epoch": 0.4439779417641261,
-      "grad_norm": 0.8465493321418762,
-      "learning_rate": 5.560102301790282e-06,
-      "loss": 0.0802,
-      "step": 2604
-    },
-    {
-      "epoch": 0.44414844020566374,
-      "grad_norm": 1.0767474174499512,
-      "learning_rate": 5.55839727195226e-06,
-      "loss": 0.0754,
-      "step": 2605
-    },
-    {
-      "epoch": 0.4443189386472014,
-      "grad_norm": 1.0401229858398438,
-      "learning_rate": 5.556692242114238e-06,
-      "loss": 0.1615,
-      "step": 2606
-    },
-    {
-      "epoch": 0.4444894370887391,
-      "grad_norm": 1.0522851943969727,
-      "learning_rate": 5.554987212276215e-06,
-      "loss": 0.1591,
-      "step": 2607
-    },
-    {
-      "epoch": 0.4446599355302768,
-      "grad_norm": 1.0689300298690796,
-      "learning_rate": 5.553282182438194e-06,
-      "loss": 0.1307,
-      "step": 2608
-    },
-    {
-      "epoch": 0.4448304339718145,
-      "grad_norm": 0.6239145398139954,
-      "learning_rate": 5.551577152600171e-06,
-      "loss": 0.0851,
-      "step": 2609
-    },
-    {
-      "epoch": 0.44500093241335215,
-      "grad_norm": 1.2811641693115234,
-      "learning_rate": 5.549872122762149e-06,
-      "loss": 0.2046,
-      "step": 2610
-    },
-    {
-      "epoch": 0.44517143085488986,
-      "grad_norm": 0.8296309113502502,
-      "learning_rate": 5.5481670929241275e-06,
-      "loss": 0.0499,
-      "step": 2611
-    },
-    {
-      "epoch": 0.4453419292964275,
-      "grad_norm": 1.7726424932479858,
-      "learning_rate": 5.546462063086105e-06,
-      "loss": 0.069,
-      "step": 2612
-    },
-    {
-      "epoch": 0.44551242773796523,
-      "grad_norm": 0.927315890789032,
-      "learning_rate": 5.5447570332480825e-06,
-      "loss": 0.0817,
-      "step": 2613
-    },
-    {
-      "epoch": 0.4456829261795029,
-      "grad_norm": 0.8551962375640869,
-      "learning_rate": 5.54305200341006e-06,
-      "loss": 0.0736,
-      "step": 2614
-    },
-    {
-      "epoch": 0.44585342462104055,
-      "grad_norm": 0.9609452486038208,
-      "learning_rate": 5.5413469735720384e-06,
-      "loss": 0.1334,
-      "step": 2615
-    },
-    {
-      "epoch": 0.44602392306257826,
-      "grad_norm": 0.8067399263381958,
-      "learning_rate": 5.5396419437340155e-06,
-      "loss": 0.1182,
-      "step": 2616
-    },
-    {
-      "epoch": 0.4461944215041159,
-      "grad_norm": 1.0610885620117188,
-      "learning_rate": 5.5379369138959935e-06,
-      "loss": 0.1351,
-      "step": 2617
-    },
-    {
-      "epoch": 0.44636491994565364,
-      "grad_norm": 1.2180925607681274,
-      "learning_rate": 5.536231884057971e-06,
-      "loss": 0.1059,
-      "step": 2618
-    },
-    {
-      "epoch": 0.4465354183871913,
-      "grad_norm": 0.5902736186981201,
-      "learning_rate": 5.534526854219949e-06,
-      "loss": 0.0571,
-      "step": 2619
-    },
-    {
-      "epoch": 0.446705916828729,
-      "grad_norm": 1.0947083234786987,
-      "learning_rate": 5.532821824381927e-06,
-      "loss": 0.1059,
-      "step": 2620
-    },
-    {
-      "epoch": 0.44687641527026667,
-      "grad_norm": 0.7008499503135681,
-      "learning_rate": 5.5311167945439044e-06,
-      "loss": 0.0926,
-      "step": 2621
-    },
-    {
-      "epoch": 0.44704691371180433,
-      "grad_norm": 1.785274863243103,
-      "learning_rate": 5.529411764705883e-06,
-      "loss": 0.2751,
-      "step": 2622
-    },
-    {
-      "epoch": 0.44721741215334204,
-      "grad_norm": 0.6739360690116882,
-      "learning_rate": 5.52770673486786e-06,
-      "loss": 0.0731,
-      "step": 2623
-    },
-    {
-      "epoch": 0.4473879105948797,
-      "grad_norm": 0.648942768573761,
-      "learning_rate": 5.526001705029838e-06,
-      "loss": 0.0356,
-      "step": 2624
-    },
-    {
-      "epoch": 0.4475584090364174,
-      "grad_norm": 0.6886118054389954,
-      "learning_rate": 5.524296675191816e-06,
-      "loss": 0.0757,
-      "step": 2625
-    },
-    {
-      "epoch": 0.4477289074779551,
-      "grad_norm": 0.6833609938621521,
-      "learning_rate": 5.522591645353794e-06,
-      "loss": 0.0329,
-      "step": 2626
-    },
-    {
-      "epoch": 0.4478994059194928,
-      "grad_norm": 1.5278652906417847,
-      "learning_rate": 5.520886615515771e-06,
-      "loss": 0.2367,
-      "step": 2627
-    },
-    {
-      "epoch": 0.44806990436103045,
-      "grad_norm": 0.5609342455863953,
-      "learning_rate": 5.51918158567775e-06,
-      "loss": 0.0851,
-      "step": 2628
-    },
-    {
-      "epoch": 0.4482404028025681,
-      "grad_norm": 1.545210361480713,
-      "learning_rate": 5.517476555839728e-06,
-      "loss": 0.1495,
-      "step": 2629
-    },
-    {
-      "epoch": 0.4484109012441058,
-      "grad_norm": 0.9157249927520752,
-      "learning_rate": 5.515771526001705e-06,
-      "loss": 0.1217,
-      "step": 2630
-    },
-    {
-      "epoch": 0.4485813996856435,
-      "grad_norm": 0.5781745910644531,
-      "learning_rate": 5.514066496163684e-06,
-      "loss": 0.0635,
-      "step": 2631
-    },
-    {
-      "epoch": 0.4487518981271812,
-      "grad_norm": 1.195381760597229,
-      "learning_rate": 5.512361466325661e-06,
-      "loss": 0.1764,
-      "step": 2632
-    },
-    {
-      "epoch": 0.44892239656871885,
-      "grad_norm": 1.0727818012237549,
-      "learning_rate": 5.510656436487639e-06,
-      "loss": 0.1179,
-      "step": 2633
-    },
-    {
-      "epoch": 0.44909289501025657,
-      "grad_norm": 1.5280828475952148,
-      "learning_rate": 5.508951406649616e-06,
-      "loss": 0.1587,
-      "step": 2634
-    },
-    {
-      "epoch": 0.4492633934517942,
-      "grad_norm": 1.2746130228042603,
-      "learning_rate": 5.507246376811595e-06,
-      "loss": 0.1229,
-      "step": 2635
-    },
-    {
-      "epoch": 0.44943389189333194,
-      "grad_norm": 0.7133687734603882,
-      "learning_rate": 5.505541346973573e-06,
-      "loss": 0.0498,
-      "step": 2636
-    },
-    {
-      "epoch": 0.4496043903348696,
-      "grad_norm": 0.8298734426498413,
-      "learning_rate": 5.50383631713555e-06,
-      "loss": 0.0974,
-      "step": 2637
-    },
-    {
-      "epoch": 0.44977488877640726,
-      "grad_norm": 0.7758700847625732,
-      "learning_rate": 5.502131287297529e-06,
-      "loss": 0.111,
-      "step": 2638
-    },
-    {
-      "epoch": 0.44994538721794497,
-      "grad_norm": 1.034207820892334,
-      "learning_rate": 5.500426257459506e-06,
-      "loss": 0.0883,
-      "step": 2639
-    },
-    {
-      "epoch": 0.45011588565948263,
-      "grad_norm": 0.9524661898612976,
-      "learning_rate": 5.498721227621484e-06,
-      "loss": 0.1076,
-      "step": 2640
-    },
-    {
-      "epoch": 0.45028638410102034,
-      "grad_norm": 1.1032789945602417,
-      "learning_rate": 5.497016197783462e-06,
-      "loss": 0.1365,
-      "step": 2641
-    },
-    {
-      "epoch": 0.450456882542558,
-      "grad_norm": 1.4389866590499878,
-      "learning_rate": 5.49531116794544e-06,
-      "loss": 0.0961,
-      "step": 2642
-    },
-    {
-      "epoch": 0.4506273809840957,
-      "grad_norm": 1.1861354112625122,
-      "learning_rate": 5.493606138107417e-06,
-      "loss": 0.1775,
-      "step": 2643
-    },
-    {
-      "epoch": 0.4507978794256334,
-      "grad_norm": 2.20349383354187,
-      "learning_rate": 5.4919011082693955e-06,
-      "loss": 0.245,
-      "step": 2644
-    },
-    {
-      "epoch": 0.45096837786717103,
-      "grad_norm": 0.9795979857444763,
-      "learning_rate": 5.4901960784313735e-06,
-      "loss": 0.1968,
-      "step": 2645
-    },
-    {
-      "epoch": 0.45113887630870875,
-      "grad_norm": 1.4784023761749268,
-      "learning_rate": 5.488491048593351e-06,
-      "loss": 0.2209,
-      "step": 2646
-    },
-    {
-      "epoch": 0.4513093747502464,
-      "grad_norm": 1.928300380706787,
-      "learning_rate": 5.486786018755329e-06,
-      "loss": 0.1449,
-      "step": 2647
-    },
-    {
-      "epoch": 0.4514798731917841,
-      "grad_norm": 1.053138017654419,
-      "learning_rate": 5.4850809889173065e-06,
-      "loss": 0.1044,
-      "step": 2648
-    },
-    {
-      "epoch": 0.4516503716333218,
-      "grad_norm": 1.2316356897354126,
-      "learning_rate": 5.4833759590792844e-06,
-      "loss": 0.149,
-      "step": 2649
-    },
-    {
-      "epoch": 0.4518208700748595,
-      "grad_norm": 0.7047519087791443,
-      "learning_rate": 5.4816709292412615e-06,
-      "loss": 0.0785,
-      "step": 2650
-    },
-    {
-      "epoch": 0.45199136851639715,
-      "grad_norm": 1.0934659242630005,
-      "learning_rate": 5.47996589940324e-06,
-      "loss": 0.1481,
-      "step": 2651
-    },
-    {
-      "epoch": 0.4521618669579348,
-      "grad_norm": 1.102778673171997,
-      "learning_rate": 5.478260869565217e-06,
-      "loss": 0.2052,
-      "step": 2652
-    },
-    {
-      "epoch": 0.4523323653994725,
-      "grad_norm": 1.1186070442199707,
-      "learning_rate": 5.476555839727195e-06,
-      "loss": 0.1752,
-      "step": 2653
-    },
-    {
-      "epoch": 0.4525028638410102,
-      "grad_norm": 1.0531350374221802,
-      "learning_rate": 5.474850809889174e-06,
-      "loss": 0.0994,
-      "step": 2654
-    },
-    {
-      "epoch": 0.4526733622825479,
-      "grad_norm": 1.003851294517517,
-      "learning_rate": 5.473145780051151e-06,
-      "loss": 0.1072,
-      "step": 2655
-    },
-    {
-      "epoch": 0.45284386072408556,
-      "grad_norm": 1.0673835277557373,
-      "learning_rate": 5.471440750213129e-06,
-      "loss": 0.1134,
-      "step": 2656
-    },
-    {
-      "epoch": 0.45301435916562327,
-      "grad_norm": 1.0477086305618286,
-      "learning_rate": 5.469735720375106e-06,
-      "loss": 0.0982,
-      "step": 2657
-    },
-    {
-      "epoch": 0.45318485760716093,
-      "grad_norm": 1.4844733476638794,
-      "learning_rate": 5.468030690537085e-06,
-      "loss": 0.1586,
-      "step": 2658
-    },
-    {
-      "epoch": 0.45335535604869864,
-      "grad_norm": 0.7507916688919067,
-      "learning_rate": 5.466325660699062e-06,
-      "loss": 0.1154,
-      "step": 2659
-    },
-    {
-      "epoch": 0.4535258544902363,
-      "grad_norm": 0.8093088269233704,
-      "learning_rate": 5.46462063086104e-06,
-      "loss": 0.088,
-      "step": 2660
-    },
-    {
-      "epoch": 0.45369635293177396,
-      "grad_norm": 0.9286861419677734,
-      "learning_rate": 5.462915601023019e-06,
-      "loss": 0.121,
-      "step": 2661
-    },
-    {
-      "epoch": 0.4538668513733117,
-      "grad_norm": 1.1589971780776978,
-      "learning_rate": 5.461210571184996e-06,
-      "loss": 0.1477,
-      "step": 2662
-    },
-    {
-      "epoch": 0.45403734981484933,
-      "grad_norm": 1.4251331090927124,
-      "learning_rate": 5.459505541346974e-06,
-      "loss": 0.1629,
-      "step": 2663
-    },
-    {
-      "epoch": 0.45420784825638705,
-      "grad_norm": 0.899058997631073,
-      "learning_rate": 5.457800511508952e-06,
-      "loss": 0.1363,
-      "step": 2664
-    },
-    {
-      "epoch": 0.4543783466979247,
-      "grad_norm": 0.684359073638916,
-      "learning_rate": 5.45609548167093e-06,
-      "loss": 0.0957,
-      "step": 2665
-    },
-    {
-      "epoch": 0.4545488451394624,
-      "grad_norm": 1.1812783479690552,
-      "learning_rate": 5.454390451832907e-06,
-      "loss": 0.1239,
-      "step": 2666
-    },
-    {
-      "epoch": 0.4547193435810001,
-      "grad_norm": 0.7902812361717224,
-      "learning_rate": 5.452685421994886e-06,
-      "loss": 0.1446,
-      "step": 2667
-    },
-    {
-      "epoch": 0.45488984202253774,
-      "grad_norm": 0.7050306797027588,
-      "learning_rate": 5.450980392156863e-06,
-      "loss": 0.065,
-      "step": 2668
-    },
-    {
-      "epoch": 0.45506034046407545,
-      "grad_norm": 0.608092725276947,
-      "learning_rate": 5.449275362318841e-06,
-      "loss": 0.0506,
-      "step": 2669
-    },
-    {
-      "epoch": 0.4552308389056131,
-      "grad_norm": 0.7197535037994385,
-      "learning_rate": 5.44757033248082e-06,
-      "loss": 0.0762,
-      "step": 2670
-    },
-    {
-      "epoch": 0.4554013373471508,
-      "grad_norm": 1.2062654495239258,
-      "learning_rate": 5.445865302642797e-06,
-      "loss": 0.1647,
-      "step": 2671
-    },
-    {
-      "epoch": 0.4555718357886885,
-      "grad_norm": 1.2899487018585205,
-      "learning_rate": 5.444160272804775e-06,
-      "loss": 0.1509,
-      "step": 2672
-    },
-    {
-      "epoch": 0.4557423342302262,
-      "grad_norm": 0.7482165098190308,
-      "learning_rate": 5.442455242966752e-06,
-      "loss": 0.1098,
-      "step": 2673
-    },
-    {
-      "epoch": 0.45591283267176386,
-      "grad_norm": 0.8238338232040405,
-      "learning_rate": 5.4407502131287306e-06,
-      "loss": 0.071,
-      "step": 2674
-    },
-    {
-      "epoch": 0.4560833311133015,
-      "grad_norm": 0.6520843505859375,
-      "learning_rate": 5.439045183290708e-06,
-      "loss": 0.062,
-      "step": 2675
-    },
-    {
-      "epoch": 0.45625382955483923,
-      "grad_norm": 1.111775279045105,
-      "learning_rate": 5.437340153452686e-06,
-      "loss": 0.1101,
-      "step": 2676
-    },
-    {
-      "epoch": 0.4564243279963769,
-      "grad_norm": 0.7632780075073242,
-      "learning_rate": 5.4356351236146636e-06,
-      "loss": 0.0655,
-      "step": 2677
-    },
-    {
-      "epoch": 0.4565948264379146,
-      "grad_norm": 1.4325367212295532,
-      "learning_rate": 5.4339300937766415e-06,
-      "loss": 0.1902,
-      "step": 2678
-    },
-    {
-      "epoch": 0.45676532487945226,
-      "grad_norm": 2.3953022956848145,
-      "learning_rate": 5.4322250639386195e-06,
-      "loss": 0.2011,
-      "step": 2679
-    },
-    {
-      "epoch": 0.45693582332099,
-      "grad_norm": 0.8128249049186707,
-      "learning_rate": 5.430520034100597e-06,
-      "loss": 0.1284,
-      "step": 2680
-    },
-    {
-      "epoch": 0.45710632176252763,
-      "grad_norm": 1.0645301342010498,
-      "learning_rate": 5.428815004262575e-06,
-      "loss": 0.1573,
-      "step": 2681
-    },
-    {
-      "epoch": 0.45727682020406535,
-      "grad_norm": 0.9294531345367432,
-      "learning_rate": 5.4271099744245525e-06,
-      "loss": 0.1008,
-      "step": 2682
-    },
-    {
-      "epoch": 0.457447318645603,
-      "grad_norm": 0.8018505573272705,
-      "learning_rate": 5.425404944586531e-06,
-      "loss": 0.0975,
-      "step": 2683
-    },
-    {
-      "epoch": 0.45761781708714067,
-      "grad_norm": 1.1697618961334229,
-      "learning_rate": 5.423699914748508e-06,
-      "loss": 0.1187,
-      "step": 2684
-    },
-    {
-      "epoch": 0.4577883155286784,
-      "grad_norm": 0.8472014665603638,
-      "learning_rate": 5.421994884910486e-06,
-      "loss": 0.1486,
-      "step": 2685
-    },
-    {
-      "epoch": 0.45795881397021604,
-      "grad_norm": 0.9629762768745422,
-      "learning_rate": 5.420289855072465e-06,
-      "loss": 0.1444,
-      "step": 2686
-    },
-    {
-      "epoch": 0.45812931241175375,
-      "grad_norm": 0.8507773876190186,
-      "learning_rate": 5.418584825234442e-06,
-      "loss": 0.1284,
-      "step": 2687
-    },
-    {
-      "epoch": 0.4582998108532914,
-      "grad_norm": 0.8529912829399109,
-      "learning_rate": 5.41687979539642e-06,
-      "loss": 0.115,
-      "step": 2688
-    },
-    {
-      "epoch": 0.4584703092948291,
-      "grad_norm": 0.6989152431488037,
-      "learning_rate": 5.415174765558397e-06,
-      "loss": 0.0956,
-      "step": 2689
-    },
-    {
-      "epoch": 0.4586408077363668,
-      "grad_norm": 1.049714207649231,
-      "learning_rate": 5.413469735720376e-06,
-      "loss": 0.157,
-      "step": 2690
-    },
-    {
-      "epoch": 0.45881130617790444,
-      "grad_norm": 1.126670002937317,
-      "learning_rate": 5.411764705882353e-06,
-      "loss": 0.1814,
-      "step": 2691
-    },
-    {
-      "epoch": 0.45898180461944216,
-      "grad_norm": 0.8889474868774414,
-      "learning_rate": 5.410059676044331e-06,
-      "loss": 0.0571,
-      "step": 2692
-    },
-    {
-      "epoch": 0.4591523030609798,
-      "grad_norm": 1.0496242046356201,
-      "learning_rate": 5.408354646206308e-06,
-      "loss": 0.1473,
-      "step": 2693
-    },
-    {
-      "epoch": 0.45932280150251753,
-      "grad_norm": 1.1733475923538208,
-      "learning_rate": 5.406649616368287e-06,
-      "loss": 0.1563,
-      "step": 2694
-    },
-    {
-      "epoch": 0.4594932999440552,
-      "grad_norm": 1.4026211500167847,
-      "learning_rate": 5.404944586530265e-06,
-      "loss": 0.2016,
-      "step": 2695
-    },
-    {
-      "epoch": 0.4596637983855929,
-      "grad_norm": 1.75924813747406,
-      "learning_rate": 5.403239556692242e-06,
-      "loss": 0.28,
-      "step": 2696
-    },
-    {
-      "epoch": 0.45983429682713056,
-      "grad_norm": 0.8990041613578796,
-      "learning_rate": 5.401534526854221e-06,
-      "loss": 0.1298,
-      "step": 2697
-    },
-    {
-      "epoch": 0.4600047952686682,
-      "grad_norm": 1.098027229309082,
-      "learning_rate": 5.399829497016198e-06,
-      "loss": 0.1223,
-      "step": 2698
-    },
-    {
-      "epoch": 0.46017529371020593,
-      "grad_norm": 1.0936179161071777,
-      "learning_rate": 5.398124467178176e-06,
-      "loss": 0.1047,
-      "step": 2699
-    },
-    {
-      "epoch": 0.4603457921517436,
-      "grad_norm": 0.7299230098724365,
-      "learning_rate": 5.396419437340154e-06,
-      "loss": 0.117,
-      "step": 2700
-    },
-    {
-      "epoch": 0.4603457921517436,
-      "eval_f1_score": 0.2802197802197802,
-      "eval_loss": 0.1476231813430786,
-      "eval_runtime": 182.6762,
-      "eval_samples_per_second": 54.742,
-      "eval_steps_per_second": 3.421,
-      "step": 2700
-    },
-    {
-      "epoch": 0.4605162905932813,
-      "grad_norm": 0.8915308713912964,
-      "learning_rate": 5.394714407502132e-06,
-      "loss": 0.1093,
-      "step": 2701
-    },
-    {
-      "epoch": 0.46068678903481897,
-      "grad_norm": 1.1240203380584717,
-      "learning_rate": 5.393009377664109e-06,
-      "loss": 0.1667,
-      "step": 2702
-    },
-    {
-      "epoch": 0.4608572874763567,
-      "grad_norm": 0.9414085149765015,
-      "learning_rate": 5.391304347826088e-06,
-      "loss": 0.1606,
-      "step": 2703
-    },
-    {
-      "epoch": 0.46102778591789434,
-      "grad_norm": 0.9260140657424927,
-      "learning_rate": 5.389599317988066e-06,
-      "loss": 0.0523,
-      "step": 2704
-    },
-    {
-      "epoch": 0.46119828435943205,
-      "grad_norm": 0.8883020877838135,
-      "learning_rate": 5.387894288150043e-06,
-      "loss": 0.0896,
-      "step": 2705
-    },
-    {
-      "epoch": 0.4613687828009697,
-      "grad_norm": 0.8151699304580688,
-      "learning_rate": 5.3861892583120215e-06,
-      "loss": 0.1237,
-      "step": 2706
-    },
-    {
-      "epoch": 0.46153928124250737,
-      "grad_norm": 0.612257182598114,
-      "learning_rate": 5.384484228473999e-06,
-      "loss": 0.052,
-      "step": 2707
-    },
-    {
-      "epoch": 0.4617097796840451,
-      "grad_norm": 0.8854296207427979,
-      "learning_rate": 5.3827791986359766e-06,
-      "loss": 0.1263,
-      "step": 2708
-    },
-    {
-      "epoch": 0.46188027812558274,
-      "grad_norm": 0.6729491949081421,
-      "learning_rate": 5.381074168797954e-06,
-      "loss": 0.1134,
-      "step": 2709
-    },
-    {
-      "epoch": 0.46205077656712046,
-      "grad_norm": 0.995721697807312,
-      "learning_rate": 5.3793691389599325e-06,
-      "loss": 0.1188,
-      "step": 2710
-    },
-    {
-      "epoch": 0.4622212750086581,
-      "grad_norm": 2.032381772994995,
-      "learning_rate": 5.37766410912191e-06,
-      "loss": 0.2624,
-      "step": 2711
-    },
-    {
-      "epoch": 0.46239177345019583,
-      "grad_norm": 0.7103998064994812,
-      "learning_rate": 5.3759590792838875e-06,
-      "loss": 0.0889,
-      "step": 2712
-    },
-    {
-      "epoch": 0.4625622718917335,
-      "grad_norm": 1.155537724494934,
-      "learning_rate": 5.374254049445866e-06,
-      "loss": 0.1714,
-      "step": 2713
-    },
-    {
-      "epoch": 0.46273277033327115,
-      "grad_norm": 1.4818527698516846,
-      "learning_rate": 5.372549019607843e-06,
-      "loss": 0.1816,
-      "step": 2714
-    },
-    {
-      "epoch": 0.46290326877480886,
-      "grad_norm": 0.5952757000923157,
-      "learning_rate": 5.370843989769821e-06,
-      "loss": 0.0669,
-      "step": 2715
-    },
-    {
-      "epoch": 0.4630737672163465,
-      "grad_norm": 0.9609639644622803,
-      "learning_rate": 5.369138959931799e-06,
-      "loss": 0.1793,
-      "step": 2716
-    },
-    {
-      "epoch": 0.46324426565788424,
-      "grad_norm": 1.3182636499404907,
-      "learning_rate": 5.367433930093777e-06,
-      "loss": 0.1719,
-      "step": 2717
-    },
-    {
-      "epoch": 0.4634147640994219,
-      "grad_norm": 0.7964705228805542,
-      "learning_rate": 5.365728900255754e-06,
-      "loss": 0.1311,
-      "step": 2718
-    },
-    {
-      "epoch": 0.4635852625409596,
-      "grad_norm": 0.6149096488952637,
-      "learning_rate": 5.364023870417733e-06,
-      "loss": 0.0855,
-      "step": 2719
-    },
-    {
-      "epoch": 0.46375576098249727,
-      "grad_norm": 1.1285103559494019,
-      "learning_rate": 5.362318840579711e-06,
-      "loss": 0.1798,
-      "step": 2720
-    },
-    {
-      "epoch": 0.4639262594240349,
-      "grad_norm": 0.6177775263786316,
-      "learning_rate": 5.360613810741688e-06,
-      "loss": 0.1305,
-      "step": 2721
-    },
-    {
-      "epoch": 0.46409675786557264,
-      "grad_norm": 1.480620265007019,
-      "learning_rate": 5.358908780903667e-06,
-      "loss": 0.1964,
-      "step": 2722
-    },
-    {
-      "epoch": 0.4642672563071103,
-      "grad_norm": 0.7413175702095032,
-      "learning_rate": 5.357203751065644e-06,
-      "loss": 0.1041,
-      "step": 2723
-    },
-    {
-      "epoch": 0.464437754748648,
-      "grad_norm": 0.8490052223205566,
-      "learning_rate": 5.355498721227622e-06,
-      "loss": 0.1172,
-      "step": 2724
-    },
-    {
-      "epoch": 0.46460825319018567,
-      "grad_norm": 0.7542960047721863,
-      "learning_rate": 5.353793691389599e-06,
-      "loss": 0.103,
-      "step": 2725
-    },
-    {
-      "epoch": 0.4647787516317234,
-      "grad_norm": 1.2431647777557373,
-      "learning_rate": 5.352088661551578e-06,
-      "loss": 0.2465,
-      "step": 2726
-    },
-    {
-      "epoch": 0.46494925007326104,
-      "grad_norm": 1.1575838327407837,
-      "learning_rate": 5.350383631713555e-06,
-      "loss": 0.1723,
-      "step": 2727
-    },
-    {
-      "epoch": 0.46511974851479876,
-      "grad_norm": 0.9394764304161072,
-      "learning_rate": 5.348678601875533e-06,
-      "loss": 0.0999,
-      "step": 2728
-    },
-    {
-      "epoch": 0.4652902469563364,
-      "grad_norm": 1.261762261390686,
-      "learning_rate": 5.346973572037512e-06,
-      "loss": 0.1649,
-      "step": 2729
-    },
-    {
-      "epoch": 0.4654607453978741,
-      "grad_norm": 0.990092933177948,
-      "learning_rate": 5.345268542199489e-06,
-      "loss": 0.1377,
-      "step": 2730
-    },
-    {
-      "epoch": 0.4656312438394118,
-      "grad_norm": 0.90677410364151,
-      "learning_rate": 5.343563512361467e-06,
-      "loss": 0.1167,
-      "step": 2731
-    },
-    {
-      "epoch": 0.46580174228094945,
-      "grad_norm": 0.9878215789794922,
-      "learning_rate": 5.341858482523444e-06,
-      "loss": 0.1888,
-      "step": 2732
-    },
-    {
-      "epoch": 0.46597224072248716,
-      "grad_norm": 0.8524580001831055,
-      "learning_rate": 5.340153452685423e-06,
-      "loss": 0.0938,
-      "step": 2733
-    },
-    {
-      "epoch": 0.4661427391640248,
-      "grad_norm": 1.0039318799972534,
-      "learning_rate": 5.3384484228474e-06,
-      "loss": 0.061,
-      "step": 2734
-    },
-    {
-      "epoch": 0.46631323760556254,
-      "grad_norm": 0.7102029323577881,
-      "learning_rate": 5.336743393009378e-06,
-      "loss": 0.0841,
-      "step": 2735
-    },
-    {
-      "epoch": 0.4664837360471002,
-      "grad_norm": 0.9649211168289185,
-      "learning_rate": 5.3350383631713566e-06,
-      "loss": 0.1681,
-      "step": 2736
-    },
-    {
-      "epoch": 0.46665423448863785,
-      "grad_norm": 1.3707305192947388,
-      "learning_rate": 5.333333333333334e-06,
-      "loss": 0.1815,
-      "step": 2737
-    },
-    {
-      "epoch": 0.46682473293017557,
-      "grad_norm": 1.0654339790344238,
-      "learning_rate": 5.331628303495312e-06,
-      "loss": 0.1352,
-      "step": 2738
-    },
-    {
-      "epoch": 0.4669952313717132,
-      "grad_norm": 0.7589823603630066,
-      "learning_rate": 5.3299232736572896e-06,
-      "loss": 0.0867,
-      "step": 2739
-    },
-    {
-      "epoch": 0.46716572981325094,
-      "grad_norm": 1.3560172319412231,
-      "learning_rate": 5.3282182438192675e-06,
-      "loss": 0.1688,
-      "step": 2740
-    },
-    {
-      "epoch": 0.4673362282547886,
-      "grad_norm": 1.1395232677459717,
-      "learning_rate": 5.326513213981245e-06,
-      "loss": 0.1397,
-      "step": 2741
-    },
-    {
-      "epoch": 0.4675067266963263,
-      "grad_norm": 1.3840816020965576,
-      "learning_rate": 5.324808184143223e-06,
-      "loss": 0.1409,
-      "step": 2742
-    },
-    {
-      "epoch": 0.46767722513786397,
-      "grad_norm": 0.9683536887168884,
-      "learning_rate": 5.3231031543052005e-06,
-      "loss": 0.0896,
-      "step": 2743
-    },
-    {
-      "epoch": 0.46784772357940163,
-      "grad_norm": 1.3210676908493042,
-      "learning_rate": 5.3213981244671784e-06,
-      "loss": 0.1244,
-      "step": 2744
-    },
-    {
-      "epoch": 0.46801822202093935,
-      "grad_norm": 0.9296742677688599,
-      "learning_rate": 5.319693094629157e-06,
-      "loss": 0.1292,
-      "step": 2745
-    },
-    {
-      "epoch": 0.468188720462477,
-      "grad_norm": 0.8764903545379639,
-      "learning_rate": 5.317988064791134e-06,
-      "loss": 0.1354,
-      "step": 2746
-    },
-    {
-      "epoch": 0.4683592189040147,
-      "grad_norm": 1.4558213949203491,
-      "learning_rate": 5.316283034953112e-06,
-      "loss": 0.2131,
-      "step": 2747
-    },
-    {
-      "epoch": 0.4685297173455524,
-      "grad_norm": 0.6826790571212769,
-      "learning_rate": 5.314578005115089e-06,
-      "loss": 0.0777,
-      "step": 2748
-    },
-    {
-      "epoch": 0.4687002157870901,
-      "grad_norm": 1.1762099266052246,
-      "learning_rate": 5.312872975277068e-06,
-      "loss": 0.0992,
-      "step": 2749
-    },
-    {
-      "epoch": 0.46887071422862775,
-      "grad_norm": 0.9933898448944092,
-      "learning_rate": 5.311167945439045e-06,
-      "loss": 0.0883,
-      "step": 2750
-    },
-    {
-      "epoch": 0.46904121267016546,
-      "grad_norm": 0.7758609056472778,
-      "learning_rate": 5.309462915601023e-06,
-      "loss": 0.0931,
-      "step": 2751
-    },
-    {
-      "epoch": 0.4692117111117031,
-      "grad_norm": 1.1159803867340088,
-      "learning_rate": 5.307757885763001e-06,
-      "loss": 0.1881,
-      "step": 2752
-    },
-    {
-      "epoch": 0.4693822095532408,
-      "grad_norm": 1.1558820009231567,
-      "learning_rate": 5.306052855924979e-06,
-      "loss": 0.1709,
-      "step": 2753
-    },
-    {
-      "epoch": 0.4695527079947785,
-      "grad_norm": 1.1873338222503662,
-      "learning_rate": 5.304347826086957e-06,
-      "loss": 0.1503,
-      "step": 2754
-    },
-    {
-      "epoch": 0.46972320643631615,
-      "grad_norm": 1.2646960020065308,
-      "learning_rate": 5.302642796248935e-06,
-      "loss": 0.0855,
-      "step": 2755
-    },
-    {
-      "epoch": 0.46989370487785387,
-      "grad_norm": 1.0196788311004639,
-      "learning_rate": 5.300937766410913e-06,
-      "loss": 0.1259,
-      "step": 2756
-    },
-    {
-      "epoch": 0.4700642033193915,
-      "grad_norm": 0.7964122891426086,
-      "learning_rate": 5.29923273657289e-06,
-      "loss": 0.0569,
-      "step": 2757
-    },
-    {
-      "epoch": 0.47023470176092924,
-      "grad_norm": 0.9101222157478333,
-      "learning_rate": 5.297527706734869e-06,
-      "loss": 0.1175,
-      "step": 2758
-    },
-    {
-      "epoch": 0.4704052002024669,
-      "grad_norm": 1.1495468616485596,
-      "learning_rate": 5.295822676896846e-06,
-      "loss": 0.1371,
-      "step": 2759
-    },
-    {
-      "epoch": 0.47057569864400456,
-      "grad_norm": 1.1147866249084473,
-      "learning_rate": 5.294117647058824e-06,
-      "loss": 0.1404,
-      "step": 2760
-    },
-    {
-      "epoch": 0.4707461970855423,
-      "grad_norm": 0.897382378578186,
-      "learning_rate": 5.292412617220803e-06,
-      "loss": 0.0796,
-      "step": 2761
-    },
-    {
-      "epoch": 0.47091669552707993,
-      "grad_norm": 1.366015911102295,
-      "learning_rate": 5.29070758738278e-06,
-      "loss": 0.1658,
-      "step": 2762
-    },
-    {
-      "epoch": 0.47108719396861765,
-      "grad_norm": 1.143560528755188,
-      "learning_rate": 5.289002557544758e-06,
-      "loss": 0.1538,
-      "step": 2763
-    },
-    {
-      "epoch": 0.4712576924101553,
-      "grad_norm": 0.8982717394828796,
-      "learning_rate": 5.287297527706735e-06,
-      "loss": 0.1504,
-      "step": 2764
-    },
-    {
-      "epoch": 0.471428190851693,
-      "grad_norm": 0.8663648366928101,
-      "learning_rate": 5.285592497868714e-06,
-      "loss": 0.0669,
-      "step": 2765
-    },
-    {
-      "epoch": 0.4715986892932307,
-      "grad_norm": 1.088739037513733,
-      "learning_rate": 5.283887468030691e-06,
-      "loss": 0.1432,
-      "step": 2766
-    },
-    {
-      "epoch": 0.47176918773476834,
-      "grad_norm": 1.2358800172805786,
-      "learning_rate": 5.282182438192669e-06,
-      "loss": 0.0928,
-      "step": 2767
-    },
-    {
-      "epoch": 0.47193968617630605,
-      "grad_norm": 1.063958764076233,
-      "learning_rate": 5.280477408354646e-06,
-      "loss": 0.1624,
-      "step": 2768
-    },
-    {
-      "epoch": 0.4721101846178437,
-      "grad_norm": 1.3645049333572388,
-      "learning_rate": 5.278772378516625e-06,
-      "loss": 0.1668,
-      "step": 2769
-    },
-    {
-      "epoch": 0.4722806830593814,
-      "grad_norm": 1.0899858474731445,
-      "learning_rate": 5.2770673486786025e-06,
-      "loss": 0.1343,
-      "step": 2770
-    },
-    {
-      "epoch": 0.4724511815009191,
-      "grad_norm": 1.2116706371307373,
-      "learning_rate": 5.27536231884058e-06,
-      "loss": 0.1919,
-      "step": 2771
-    },
-    {
-      "epoch": 0.4726216799424568,
-      "grad_norm": 1.0772473812103271,
-      "learning_rate": 5.2736572890025584e-06,
-      "loss": 0.1613,
-      "step": 2772
-    },
-    {
-      "epoch": 0.47279217838399445,
-      "grad_norm": 0.9527348279953003,
-      "learning_rate": 5.2719522591645355e-06,
-      "loss": 0.1292,
-      "step": 2773
-    },
-    {
-      "epoch": 0.47296267682553217,
-      "grad_norm": 1.2668654918670654,
-      "learning_rate": 5.2702472293265135e-06,
-      "loss": 0.2069,
-      "step": 2774
-    },
-    {
-      "epoch": 0.4731331752670698,
-      "grad_norm": 1.488873839378357,
-      "learning_rate": 5.2685421994884914e-06,
-      "loss": 0.1628,
-      "step": 2775
-    },
-    {
-      "epoch": 0.4733036737086075,
-      "grad_norm": 0.842333972454071,
-      "learning_rate": 5.266837169650469e-06,
-      "loss": 0.1079,
-      "step": 2776
-    },
-    {
-      "epoch": 0.4734741721501452,
-      "grad_norm": 0.8590933680534363,
-      "learning_rate": 5.2651321398124465e-06,
-      "loss": 0.1045,
-      "step": 2777
-    },
-    {
-      "epoch": 0.47364467059168286,
-      "grad_norm": 1.0166113376617432,
-      "learning_rate": 5.263427109974425e-06,
-      "loss": 0.1297,
-      "step": 2778
-    },
-    {
-      "epoch": 0.4738151690332206,
-      "grad_norm": 1.1054891347885132,
-      "learning_rate": 5.261722080136403e-06,
-      "loss": 0.0694,
-      "step": 2779
-    },
-    {
-      "epoch": 0.47398566747475823,
-      "grad_norm": 1.1980565786361694,
-      "learning_rate": 5.26001705029838e-06,
-      "loss": 0.1124,
-      "step": 2780
-    },
-    {
-      "epoch": 0.47415616591629595,
-      "grad_norm": 0.9477231502532959,
-      "learning_rate": 5.258312020460359e-06,
-      "loss": 0.1674,
-      "step": 2781
-    },
-    {
-      "epoch": 0.4743266643578336,
-      "grad_norm": 1.079034686088562,
-      "learning_rate": 5.256606990622336e-06,
-      "loss": 0.1674,
-      "step": 2782
-    },
-    {
-      "epoch": 0.47449716279937126,
-      "grad_norm": 1.0305124521255493,
-      "learning_rate": 5.254901960784314e-06,
-      "loss": 0.1413,
-      "step": 2783
-    },
-    {
-      "epoch": 0.474667661240909,
-      "grad_norm": 0.8747472763061523,
-      "learning_rate": 5.253196930946291e-06,
-      "loss": 0.1355,
-      "step": 2784
-    },
-    {
-      "epoch": 0.47483815968244664,
-      "grad_norm": 0.9253730177879333,
-      "learning_rate": 5.25149190110827e-06,
-      "loss": 0.1316,
-      "step": 2785
-    },
-    {
-      "epoch": 0.47500865812398435,
-      "grad_norm": 0.8443253636360168,
-      "learning_rate": 5.249786871270247e-06,
-      "loss": 0.1079,
-      "step": 2786
-    },
-    {
-      "epoch": 0.475179156565522,
-      "grad_norm": 1.0115752220153809,
-      "learning_rate": 5.248081841432225e-06,
-      "loss": 0.1634,
-      "step": 2787
-    },
-    {
-      "epoch": 0.4753496550070597,
-      "grad_norm": 0.876071572303772,
-      "learning_rate": 5.246376811594204e-06,
-      "loss": 0.09,
-      "step": 2788
-    },
-    {
-      "epoch": 0.4755201534485974,
-      "grad_norm": 0.644968569278717,
-      "learning_rate": 5.244671781756181e-06,
-      "loss": 0.0792,
-      "step": 2789
-    },
-    {
-      "epoch": 0.47569065189013504,
-      "grad_norm": 0.8539202213287354,
-      "learning_rate": 5.242966751918159e-06,
-      "loss": 0.11,
-      "step": 2790
-    },
-    {
-      "epoch": 0.47586115033167276,
-      "grad_norm": 1.7425650358200073,
-      "learning_rate": 5.241261722080137e-06,
-      "loss": 0.2118,
-      "step": 2791
-    },
-    {
-      "epoch": 0.4760316487732104,
-      "grad_norm": 1.3849416971206665,
-      "learning_rate": 5.239556692242115e-06,
-      "loss": 0.1975,
-      "step": 2792
-    },
-    {
-      "epoch": 0.47620214721474813,
-      "grad_norm": 0.8895875215530396,
-      "learning_rate": 5.237851662404092e-06,
-      "loss": 0.1633,
-      "step": 2793
-    },
-    {
-      "epoch": 0.4763726456562858,
-      "grad_norm": 0.7971814870834351,
-      "learning_rate": 5.236146632566071e-06,
-      "loss": 0.0991,
-      "step": 2794
-    },
-    {
-      "epoch": 0.4765431440978235,
-      "grad_norm": 1.0921452045440674,
-      "learning_rate": 5.234441602728049e-06,
-      "loss": 0.1087,
-      "step": 2795
-    },
-    {
-      "epoch": 0.47671364253936116,
-      "grad_norm": 1.0986230373382568,
-      "learning_rate": 5.232736572890026e-06,
-      "loss": 0.1827,
-      "step": 2796
-    },
-    {
-      "epoch": 0.4768841409808989,
-      "grad_norm": 1.2531445026397705,
-      "learning_rate": 5.231031543052005e-06,
-      "loss": 0.2311,
-      "step": 2797
-    },
-    {
-      "epoch": 0.47705463942243653,
-      "grad_norm": 0.633826732635498,
-      "learning_rate": 5.229326513213982e-06,
-      "loss": 0.0877,
-      "step": 2798
-    },
-    {
-      "epoch": 0.4772251378639742,
-      "grad_norm": 1.069523811340332,
-      "learning_rate": 5.22762148337596e-06,
-      "loss": 0.1876,
-      "step": 2799
-    },
-    {
-      "epoch": 0.4773956363055119,
-      "grad_norm": 0.5807836055755615,
-      "learning_rate": 5.225916453537937e-06,
-      "loss": 0.0568,
-      "step": 2800
-    },
-    {
-      "epoch": 0.4773956363055119,
-      "eval_f1_score": 0.2947976878612717,
-      "eval_loss": 0.14735326170921326,
-      "eval_runtime": 182.6683,
-      "eval_samples_per_second": 54.744,
-      "eval_steps_per_second": 3.422,
-      "step": 2800
-    },
-    {
-      "epoch": 0.47756613474704956,
-      "grad_norm": 1.1166603565216064,
-      "learning_rate": 5.2242114236999155e-06,
-      "loss": 0.2046,
-      "step": 2801
-    },
-    {
-      "epoch": 0.4777366331885873,
-      "grad_norm": 0.9348694086074829,
-      "learning_rate": 5.222506393861893e-06,
-      "loss": 0.0616,
-      "step": 2802
-    },
-    {
-      "epoch": 0.47790713163012494,
-      "grad_norm": 0.8302772045135498,
-      "learning_rate": 5.220801364023871e-06,
-      "loss": 0.1194,
-      "step": 2803
-    },
-    {
-      "epoch": 0.47807763007166265,
-      "grad_norm": 1.2290252447128296,
-      "learning_rate": 5.219096334185849e-06,
-      "loss": 0.1349,
-      "step": 2804
-    },
-    {
-      "epoch": 0.4782481285132003,
-      "grad_norm": 0.9503166079521179,
-      "learning_rate": 5.2173913043478265e-06,
-      "loss": 0.138,
-      "step": 2805
-    },
-    {
-      "epoch": 0.47841862695473797,
-      "grad_norm": 0.8801444172859192,
-      "learning_rate": 5.2156862745098044e-06,
-      "loss": 0.1008,
-      "step": 2806
-    },
-    {
-      "epoch": 0.4785891253962757,
-      "grad_norm": 1.240086555480957,
-      "learning_rate": 5.2139812446717815e-06,
-      "loss": 0.1937,
-      "step": 2807
-    },
-    {
-      "epoch": 0.47875962383781334,
-      "grad_norm": 0.7774300575256348,
-      "learning_rate": 5.21227621483376e-06,
-      "loss": 0.061,
-      "step": 2808
-    },
-    {
-      "epoch": 0.47893012227935106,
-      "grad_norm": 0.9553461670875549,
-      "learning_rate": 5.210571184995737e-06,
-      "loss": 0.0696,
-      "step": 2809
-    },
-    {
-      "epoch": 0.4791006207208887,
-      "grad_norm": 0.8926774263381958,
-      "learning_rate": 5.208866155157715e-06,
-      "loss": 0.1474,
-      "step": 2810
-    },
-    {
-      "epoch": 0.47927111916242643,
-      "grad_norm": 0.8952902555465698,
-      "learning_rate": 5.207161125319693e-06,
-      "loss": 0.1221,
-      "step": 2811
-    },
-    {
-      "epoch": 0.4794416176039641,
-      "grad_norm": 0.8415282964706421,
-      "learning_rate": 5.205456095481671e-06,
-      "loss": 0.0966,
-      "step": 2812
-    },
-    {
-      "epoch": 0.47961211604550175,
-      "grad_norm": 0.746725857257843,
-      "learning_rate": 5.203751065643649e-06,
-      "loss": 0.067,
-      "step": 2813
-    },
-    {
-      "epoch": 0.47978261448703946,
-      "grad_norm": 0.9865466952323914,
-      "learning_rate": 5.202046035805627e-06,
-      "loss": 0.1327,
-      "step": 2814
-    },
-    {
-      "epoch": 0.4799531129285771,
-      "grad_norm": 0.7144932150840759,
-      "learning_rate": 5.200341005967605e-06,
-      "loss": 0.1007,
-      "step": 2815
-    },
-    {
-      "epoch": 0.48012361137011483,
-      "grad_norm": 0.7237259745597839,
-      "learning_rate": 5.198635976129582e-06,
-      "loss": 0.1027,
-      "step": 2816
-    },
-    {
-      "epoch": 0.4802941098116525,
-      "grad_norm": 1.5335054397583008,
-      "learning_rate": 5.196930946291561e-06,
-      "loss": 0.1574,
-      "step": 2817
-    },
-    {
-      "epoch": 0.4804646082531902,
-      "grad_norm": 1.2245336771011353,
-      "learning_rate": 5.195225916453538e-06,
-      "loss": 0.136,
-      "step": 2818
-    },
-    {
-      "epoch": 0.48063510669472786,
-      "grad_norm": 1.8195768594741821,
-      "learning_rate": 5.193520886615516e-06,
-      "loss": 0.2907,
-      "step": 2819
-    },
-    {
-      "epoch": 0.4808056051362655,
-      "grad_norm": 1.5869253873825073,
-      "learning_rate": 5.191815856777495e-06,
-      "loss": 0.1904,
-      "step": 2820
-    },
-    {
-      "epoch": 0.48097610357780324,
-      "grad_norm": 1.2377763986587524,
-      "learning_rate": 5.190110826939472e-06,
-      "loss": 0.1379,
-      "step": 2821
-    },
-    {
-      "epoch": 0.4811466020193409,
-      "grad_norm": 1.3938157558441162,
-      "learning_rate": 5.18840579710145e-06,
-      "loss": 0.1687,
-      "step": 2822
-    },
-    {
-      "epoch": 0.4813171004608786,
-      "grad_norm": 0.8821825981140137,
-      "learning_rate": 5.186700767263427e-06,
-      "loss": 0.1219,
-      "step": 2823
-    },
-    {
-      "epoch": 0.48148759890241627,
-      "grad_norm": 1.9296631813049316,
-      "learning_rate": 5.184995737425406e-06,
-      "loss": 0.1452,
-      "step": 2824
-    },
-    {
-      "epoch": 0.481658097343954,
-      "grad_norm": 1.841987133026123,
-      "learning_rate": 5.183290707587383e-06,
-      "loss": 0.2546,
-      "step": 2825
-    },
-    {
-      "epoch": 0.48182859578549164,
-      "grad_norm": 0.8879536390304565,
-      "learning_rate": 5.181585677749361e-06,
-      "loss": 0.0895,
-      "step": 2826
-    },
-    {
-      "epoch": 0.48199909422702936,
-      "grad_norm": 1.2907721996307373,
-      "learning_rate": 5.179880647911339e-06,
-      "loss": 0.1423,
-      "step": 2827
-    },
-    {
-      "epoch": 0.482169592668567,
-      "grad_norm": 1.2071154117584229,
-      "learning_rate": 5.178175618073317e-06,
-      "loss": 0.1553,
-      "step": 2828
-    },
-    {
-      "epoch": 0.4823400911101047,
-      "grad_norm": 0.9301887154579163,
-      "learning_rate": 5.176470588235295e-06,
-      "loss": 0.1438,
-      "step": 2829
-    },
-    {
-      "epoch": 0.4825105895516424,
-      "grad_norm": 1.1748899221420288,
-      "learning_rate": 5.174765558397273e-06,
-      "loss": 0.1349,
-      "step": 2830
-    },
-    {
-      "epoch": 0.48268108799318005,
-      "grad_norm": 1.4293502569198608,
-      "learning_rate": 5.1730605285592506e-06,
-      "loss": 0.2146,
-      "step": 2831
-    },
-    {
-      "epoch": 0.48285158643471776,
-      "grad_norm": 1.4519188404083252,
-      "learning_rate": 5.171355498721228e-06,
-      "loss": 0.2329,
-      "step": 2832
-    },
-    {
-      "epoch": 0.4830220848762554,
-      "grad_norm": 1.1554458141326904,
-      "learning_rate": 5.1696504688832065e-06,
-      "loss": 0.186,
-      "step": 2833
-    },
-    {
-      "epoch": 0.48319258331779313,
-      "grad_norm": 0.9985054731369019,
-      "learning_rate": 5.1679454390451836e-06,
-      "loss": 0.1415,
-      "step": 2834
-    },
-    {
-      "epoch": 0.4833630817593308,
-      "grad_norm": 1.1093112230300903,
-      "learning_rate": 5.1662404092071615e-06,
-      "loss": 0.1449,
-      "step": 2835
-    },
-    {
-      "epoch": 0.48353358020086845,
-      "grad_norm": 1.0962409973144531,
-      "learning_rate": 5.164535379369139e-06,
-      "loss": 0.1308,
-      "step": 2836
-    },
-    {
-      "epoch": 0.48370407864240617,
-      "grad_norm": 0.9815444946289062,
-      "learning_rate": 5.162830349531117e-06,
-      "loss": 0.0833,
-      "step": 2837
-    },
-    {
-      "epoch": 0.4838745770839438,
-      "grad_norm": 0.8832153081893921,
-      "learning_rate": 5.161125319693095e-06,
-      "loss": 0.135,
-      "step": 2838
-    },
-    {
-      "epoch": 0.48404507552548154,
-      "grad_norm": 1.1795408725738525,
-      "learning_rate": 5.1594202898550725e-06,
-      "loss": 0.1517,
-      "step": 2839
-    },
-    {
-      "epoch": 0.4842155739670192,
-      "grad_norm": 0.7381297945976257,
-      "learning_rate": 5.157715260017051e-06,
-      "loss": 0.1156,
-      "step": 2840
-    },
-    {
-      "epoch": 0.4843860724085569,
-      "grad_norm": 0.8825168609619141,
-      "learning_rate": 5.156010230179028e-06,
-      "loss": 0.1181,
-      "step": 2841
-    },
-    {
-      "epoch": 0.48455657085009457,
-      "grad_norm": 0.8557326197624207,
-      "learning_rate": 5.154305200341006e-06,
-      "loss": 0.1357,
-      "step": 2842
-    },
-    {
-      "epoch": 0.48472706929163223,
-      "grad_norm": 1.8939331769943237,
-      "learning_rate": 5.152600170502983e-06,
-      "loss": 0.1252,
-      "step": 2843
-    },
-    {
-      "epoch": 0.48489756773316994,
-      "grad_norm": 0.8608604669570923,
-      "learning_rate": 5.150895140664962e-06,
-      "loss": 0.0843,
-      "step": 2844
-    },
-    {
-      "epoch": 0.4850680661747076,
-      "grad_norm": 0.7309491038322449,
-      "learning_rate": 5.14919011082694e-06,
-      "loss": 0.1124,
-      "step": 2845
-    },
-    {
-      "epoch": 0.4852385646162453,
-      "grad_norm": 1.0825344324111938,
-      "learning_rate": 5.147485080988917e-06,
-      "loss": 0.171,
-      "step": 2846
-    },
-    {
-      "epoch": 0.485409063057783,
-      "grad_norm": 0.9518781900405884,
-      "learning_rate": 5.145780051150896e-06,
-      "loss": 0.1327,
-      "step": 2847
-    },
-    {
-      "epoch": 0.4855795614993207,
-      "grad_norm": 0.7426133155822754,
-      "learning_rate": 5.144075021312873e-06,
-      "loss": 0.1078,
-      "step": 2848
-    },
-    {
-      "epoch": 0.48575005994085835,
-      "grad_norm": 1.0284055471420288,
-      "learning_rate": 5.142369991474851e-06,
-      "loss": 0.103,
-      "step": 2849
-    },
-    {
-      "epoch": 0.48592055838239606,
-      "grad_norm": 1.573033094406128,
-      "learning_rate": 5.140664961636829e-06,
-      "loss": 0.1213,
-      "step": 2850
-    },
-    {
-      "epoch": 0.4860910568239337,
-      "grad_norm": 1.4963144063949585,
-      "learning_rate": 5.138959931798807e-06,
-      "loss": 0.1755,
-      "step": 2851
-    },
-    {
-      "epoch": 0.4862615552654714,
-      "grad_norm": 1.6056444644927979,
-      "learning_rate": 5.137254901960784e-06,
-      "loss": 0.2098,
-      "step": 2852
-    },
-    {
-      "epoch": 0.4864320537070091,
-      "grad_norm": 1.5847902297973633,
-      "learning_rate": 5.135549872122763e-06,
-      "loss": 0.2086,
-      "step": 2853
-    },
-    {
-      "epoch": 0.48660255214854675,
-      "grad_norm": 1.0895462036132812,
-      "learning_rate": 5.133844842284741e-06,
-      "loss": 0.1435,
-      "step": 2854
-    },
-    {
-      "epoch": 0.48677305059008447,
-      "grad_norm": 0.9047597646713257,
-      "learning_rate": 5.132139812446718e-06,
-      "loss": 0.106,
-      "step": 2855
-    },
-    {
-      "epoch": 0.4869435490316221,
-      "grad_norm": 0.8072808384895325,
-      "learning_rate": 5.130434782608697e-06,
-      "loss": 0.103,
-      "step": 2856
-    },
-    {
-      "epoch": 0.48711404747315984,
-      "grad_norm": 1.1774007081985474,
-      "learning_rate": 5.128729752770674e-06,
-      "loss": 0.1697,
-      "step": 2857
-    },
-    {
-      "epoch": 0.4872845459146975,
-      "grad_norm": 1.1065585613250732,
-      "learning_rate": 5.127024722932652e-06,
-      "loss": 0.1599,
-      "step": 2858
-    },
-    {
-      "epoch": 0.48745504435623516,
-      "grad_norm": 1.0990142822265625,
-      "learning_rate": 5.125319693094629e-06,
-      "loss": 0.0708,
-      "step": 2859
-    },
-    {
-      "epoch": 0.48762554279777287,
-      "grad_norm": 2.216906785964966,
-      "learning_rate": 5.123614663256608e-06,
-      "loss": 0.1162,
-      "step": 2860
-    },
-    {
-      "epoch": 0.48779604123931053,
-      "grad_norm": 0.9633966684341431,
-      "learning_rate": 5.121909633418585e-06,
-      "loss": 0.0864,
-      "step": 2861
-    },
-    {
-      "epoch": 0.48796653968084824,
-      "grad_norm": 0.914915919303894,
-      "learning_rate": 5.120204603580563e-06,
-      "loss": 0.1159,
-      "step": 2862
-    },
-    {
-      "epoch": 0.4881370381223859,
-      "grad_norm": 0.8440526723861694,
-      "learning_rate": 5.1184995737425415e-06,
-      "loss": 0.1087,
-      "step": 2863
-    },
-    {
-      "epoch": 0.4883075365639236,
-      "grad_norm": 1.1910771131515503,
-      "learning_rate": 5.116794543904519e-06,
-      "loss": 0.1818,
-      "step": 2864
-    },
-    {
-      "epoch": 0.4884780350054613,
-      "grad_norm": 0.9938591122627258,
-      "learning_rate": 5.1150895140664966e-06,
-      "loss": 0.1431,
-      "step": 2865
-    },
-    {
-      "epoch": 0.48864853344699893,
-      "grad_norm": 1.0255413055419922,
-      "learning_rate": 5.1133844842284745e-06,
-      "loss": 0.1353,
-      "step": 2866
-    },
-    {
-      "epoch": 0.48881903188853665,
-      "grad_norm": 1.0184707641601562,
-      "learning_rate": 5.1116794543904525e-06,
-      "loss": 0.0948,
-      "step": 2867
-    },
-    {
-      "epoch": 0.4889895303300743,
-      "grad_norm": 0.8346421718597412,
-      "learning_rate": 5.1099744245524296e-06,
-      "loss": 0.1123,
-      "step": 2868
-    },
-    {
-      "epoch": 0.489160028771612,
-      "grad_norm": 1.4496159553527832,
-      "learning_rate": 5.108269394714408e-06,
-      "loss": 0.1994,
-      "step": 2869
-    },
-    {
-      "epoch": 0.4893305272131497,
-      "grad_norm": 1.0109056234359741,
-      "learning_rate": 5.106564364876386e-06,
-      "loss": 0.1317,
-      "step": 2870
-    },
-    {
-      "epoch": 0.4895010256546874,
-      "grad_norm": 1.2369807958602905,
-      "learning_rate": 5.104859335038363e-06,
-      "loss": 0.165,
-      "step": 2871
-    },
-    {
-      "epoch": 0.48967152409622505,
-      "grad_norm": 0.7144286036491394,
-      "learning_rate": 5.103154305200342e-06,
-      "loss": 0.0609,
-      "step": 2872
-    },
-    {
-      "epoch": 0.48984202253776277,
-      "grad_norm": 0.9330503940582275,
-      "learning_rate": 5.101449275362319e-06,
-      "loss": 0.1263,
-      "step": 2873
-    },
-    {
-      "epoch": 0.4900125209793004,
-      "grad_norm": 1.3567122220993042,
-      "learning_rate": 5.099744245524297e-06,
-      "loss": 0.2209,
-      "step": 2874
-    },
-    {
-      "epoch": 0.4901830194208381,
-      "grad_norm": 0.6284449100494385,
-      "learning_rate": 5.098039215686274e-06,
-      "loss": 0.0759,
-      "step": 2875
-    },
-    {
-      "epoch": 0.4903535178623758,
-      "grad_norm": 0.9885140061378479,
-      "learning_rate": 5.096334185848253e-06,
-      "loss": 0.1227,
-      "step": 2876
-    },
-    {
-      "epoch": 0.49052401630391346,
-      "grad_norm": 1.0801233053207397,
-      "learning_rate": 5.09462915601023e-06,
-      "loss": 0.1315,
-      "step": 2877
-    },
-    {
-      "epoch": 0.49069451474545117,
-      "grad_norm": 0.9521180391311646,
-      "learning_rate": 5.092924126172208e-06,
-      "loss": 0.0846,
-      "step": 2878
-    },
-    {
-      "epoch": 0.49086501318698883,
-      "grad_norm": 1.128046989440918,
-      "learning_rate": 5.091219096334187e-06,
-      "loss": 0.0865,
-      "step": 2879
-    },
-    {
-      "epoch": 0.49103551162852654,
-      "grad_norm": 0.8198946714401245,
-      "learning_rate": 5.089514066496164e-06,
-      "loss": 0.1146,
-      "step": 2880
-    },
-    {
-      "epoch": 0.4912060100700642,
-      "grad_norm": 1.3413338661193848,
-      "learning_rate": 5.087809036658142e-06,
-      "loss": 0.1639,
-      "step": 2881
-    },
-    {
-      "epoch": 0.49137650851160186,
-      "grad_norm": 1.301024317741394,
-      "learning_rate": 5.086104006820119e-06,
-      "loss": 0.1802,
-      "step": 2882
-    },
-    {
-      "epoch": 0.4915470069531396,
-      "grad_norm": 1.308898687362671,
-      "learning_rate": 5.084398976982098e-06,
-      "loss": 0.225,
-      "step": 2883
-    },
-    {
-      "epoch": 0.49171750539467723,
-      "grad_norm": 1.2367159128189087,
-      "learning_rate": 5.082693947144075e-06,
-      "loss": 0.1484,
-      "step": 2884
-    },
-    {
-      "epoch": 0.49188800383621495,
-      "grad_norm": 0.9292266964912415,
-      "learning_rate": 5.080988917306053e-06,
-      "loss": 0.1084,
-      "step": 2885
-    },
-    {
-      "epoch": 0.4920585022777526,
-      "grad_norm": 1.3165147304534912,
-      "learning_rate": 5.079283887468031e-06,
-      "loss": 0.2169,
-      "step": 2886
-    },
-    {
-      "epoch": 0.4922290007192903,
-      "grad_norm": 1.9130513668060303,
-      "learning_rate": 5.077578857630009e-06,
-      "loss": 0.2492,
-      "step": 2887
-    },
-    {
-      "epoch": 0.492399499160828,
-      "grad_norm": 0.8905692100524902,
-      "learning_rate": 5.075873827791987e-06,
-      "loss": 0.114,
-      "step": 2888
-    },
-    {
-      "epoch": 0.49256999760236564,
-      "grad_norm": 1.8215012550354004,
-      "learning_rate": 5.074168797953965e-06,
-      "loss": 0.1037,
-      "step": 2889
-    },
-    {
-      "epoch": 0.49274049604390335,
-      "grad_norm": 1.8712091445922852,
-      "learning_rate": 5.072463768115943e-06,
-      "loss": 0.1958,
-      "step": 2890
-    },
-    {
-      "epoch": 0.492910994485441,
-      "grad_norm": 0.8169884085655212,
-      "learning_rate": 5.07075873827792e-06,
-      "loss": 0.098,
-      "step": 2891
-    },
-    {
-      "epoch": 0.4930814929269787,
-      "grad_norm": 1.1399061679840088,
-      "learning_rate": 5.069053708439899e-06,
-      "loss": 0.1082,
-      "step": 2892
-    },
-    {
-      "epoch": 0.4932519913685164,
-      "grad_norm": 1.0081355571746826,
-      "learning_rate": 5.067348678601876e-06,
-      "loss": 0.0983,
-      "step": 2893
-    },
-    {
-      "epoch": 0.4934224898100541,
-      "grad_norm": 1.1947318315505981,
-      "learning_rate": 5.065643648763854e-06,
-      "loss": 0.1476,
-      "step": 2894
-    },
-    {
-      "epoch": 0.49359298825159176,
-      "grad_norm": 1.250249981880188,
-      "learning_rate": 5.0639386189258325e-06,
-      "loss": 0.1711,
-      "step": 2895
-    },
-    {
-      "epoch": 0.49376348669312947,
-      "grad_norm": 1.09302818775177,
-      "learning_rate": 5.0622335890878096e-06,
-      "loss": 0.1568,
-      "step": 2896
-    },
-    {
-      "epoch": 0.49393398513466713,
-      "grad_norm": 0.880730390548706,
-      "learning_rate": 5.0605285592497875e-06,
-      "loss": 0.1393,
-      "step": 2897
-    },
-    {
-      "epoch": 0.4941044835762048,
-      "grad_norm": 1.1225374937057495,
-      "learning_rate": 5.058823529411765e-06,
-      "loss": 0.1444,
-      "step": 2898
-    },
-    {
-      "epoch": 0.4942749820177425,
-      "grad_norm": 1.02659273147583,
-      "learning_rate": 5.057118499573743e-06,
-      "loss": 0.1071,
-      "step": 2899
-    },
-    {
-      "epoch": 0.49444548045928016,
-      "grad_norm": 0.6902765035629272,
-      "learning_rate": 5.0554134697357205e-06,
-      "loss": 0.0797,
-      "step": 2900
-    },
-    {
-      "epoch": 0.49444548045928016,
-      "eval_f1_score": 0.32275132275132273,
-      "eval_loss": 0.14594615995883942,
-      "eval_runtime": 182.6881,
-      "eval_samples_per_second": 54.738,
-      "eval_steps_per_second": 3.421,
-      "step": 2900
-    },
-    {
-      "epoch": 0.4946159789008179,
-      "grad_norm": 1.128650188446045,
-      "learning_rate": 5.0537084398976984e-06,
-      "loss": 0.1338,
-      "step": 2901
-    },
-    {
-      "epoch": 0.49478647734235554,
-      "grad_norm": 0.9671938419342041,
-      "learning_rate": 5.052003410059676e-06,
-      "loss": 0.1105,
-      "step": 2902
-    },
-    {
-      "epoch": 0.49495697578389325,
-      "grad_norm": 0.7639597058296204,
-      "learning_rate": 5.050298380221654e-06,
-      "loss": 0.0725,
-      "step": 2903
-    },
-    {
-      "epoch": 0.4951274742254309,
-      "grad_norm": 0.9767041206359863,
-      "learning_rate": 5.048593350383632e-06,
-      "loss": 0.0928,
-      "step": 2904
-    },
-    {
-      "epoch": 0.49529797266696857,
-      "grad_norm": 0.9453780651092529,
-      "learning_rate": 5.04688832054561e-06,
-      "loss": 0.1052,
-      "step": 2905
-    },
-    {
-      "epoch": 0.4954684711085063,
-      "grad_norm": 1.1028945446014404,
-      "learning_rate": 5.045183290707588e-06,
-      "loss": 0.0835,
-      "step": 2906
-    },
-    {
-      "epoch": 0.49563896955004394,
-      "grad_norm": 0.7256104350090027,
-      "learning_rate": 5.043478260869565e-06,
-      "loss": 0.0648,
-      "step": 2907
-    },
-    {
-      "epoch": 0.49580946799158165,
-      "grad_norm": 1.5962961912155151,
-      "learning_rate": 5.041773231031544e-06,
-      "loss": 0.1047,
-      "step": 2908
-    },
-    {
-      "epoch": 0.4959799664331193,
-      "grad_norm": 0.8366924524307251,
-      "learning_rate": 5.040068201193521e-06,
-      "loss": 0.1633,
-      "step": 2909
-    },
-    {
-      "epoch": 0.496150464874657,
-      "grad_norm": 0.9724977016448975,
-      "learning_rate": 5.038363171355499e-06,
-      "loss": 0.1732,
-      "step": 2910
-    },
-    {
-      "epoch": 0.4963209633161947,
-      "grad_norm": 1.251202940940857,
-      "learning_rate": 5.036658141517476e-06,
-      "loss": 0.1237,
-      "step": 2911
-    },
-    {
-      "epoch": 0.49649146175773234,
-      "grad_norm": 1.2317320108413696,
-      "learning_rate": 5.034953111679455e-06,
-      "loss": 0.1581,
-      "step": 2912
-    },
-    {
-      "epoch": 0.49666196019927006,
-      "grad_norm": 1.3477345705032349,
-      "learning_rate": 5.033248081841433e-06,
-      "loss": 0.1353,
-      "step": 2913
-    },
-    {
-      "epoch": 0.4968324586408077,
-      "grad_norm": 0.7249893546104431,
-      "learning_rate": 5.03154305200341e-06,
-      "loss": 0.0885,
-      "step": 2914
-    },
-    {
-      "epoch": 0.49700295708234543,
-      "grad_norm": 1.0673930644989014,
-      "learning_rate": 5.029838022165389e-06,
-      "loss": 0.1094,
-      "step": 2915
-    },
-    {
-      "epoch": 0.4971734555238831,
-      "grad_norm": 0.9129320383071899,
-      "learning_rate": 5.028132992327366e-06,
-      "loss": 0.0932,
-      "step": 2916
-    },
-    {
-      "epoch": 0.4973439539654208,
-      "grad_norm": 1.1062793731689453,
-      "learning_rate": 5.026427962489344e-06,
-      "loss": 0.2301,
-      "step": 2917
-    },
-    {
-      "epoch": 0.49751445240695846,
-      "grad_norm": 1.1716173887252808,
-      "learning_rate": 5.024722932651321e-06,
-      "loss": 0.1204,
-      "step": 2918
-    },
-    {
-      "epoch": 0.4976849508484962,
-      "grad_norm": 0.6853263974189758,
-      "learning_rate": 5.0230179028133e-06,
-      "loss": 0.0942,
-      "step": 2919
-    },
-    {
-      "epoch": 0.49785544929003384,
-      "grad_norm": 0.8545072078704834,
-      "learning_rate": 5.021312872975278e-06,
-      "loss": 0.1283,
-      "step": 2920
-    },
-    {
-      "epoch": 0.4980259477315715,
-      "grad_norm": 1.072073221206665,
-      "learning_rate": 5.019607843137255e-06,
-      "loss": 0.1356,
-      "step": 2921
-    },
-    {
-      "epoch": 0.4981964461731092,
-      "grad_norm": 1.1019538640975952,
-      "learning_rate": 5.017902813299234e-06,
-      "loss": 0.0915,
-      "step": 2922
-    },
-    {
-      "epoch": 0.49836694461464687,
-      "grad_norm": 1.5704419612884521,
-      "learning_rate": 5.016197783461211e-06,
-      "loss": 0.2405,
-      "step": 2923
-    },
-    {
-      "epoch": 0.4985374430561846,
-      "grad_norm": 1.09921395778656,
-      "learning_rate": 5.014492753623189e-06,
-      "loss": 0.1231,
-      "step": 2924
-    },
-    {
-      "epoch": 0.49870794149772224,
-      "grad_norm": 0.9660294055938721,
-      "learning_rate": 5.012787723785167e-06,
-      "loss": 0.0873,
-      "step": 2925
-    },
-    {
-      "epoch": 0.49887843993925995,
-      "grad_norm": 0.7957599759101868,
-      "learning_rate": 5.011082693947145e-06,
-      "loss": 0.097,
-      "step": 2926
-    },
-    {
-      "epoch": 0.4990489383807976,
-      "grad_norm": 1.182488203048706,
-      "learning_rate": 5.009377664109122e-06,
-      "loss": 0.1552,
-      "step": 2927
-    },
-    {
-      "epoch": 0.49921943682233527,
-      "grad_norm": 1.1175565719604492,
-      "learning_rate": 5.0076726342711005e-06,
-      "loss": 0.1212,
-      "step": 2928
-    },
-    {
-      "epoch": 0.499389935263873,
-      "grad_norm": 1.1008256673812866,
-      "learning_rate": 5.0059676044330784e-06,
-      "loss": 0.106,
-      "step": 2929
-    },
-    {
-      "epoch": 0.49956043370541064,
-      "grad_norm": 1.1122069358825684,
-      "learning_rate": 5.0042625745950555e-06,
-      "loss": 0.1277,
-      "step": 2930
-    },
-    {
-      "epoch": 0.49973093214694836,
-      "grad_norm": 0.8675044178962708,
-      "learning_rate": 5.002557544757034e-06,
-      "loss": 0.0812,
-      "step": 2931
-    },
-    {
-      "epoch": 0.499901430588486,
-      "grad_norm": 1.1391392946243286,
-      "learning_rate": 5.0008525149190114e-06,
-      "loss": 0.1313,
-      "step": 2932
-    },
-    {
-      "epoch": 0.5000719290300237,
-      "grad_norm": 1.2596222162246704,
-      "learning_rate": 4.999147485080989e-06,
-      "loss": 0.1338,
-      "step": 2933
-    },
-    {
-      "epoch": 0.5002424274715614,
-      "grad_norm": 1.128591775894165,
-      "learning_rate": 4.997442455242967e-06,
-      "loss": 0.1136,
-      "step": 2934
-    },
-    {
-      "epoch": 0.500412925913099,
-      "grad_norm": 1.137433648109436,
-      "learning_rate": 4.995737425404945e-06,
-      "loss": 0.1382,
-      "step": 2935
-    },
-    {
-      "epoch": 0.5005834243546368,
-      "grad_norm": 1.3177850246429443,
-      "learning_rate": 4.994032395566923e-06,
-      "loss": 0.1373,
-      "step": 2936
-    },
-    {
-      "epoch": 0.5007539227961745,
-      "grad_norm": 1.0512694120407104,
-      "learning_rate": 4.9923273657289e-06,
-      "loss": 0.1067,
-      "step": 2937
-    },
-    {
-      "epoch": 0.5009244212377121,
-      "grad_norm": 1.1488574743270874,
-      "learning_rate": 4.990622335890878e-06,
-      "loss": 0.1672,
-      "step": 2938
-    },
-    {
-      "epoch": 0.5010949196792498,
-      "grad_norm": 0.9468741416931152,
-      "learning_rate": 4.988917306052856e-06,
-      "loss": 0.096,
-      "step": 2939
-    },
-    {
-      "epoch": 0.5012654181207875,
-      "grad_norm": 1.8275294303894043,
-      "learning_rate": 4.987212276214834e-06,
-      "loss": 0.1998,
-      "step": 2940
-    },
-    {
-      "epoch": 0.5014359165623252,
-      "grad_norm": 0.7763018608093262,
-      "learning_rate": 4.985507246376812e-06,
-      "loss": 0.0775,
-      "step": 2941
-    },
-    {
-      "epoch": 0.5016064150038628,
-      "grad_norm": 0.9659168720245361,
-      "learning_rate": 4.98380221653879e-06,
-      "loss": 0.1099,
-      "step": 2942
-    },
-    {
-      "epoch": 0.5017769134454005,
-      "grad_norm": 1.334991455078125,
-      "learning_rate": 4.982097186700768e-06,
-      "loss": 0.15,
-      "step": 2943
-    },
-    {
-      "epoch": 0.5019474118869383,
-      "grad_norm": 1.7190552949905396,
-      "learning_rate": 4.980392156862746e-06,
-      "loss": 0.202,
-      "step": 2944
-    },
-    {
-      "epoch": 0.5021179103284759,
-      "grad_norm": 0.994452714920044,
-      "learning_rate": 4.978687127024723e-06,
-      "loss": 0.086,
-      "step": 2945
-    },
-    {
-      "epoch": 0.5022884087700136,
-      "grad_norm": 1.280116081237793,
-      "learning_rate": 4.976982097186701e-06,
-      "loss": 0.2435,
-      "step": 2946
-    },
-    {
-      "epoch": 0.5024589072115513,
-      "grad_norm": 1.1104083061218262,
-      "learning_rate": 4.975277067348679e-06,
-      "loss": 0.1192,
-      "step": 2947
-    },
-    {
-      "epoch": 0.502629405653089,
-      "grad_norm": 1.1849713325500488,
-      "learning_rate": 4.973572037510657e-06,
-      "loss": 0.1277,
-      "step": 2948
-    },
-    {
-      "epoch": 0.5027999040946266,
-      "grad_norm": 1.0629472732543945,
-      "learning_rate": 4.971867007672634e-06,
-      "loss": 0.0563,
-      "step": 2949
-    },
-    {
-      "epoch": 0.5029704025361643,
-      "grad_norm": 1.0110509395599365,
-      "learning_rate": 4.970161977834613e-06,
-      "loss": 0.1684,
-      "step": 2950
-    },
-    {
-      "epoch": 0.503140900977702,
-      "grad_norm": 1.2462400197982788,
-      "learning_rate": 4.968456947996591e-06,
-      "loss": 0.1531,
-      "step": 2951
-    },
-    {
-      "epoch": 0.5033113994192396,
-      "grad_norm": 1.3758798837661743,
-      "learning_rate": 4.966751918158569e-06,
-      "loss": 0.107,
-      "step": 2952
-    },
-    {
-      "epoch": 0.5034818978607773,
-      "grad_norm": 0.9632350206375122,
-      "learning_rate": 4.965046888320546e-06,
-      "loss": 0.1426,
-      "step": 2953
-    },
-    {
-      "epoch": 0.5036523963023151,
-      "grad_norm": 0.9977644681930542,
-      "learning_rate": 4.963341858482524e-06,
-      "loss": 0.1266,
-      "step": 2954
-    },
-    {
-      "epoch": 0.5038228947438528,
-      "grad_norm": 1.0415880680084229,
-      "learning_rate": 4.961636828644502e-06,
-      "loss": 0.1329,
-      "step": 2955
-    },
-    {
-      "epoch": 0.5039933931853904,
-      "grad_norm": 1.1172159910202026,
-      "learning_rate": 4.95993179880648e-06,
-      "loss": 0.1161,
-      "step": 2956
-    },
-    {
-      "epoch": 0.5041638916269281,
-      "grad_norm": 1.2988855838775635,
-      "learning_rate": 4.958226768968457e-06,
-      "loss": 0.1676,
-      "step": 2957
-    },
-    {
-      "epoch": 0.5043343900684658,
-      "grad_norm": 1.8825470209121704,
-      "learning_rate": 4.9565217391304355e-06,
-      "loss": 0.216,
-      "step": 2958
-    },
-    {
-      "epoch": 0.5045048885100034,
-      "grad_norm": 1.2938297986984253,
-      "learning_rate": 4.9548167092924135e-06,
-      "loss": 0.1605,
-      "step": 2959
-    },
-    {
-      "epoch": 0.5046753869515411,
-      "grad_norm": 1.0553568601608276,
-      "learning_rate": 4.953111679454391e-06,
-      "loss": 0.1528,
-      "step": 2960
-    },
-    {
-      "epoch": 0.5048458853930788,
-      "grad_norm": 0.8495917916297913,
-      "learning_rate": 4.9514066496163685e-06,
-      "loss": 0.1034,
-      "step": 2961
-    },
-    {
-      "epoch": 0.5050163838346166,
-      "grad_norm": 1.1607139110565186,
-      "learning_rate": 4.9497016197783465e-06,
-      "loss": 0.1775,
-      "step": 2962
-    },
-    {
-      "epoch": 0.5051868822761542,
-      "grad_norm": 0.98664391040802,
-      "learning_rate": 4.9479965899403244e-06,
-      "loss": 0.1173,
-      "step": 2963
-    },
-    {
-      "epoch": 0.5053573807176919,
-      "grad_norm": 1.0900707244873047,
-      "learning_rate": 4.946291560102302e-06,
-      "loss": 0.1118,
-      "step": 2964
-    },
-    {
-      "epoch": 0.5055278791592296,
-      "grad_norm": 1.483394980430603,
-      "learning_rate": 4.9445865302642795e-06,
-      "loss": 0.2319,
-      "step": 2965
-    },
-    {
-      "epoch": 0.5056983776007672,
-      "grad_norm": 1.5776678323745728,
-      "learning_rate": 4.942881500426258e-06,
-      "loss": 0.1674,
-      "step": 2966
-    },
-    {
-      "epoch": 0.5058688760423049,
-      "grad_norm": 0.9520050883293152,
-      "learning_rate": 4.941176470588236e-06,
-      "loss": 0.1343,
-      "step": 2967
-    },
-    {
-      "epoch": 0.5060393744838426,
-      "grad_norm": 0.9713250398635864,
-      "learning_rate": 4.939471440750213e-06,
-      "loss": 0.163,
-      "step": 2968
-    },
-    {
-      "epoch": 0.5062098729253803,
-      "grad_norm": 1.0562556982040405,
-      "learning_rate": 4.937766410912191e-06,
-      "loss": 0.1187,
-      "step": 2969
-    },
-    {
-      "epoch": 0.5063803713669179,
-      "grad_norm": 0.8522936105728149,
-      "learning_rate": 4.936061381074169e-06,
-      "loss": 0.1625,
-      "step": 2970
-    },
-    {
-      "epoch": 0.5065508698084557,
-      "grad_norm": 0.9506287574768066,
-      "learning_rate": 4.934356351236147e-06,
-      "loss": 0.0804,
-      "step": 2971
-    },
-    {
-      "epoch": 0.5067213682499934,
-      "grad_norm": 0.8114155530929565,
-      "learning_rate": 4.932651321398125e-06,
-      "loss": 0.0813,
-      "step": 2972
-    },
-    {
-      "epoch": 0.506891866691531,
-      "grad_norm": 0.8814626336097717,
-      "learning_rate": 4.930946291560102e-06,
-      "loss": 0.1039,
-      "step": 2973
-    },
-    {
-      "epoch": 0.5070623651330687,
-      "grad_norm": 0.7418640851974487,
-      "learning_rate": 4.92924126172208e-06,
-      "loss": 0.0652,
-      "step": 2974
-    },
-    {
-      "epoch": 0.5072328635746064,
-      "grad_norm": 0.748171865940094,
-      "learning_rate": 4.927536231884059e-06,
-      "loss": 0.0882,
-      "step": 2975
-    },
-    {
-      "epoch": 0.5074033620161441,
-      "grad_norm": 1.8255096673965454,
-      "learning_rate": 4.925831202046036e-06,
-      "loss": 0.237,
-      "step": 2976
-    },
-    {
-      "epoch": 0.5075738604576817,
-      "grad_norm": 1.0054211616516113,
-      "learning_rate": 4.924126172208014e-06,
-      "loss": 0.0576,
-      "step": 2977
-    },
-    {
-      "epoch": 0.5077443588992194,
-      "grad_norm": 1.8089407682418823,
-      "learning_rate": 4.922421142369992e-06,
-      "loss": 0.291,
-      "step": 2978
-    },
-    {
-      "epoch": 0.5079148573407571,
-      "grad_norm": 1.2715364694595337,
-      "learning_rate": 4.92071611253197e-06,
-      "loss": 0.1506,
-      "step": 2979
-    },
-    {
-      "epoch": 0.5080853557822949,
-      "grad_norm": 1.0056416988372803,
-      "learning_rate": 4.919011082693948e-06,
-      "loss": 0.1127,
-      "step": 2980
-    },
-    {
-      "epoch": 0.5082558542238325,
-      "grad_norm": 2.1583173274993896,
-      "learning_rate": 4.917306052855925e-06,
-      "loss": 0.2778,
-      "step": 2981
-    },
-    {
-      "epoch": 0.5084263526653702,
-      "grad_norm": 1.2089869976043701,
-      "learning_rate": 4.915601023017903e-06,
-      "loss": 0.1245,
-      "step": 2982
-    },
-    {
-      "epoch": 0.5085968511069079,
-      "grad_norm": 1.3445782661437988,
-      "learning_rate": 4.913895993179882e-06,
-      "loss": 0.1726,
-      "step": 2983
-    },
-    {
-      "epoch": 0.5087673495484455,
-      "grad_norm": 0.9058725237846375,
-      "learning_rate": 4.912190963341859e-06,
-      "loss": 0.1508,
-      "step": 2984
-    },
-    {
-      "epoch": 0.5089378479899832,
-      "grad_norm": 0.7840000987052917,
-      "learning_rate": 4.910485933503837e-06,
-      "loss": 0.0673,
-      "step": 2985
-    },
-    {
-      "epoch": 0.5091083464315209,
-      "grad_norm": 0.9684233069419861,
-      "learning_rate": 4.908780903665815e-06,
-      "loss": 0.1054,
-      "step": 2986
-    },
-    {
-      "epoch": 0.5092788448730586,
-      "grad_norm": 1.2829673290252686,
-      "learning_rate": 4.907075873827793e-06,
-      "loss": 0.1898,
-      "step": 2987
-    },
-    {
-      "epoch": 0.5094493433145962,
-      "grad_norm": 1.809422492980957,
-      "learning_rate": 4.90537084398977e-06,
-      "loss": 0.1678,
-      "step": 2988
-    },
-    {
-      "epoch": 0.509619841756134,
-      "grad_norm": 0.8061268925666809,
-      "learning_rate": 4.903665814151748e-06,
-      "loss": 0.1195,
-      "step": 2989
-    },
-    {
-      "epoch": 0.5097903401976717,
-      "grad_norm": 0.7315105199813843,
-      "learning_rate": 4.901960784313726e-06,
-      "loss": 0.1081,
-      "step": 2990
-    },
-    {
-      "epoch": 0.5099608386392093,
-      "grad_norm": 0.9622489213943481,
-      "learning_rate": 4.9002557544757036e-06,
-      "loss": 0.1252,
-      "step": 2991
-    },
-    {
-      "epoch": 0.510131337080747,
-      "grad_norm": 1.0779286623001099,
-      "learning_rate": 4.8985507246376815e-06,
-      "loss": 0.2076,
-      "step": 2992
-    },
-    {
-      "epoch": 0.5103018355222847,
-      "grad_norm": 0.7318807244300842,
-      "learning_rate": 4.8968456947996595e-06,
-      "loss": 0.0755,
-      "step": 2993
-    },
-    {
-      "epoch": 0.5104723339638224,
-      "grad_norm": 0.7845926880836487,
-      "learning_rate": 4.895140664961637e-06,
-      "loss": 0.1215,
-      "step": 2994
-    },
-    {
-      "epoch": 0.51064283240536,
-      "grad_norm": 0.8664889335632324,
-      "learning_rate": 4.893435635123615e-06,
-      "loss": 0.0785,
-      "step": 2995
-    },
-    {
-      "epoch": 0.5108133308468977,
-      "grad_norm": 1.0492292642593384,
-      "learning_rate": 4.8917306052855925e-06,
-      "loss": 0.0691,
-      "step": 2996
-    },
-    {
-      "epoch": 0.5109838292884354,
-      "grad_norm": 1.4236756563186646,
-      "learning_rate": 4.89002557544757e-06,
-      "loss": 0.1649,
-      "step": 2997
-    },
-    {
-      "epoch": 0.511154327729973,
-      "grad_norm": 0.8544982671737671,
-      "learning_rate": 4.888320545609548e-06,
-      "loss": 0.1691,
-      "step": 2998
-    },
-    {
-      "epoch": 0.5113248261715108,
-      "grad_norm": 1.7541369199752808,
-      "learning_rate": 4.886615515771526e-06,
-      "loss": 0.234,
-      "step": 2999
-    },
-    {
-      "epoch": 0.5114953246130485,
-      "grad_norm": 1.8814383745193481,
-      "learning_rate": 4.884910485933504e-06,
-      "loss": 0.2522,
-      "step": 3000
-    },
-    {
-      "epoch": 0.5114953246130485,
-      "eval_f1_score": 0.36180904522613067,
-      "eval_loss": 0.14497753977775574,
-      "eval_runtime": 182.6894,
-      "eval_samples_per_second": 54.738,
-      "eval_steps_per_second": 3.421,
-      "step": 3000
-    },
-    {
-      "epoch": 0.5116658230545862,
-      "grad_norm": 1.3755370378494263,
-      "learning_rate": 4.883205456095482e-06,
-      "loss": 0.1503,
-      "step": 3001
-    },
-    {
-      "epoch": 0.5118363214961238,
-      "grad_norm": 1.4907002449035645,
-      "learning_rate": 4.88150042625746e-06,
-      "loss": 0.1312,
-      "step": 3002
-    },
-    {
-      "epoch": 0.5120068199376615,
-      "grad_norm": 0.9052005410194397,
-      "learning_rate": 4.879795396419438e-06,
-      "loss": 0.1519,
-      "step": 3003
-    },
-    {
-      "epoch": 0.5121773183791992,
-      "grad_norm": 0.86555415391922,
-      "learning_rate": 4.878090366581415e-06,
-      "loss": 0.1146,
-      "step": 3004
-    },
-    {
-      "epoch": 0.5123478168207368,
-      "grad_norm": 1.230872631072998,
-      "learning_rate": 4.876385336743393e-06,
-      "loss": 0.1394,
-      "step": 3005
-    },
-    {
-      "epoch": 0.5125183152622745,
-      "grad_norm": 1.2168004512786865,
-      "learning_rate": 4.874680306905371e-06,
-      "loss": 0.2157,
-      "step": 3006
-    },
-    {
-      "epoch": 0.5126888137038123,
-      "grad_norm": 1.3998253345489502,
-      "learning_rate": 4.872975277067349e-06,
-      "loss": 0.1221,
-      "step": 3007
-    },
-    {
-      "epoch": 0.51285931214535,
-      "grad_norm": 1.2359734773635864,
-      "learning_rate": 4.871270247229327e-06,
-      "loss": 0.1726,
-      "step": 3008
-    },
-    {
-      "epoch": 0.5130298105868876,
-      "grad_norm": 1.387755274772644,
-      "learning_rate": 4.869565217391305e-06,
-      "loss": 0.2025,
-      "step": 3009
-    },
-    {
-      "epoch": 0.5132003090284253,
-      "grad_norm": 1.1594918966293335,
-      "learning_rate": 4.867860187553283e-06,
-      "loss": 0.1617,
-      "step": 3010
-    },
-    {
-      "epoch": 0.513370807469963,
-      "grad_norm": 0.8933932185173035,
-      "learning_rate": 4.866155157715261e-06,
-      "loss": 0.1102,
-      "step": 3011
-    },
-    {
-      "epoch": 0.5135413059115006,
-      "grad_norm": 1.0764583349227905,
-      "learning_rate": 4.864450127877238e-06,
-      "loss": 0.1151,
-      "step": 3012
-    },
-    {
-      "epoch": 0.5137118043530383,
-      "grad_norm": 0.9294535517692566,
-      "learning_rate": 4.862745098039216e-06,
-      "loss": 0.1442,
-      "step": 3013
-    },
-    {
-      "epoch": 0.513882302794576,
-      "grad_norm": 0.8940418362617493,
-      "learning_rate": 4.861040068201194e-06,
-      "loss": 0.1079,
-      "step": 3014
-    },
-    {
-      "epoch": 0.5140528012361137,
-      "grad_norm": 0.8228653073310852,
-      "learning_rate": 4.859335038363172e-06,
-      "loss": 0.0792,
-      "step": 3015
-    },
-    {
-      "epoch": 0.5142232996776513,
-      "grad_norm": 0.860935389995575,
-      "learning_rate": 4.85763000852515e-06,
-      "loss": 0.0919,
-      "step": 3016
-    },
-    {
-      "epoch": 0.5143937981191891,
-      "grad_norm": 0.9354767203330994,
-      "learning_rate": 4.855924978687128e-06,
-      "loss": 0.145,
-      "step": 3017
-    },
-    {
-      "epoch": 0.5145642965607268,
-      "grad_norm": 0.8232112526893616,
-      "learning_rate": 4.854219948849106e-06,
-      "loss": 0.1096,
-      "step": 3018
-    },
-    {
-      "epoch": 0.5147347950022644,
-      "grad_norm": 1.0975641012191772,
-      "learning_rate": 4.8525149190110836e-06,
-      "loss": 0.0981,
-      "step": 3019
-    },
-    {
-      "epoch": 0.5149052934438021,
-      "grad_norm": 1.0185004472732544,
-      "learning_rate": 4.850809889173061e-06,
-      "loss": 0.1236,
-      "step": 3020
-    },
-    {
-      "epoch": 0.5150757918853398,
-      "grad_norm": 0.8207215666770935,
-      "learning_rate": 4.849104859335039e-06,
-      "loss": 0.0702,
-      "step": 3021
-    },
-    {
-      "epoch": 0.5152462903268775,
-      "grad_norm": 1.1175296306610107,
-      "learning_rate": 4.8473998294970166e-06,
-      "loss": 0.1325,
-      "step": 3022
-    },
-    {
-      "epoch": 0.5154167887684151,
-      "grad_norm": 0.7806240916252136,
-      "learning_rate": 4.8456947996589945e-06,
-      "loss": 0.0861,
-      "step": 3023
-    },
-    {
-      "epoch": 0.5155872872099528,
-      "grad_norm": 1.094958782196045,
-      "learning_rate": 4.843989769820972e-06,
-      "loss": 0.0783,
-      "step": 3024
-    },
-    {
-      "epoch": 0.5157577856514906,
-      "grad_norm": 1.0804235935211182,
-      "learning_rate": 4.84228473998295e-06,
-      "loss": 0.1074,
-      "step": 3025
-    },
-    {
-      "epoch": 0.5159282840930283,
-      "grad_norm": 0.8760350346565247,
-      "learning_rate": 4.840579710144928e-06,
-      "loss": 0.0962,
-      "step": 3026
-    },
-    {
-      "epoch": 0.5160987825345659,
-      "grad_norm": 1.3372377157211304,
-      "learning_rate": 4.838874680306906e-06,
-      "loss": 0.1071,
-      "step": 3027
-    },
-    {
-      "epoch": 0.5162692809761036,
-      "grad_norm": 1.0021148920059204,
-      "learning_rate": 4.837169650468883e-06,
-      "loss": 0.119,
-      "step": 3028
-    },
-    {
-      "epoch": 0.5164397794176413,
-      "grad_norm": 1.2906092405319214,
-      "learning_rate": 4.835464620630861e-06,
-      "loss": 0.1074,
-      "step": 3029
-    },
-    {
-      "epoch": 0.5166102778591789,
-      "grad_norm": 1.09266197681427,
-      "learning_rate": 4.833759590792839e-06,
-      "loss": 0.1086,
-      "step": 3030
-    },
-    {
-      "epoch": 0.5167807763007166,
-      "grad_norm": 1.7703126668930054,
-      "learning_rate": 4.832054560954817e-06,
-      "loss": 0.2657,
-      "step": 3031
-    },
-    {
-      "epoch": 0.5169512747422543,
-      "grad_norm": 0.8526506423950195,
-      "learning_rate": 4.830349531116794e-06,
-      "loss": 0.0627,
-      "step": 3032
-    },
-    {
-      "epoch": 0.517121773183792,
-      "grad_norm": 1.1537318229675293,
-      "learning_rate": 4.828644501278773e-06,
-      "loss": 0.118,
-      "step": 3033
-    },
-    {
-      "epoch": 0.5172922716253296,
-      "grad_norm": 1.1283197402954102,
-      "learning_rate": 4.826939471440751e-06,
-      "loss": 0.0813,
-      "step": 3034
-    },
-    {
-      "epoch": 0.5174627700668674,
-      "grad_norm": 0.9512043595314026,
-      "learning_rate": 4.825234441602728e-06,
-      "loss": 0.0994,
-      "step": 3035
-    },
-    {
-      "epoch": 0.5176332685084051,
-      "grad_norm": 1.416027545928955,
-      "learning_rate": 4.823529411764706e-06,
-      "loss": 0.1292,
-      "step": 3036
-    },
-    {
-      "epoch": 0.5178037669499427,
-      "grad_norm": 1.2295200824737549,
-      "learning_rate": 4.821824381926684e-06,
-      "loss": 0.1406,
-      "step": 3037
-    },
-    {
-      "epoch": 0.5179742653914804,
-      "grad_norm": 1.3271145820617676,
-      "learning_rate": 4.820119352088662e-06,
-      "loss": 0.1723,
-      "step": 3038
-    },
-    {
-      "epoch": 0.5181447638330181,
-      "grad_norm": 0.9563559293746948,
-      "learning_rate": 4.81841432225064e-06,
-      "loss": 0.0892,
-      "step": 3039
-    },
-    {
-      "epoch": 0.5183152622745558,
-      "grad_norm": 1.0865267515182495,
-      "learning_rate": 4.816709292412617e-06,
-      "loss": 0.1565,
-      "step": 3040
-    },
-    {
-      "epoch": 0.5184857607160934,
-      "grad_norm": 0.8583990931510925,
-      "learning_rate": 4.815004262574595e-06,
-      "loss": 0.0959,
-      "step": 3041
-    },
-    {
-      "epoch": 0.5186562591576311,
-      "grad_norm": 1.1742157936096191,
-      "learning_rate": 4.813299232736574e-06,
-      "loss": 0.1373,
-      "step": 3042
-    },
-    {
-      "epoch": 0.5188267575991689,
-      "grad_norm": 0.895906925201416,
-      "learning_rate": 4.811594202898551e-06,
-      "loss": 0.0875,
-      "step": 3043
-    },
-    {
-      "epoch": 0.5189972560407065,
-      "grad_norm": 0.9671127200126648,
-      "learning_rate": 4.809889173060529e-06,
-      "loss": 0.1015,
-      "step": 3044
-    },
-    {
-      "epoch": 0.5191677544822442,
-      "grad_norm": 1.157288670539856,
-      "learning_rate": 4.808184143222507e-06,
-      "loss": 0.1169,
-      "step": 3045
-    },
-    {
-      "epoch": 0.5193382529237819,
-      "grad_norm": 1.1225751638412476,
-      "learning_rate": 4.806479113384485e-06,
-      "loss": 0.1367,
-      "step": 3046
-    },
-    {
-      "epoch": 0.5195087513653196,
-      "grad_norm": 0.8157687187194824,
-      "learning_rate": 4.804774083546463e-06,
-      "loss": 0.0866,
-      "step": 3047
-    },
-    {
-      "epoch": 0.5196792498068572,
-      "grad_norm": 0.908694863319397,
-      "learning_rate": 4.80306905370844e-06,
-      "loss": 0.0787,
-      "step": 3048
-    },
-    {
-      "epoch": 0.5198497482483949,
-      "grad_norm": 1.340047836303711,
-      "learning_rate": 4.801364023870418e-06,
-      "loss": 0.1833,
-      "step": 3049
-    },
-    {
-      "epoch": 0.5200202466899326,
-      "grad_norm": 0.9516338109970093,
-      "learning_rate": 4.7996589940323966e-06,
-      "loss": 0.159,
-      "step": 3050
-    },
-    {
-      "epoch": 0.5201907451314702,
-      "grad_norm": 0.9492541551589966,
-      "learning_rate": 4.797953964194374e-06,
-      "loss": 0.1062,
-      "step": 3051
-    },
-    {
-      "epoch": 0.520361243573008,
-      "grad_norm": 0.9833723306655884,
-      "learning_rate": 4.796248934356352e-06,
-      "loss": 0.137,
-      "step": 3052
-    },
-    {
-      "epoch": 0.5205317420145457,
-      "grad_norm": 0.9533904790878296,
-      "learning_rate": 4.7945439045183296e-06,
-      "loss": 0.1134,
-      "step": 3053
-    },
-    {
-      "epoch": 0.5207022404560834,
-      "grad_norm": 1.4486286640167236,
-      "learning_rate": 4.7928388746803075e-06,
-      "loss": 0.193,
-      "step": 3054
-    },
-    {
-      "epoch": 0.520872738897621,
-      "grad_norm": 1.293747901916504,
-      "learning_rate": 4.7911338448422854e-06,
-      "loss": 0.0931,
-      "step": 3055
-    },
-    {
-      "epoch": 0.5210432373391587,
-      "grad_norm": 0.9425202012062073,
-      "learning_rate": 4.7894288150042626e-06,
-      "loss": 0.0833,
-      "step": 3056
-    },
-    {
-      "epoch": 0.5212137357806964,
-      "grad_norm": 1.0497663021087646,
-      "learning_rate": 4.7877237851662405e-06,
-      "loss": 0.1519,
-      "step": 3057
-    },
-    {
-      "epoch": 0.521384234222234,
-      "grad_norm": 1.0354875326156616,
-      "learning_rate": 4.786018755328219e-06,
-      "loss": 0.0838,
-      "step": 3058
-    },
-    {
-      "epoch": 0.5215547326637717,
-      "grad_norm": 1.1730378866195679,
-      "learning_rate": 4.784313725490196e-06,
-      "loss": 0.151,
-      "step": 3059
-    },
-    {
-      "epoch": 0.5217252311053094,
-      "grad_norm": 2.1272621154785156,
-      "learning_rate": 4.782608695652174e-06,
-      "loss": 0.2293,
-      "step": 3060
-    },
-    {
-      "epoch": 0.5218957295468472,
-      "grad_norm": 1.4439617395401,
-      "learning_rate": 4.780903665814152e-06,
-      "loss": 0.1704,
-      "step": 3061
-    },
-    {
-      "epoch": 0.5220662279883848,
-      "grad_norm": 1.2619868516921997,
-      "learning_rate": 4.77919863597613e-06,
-      "loss": 0.1558,
-      "step": 3062
-    },
-    {
-      "epoch": 0.5222367264299225,
-      "grad_norm": 2.5501723289489746,
-      "learning_rate": 4.777493606138107e-06,
-      "loss": 0.2912,
-      "step": 3063
-    },
-    {
-      "epoch": 0.5224072248714602,
-      "grad_norm": 1.7127344608306885,
-      "learning_rate": 4.775788576300085e-06,
-      "loss": 0.2276,
-      "step": 3064
-    },
-    {
-      "epoch": 0.5225777233129978,
-      "grad_norm": 0.9259654879570007,
-      "learning_rate": 4.774083546462063e-06,
-      "loss": 0.1232,
-      "step": 3065
-    },
-    {
-      "epoch": 0.5227482217545355,
-      "grad_norm": 1.1065436601638794,
-      "learning_rate": 4.772378516624041e-06,
-      "loss": 0.1829,
-      "step": 3066
-    },
-    {
-      "epoch": 0.5229187201960732,
-      "grad_norm": 1.0913331508636475,
-      "learning_rate": 4.770673486786019e-06,
-      "loss": 0.1666,
-      "step": 3067
-    },
-    {
-      "epoch": 0.5230892186376109,
-      "grad_norm": 1.3781788349151611,
-      "learning_rate": 4.768968456947997e-06,
-      "loss": 0.1954,
-      "step": 3068
-    },
-    {
-      "epoch": 0.5232597170791485,
-      "grad_norm": 1.0182466506958008,
-      "learning_rate": 4.767263427109975e-06,
-      "loss": 0.1347,
-      "step": 3069
-    },
-    {
-      "epoch": 0.5234302155206862,
-      "grad_norm": 1.5784564018249512,
-      "learning_rate": 4.765558397271953e-06,
-      "loss": 0.1382,
-      "step": 3070
-    },
-    {
-      "epoch": 0.523600713962224,
-      "grad_norm": 1.2984943389892578,
-      "learning_rate": 4.76385336743393e-06,
-      "loss": 0.0943,
-      "step": 3071
-    },
-    {
-      "epoch": 0.5237712124037617,
-      "grad_norm": 1.1550954580307007,
-      "learning_rate": 4.762148337595908e-06,
-      "loss": 0.1466,
-      "step": 3072
-    },
-    {
-      "epoch": 0.5239417108452993,
-      "grad_norm": 1.4043045043945312,
-      "learning_rate": 4.760443307757886e-06,
-      "loss": 0.1147,
-      "step": 3073
-    },
-    {
-      "epoch": 0.524112209286837,
-      "grad_norm": 1.1770408153533936,
-      "learning_rate": 4.758738277919864e-06,
-      "loss": 0.168,
-      "step": 3074
-    },
-    {
-      "epoch": 0.5242827077283747,
-      "grad_norm": 0.9252126812934875,
-      "learning_rate": 4.757033248081842e-06,
-      "loss": 0.153,
-      "step": 3075
-    },
-    {
-      "epoch": 0.5244532061699123,
-      "grad_norm": 1.0992802381515503,
-      "learning_rate": 4.75532821824382e-06,
-      "loss": 0.1047,
-      "step": 3076
-    },
-    {
-      "epoch": 0.52462370461145,
-      "grad_norm": 1.1755380630493164,
-      "learning_rate": 4.753623188405798e-06,
-      "loss": 0.0625,
-      "step": 3077
-    },
-    {
-      "epoch": 0.5247942030529877,
-      "grad_norm": 1.2509602308273315,
-      "learning_rate": 4.751918158567776e-06,
-      "loss": 0.179,
-      "step": 3078
-    },
-    {
-      "epoch": 0.5249647014945255,
-      "grad_norm": 1.23047935962677,
-      "learning_rate": 4.750213128729753e-06,
-      "loss": 0.1415,
-      "step": 3079
-    },
-    {
-      "epoch": 0.5251351999360631,
-      "grad_norm": 1.2384694814682007,
-      "learning_rate": 4.748508098891731e-06,
-      "loss": 0.166,
-      "step": 3080
-    },
-    {
-      "epoch": 0.5253056983776008,
-      "grad_norm": 0.8851529359817505,
-      "learning_rate": 4.746803069053709e-06,
-      "loss": 0.105,
-      "step": 3081
-    },
-    {
-      "epoch": 0.5254761968191385,
-      "grad_norm": 0.9816237688064575,
-      "learning_rate": 4.745098039215687e-06,
-      "loss": 0.1696,
-      "step": 3082
-    },
-    {
-      "epoch": 0.5256466952606761,
-      "grad_norm": 1.4306472539901733,
-      "learning_rate": 4.743393009377665e-06,
-      "loss": 0.1873,
-      "step": 3083
-    },
-    {
-      "epoch": 0.5258171937022138,
-      "grad_norm": 0.8195698857307434,
-      "learning_rate": 4.7416879795396425e-06,
-      "loss": 0.1025,
-      "step": 3084
-    },
-    {
-      "epoch": 0.5259876921437515,
-      "grad_norm": 0.8986068367958069,
-      "learning_rate": 4.7399829497016205e-06,
-      "loss": 0.1152,
-      "step": 3085
-    },
-    {
-      "epoch": 0.5261581905852892,
-      "grad_norm": 1.3748445510864258,
-      "learning_rate": 4.7382779198635984e-06,
-      "loss": 0.1902,
-      "step": 3086
-    },
-    {
-      "epoch": 0.5263286890268268,
-      "grad_norm": 0.8912947773933411,
-      "learning_rate": 4.7365728900255755e-06,
-      "loss": 0.0736,
-      "step": 3087
-    },
-    {
-      "epoch": 0.5264991874683645,
-      "grad_norm": 1.600932002067566,
-      "learning_rate": 4.7348678601875535e-06,
-      "loss": 0.2012,
-      "step": 3088
-    },
-    {
-      "epoch": 0.5266696859099023,
-      "grad_norm": 1.0017476081848145,
-      "learning_rate": 4.7331628303495314e-06,
-      "loss": 0.1378,
-      "step": 3089
-    },
-    {
-      "epoch": 0.5268401843514399,
-      "grad_norm": 0.7620334029197693,
-      "learning_rate": 4.731457800511509e-06,
-      "loss": 0.0362,
-      "step": 3090
-    },
-    {
-      "epoch": 0.5270106827929776,
-      "grad_norm": 1.2506636381149292,
-      "learning_rate": 4.729752770673487e-06,
-      "loss": 0.1245,
-      "step": 3091
-    },
-    {
-      "epoch": 0.5271811812345153,
-      "grad_norm": 1.176638126373291,
-      "learning_rate": 4.728047740835465e-06,
-      "loss": 0.1591,
-      "step": 3092
-    },
-    {
-      "epoch": 0.527351679676053,
-      "grad_norm": 0.878821611404419,
-      "learning_rate": 4.726342710997443e-06,
-      "loss": 0.0863,
-      "step": 3093
-    },
-    {
-      "epoch": 0.5275221781175906,
-      "grad_norm": 1.617058515548706,
-      "learning_rate": 4.724637681159421e-06,
-      "loss": 0.1962,
-      "step": 3094
-    },
-    {
-      "epoch": 0.5276926765591283,
-      "grad_norm": 0.9386082887649536,
-      "learning_rate": 4.722932651321398e-06,
-      "loss": 0.0925,
-      "step": 3095
-    },
-    {
-      "epoch": 0.527863175000666,
-      "grad_norm": 1.0812885761260986,
-      "learning_rate": 4.721227621483376e-06,
-      "loss": 0.0651,
-      "step": 3096
-    },
-    {
-      "epoch": 0.5280336734422036,
-      "grad_norm": 0.925205409526825,
-      "learning_rate": 4.719522591645354e-06,
-      "loss": 0.0984,
-      "step": 3097
-    },
-    {
-      "epoch": 0.5282041718837414,
-      "grad_norm": 1.1007323265075684,
-      "learning_rate": 4.717817561807332e-06,
-      "loss": 0.1695,
-      "step": 3098
-    },
-    {
-      "epoch": 0.5283746703252791,
-      "grad_norm": 0.8048510551452637,
-      "learning_rate": 4.716112531969309e-06,
-      "loss": 0.0666,
-      "step": 3099
-    },
-    {
-      "epoch": 0.5285451687668168,
-      "grad_norm": 0.813273549079895,
-      "learning_rate": 4.714407502131288e-06,
-      "loss": 0.0776,
-      "step": 3100
-    },
-    {
-      "epoch": 0.5285451687668168,
-      "eval_f1_score": 0.3657957244655582,
-      "eval_loss": 0.1428575962781906,
-      "eval_runtime": 182.6607,
-      "eval_samples_per_second": 54.746,
-      "eval_steps_per_second": 3.422,
-      "step": 3100
-    },
-    {
-      "epoch": 0.5287156672083544,
-      "grad_norm": 1.064518928527832,
-      "learning_rate": 4.712702472293266e-06,
-      "loss": 0.1235,
-      "step": 3101
-    },
-    {
-      "epoch": 0.5288861656498921,
-      "grad_norm": 1.4153354167938232,
-      "learning_rate": 4.710997442455244e-06,
-      "loss": 0.2323,
-      "step": 3102
-    },
-    {
-      "epoch": 0.5290566640914298,
-      "grad_norm": 1.1358619928359985,
-      "learning_rate": 4.709292412617221e-06,
-      "loss": 0.1239,
-      "step": 3103
-    },
-    {
-      "epoch": 0.5292271625329674,
-      "grad_norm": 1.07460618019104,
-      "learning_rate": 4.707587382779199e-06,
-      "loss": 0.1017,
-      "step": 3104
-    },
-    {
-      "epoch": 0.5293976609745051,
-      "grad_norm": 1.2761073112487793,
-      "learning_rate": 4.705882352941177e-06,
-      "loss": 0.1273,
-      "step": 3105
-    },
-    {
-      "epoch": 0.5295681594160428,
-      "grad_norm": 1.2396087646484375,
-      "learning_rate": 4.704177323103155e-06,
-      "loss": 0.1489,
-      "step": 3106
-    },
-    {
-      "epoch": 0.5297386578575806,
-      "grad_norm": 1.3469845056533813,
-      "learning_rate": 4.702472293265132e-06,
-      "loss": 0.1379,
-      "step": 3107
-    },
-    {
-      "epoch": 0.5299091562991182,
-      "grad_norm": 0.8891554474830627,
-      "learning_rate": 4.70076726342711e-06,
-      "loss": 0.0653,
-      "step": 3108
-    },
-    {
-      "epoch": 0.5300796547406559,
-      "grad_norm": 1.4598875045776367,
-      "learning_rate": 4.699062233589089e-06,
-      "loss": 0.2002,
-      "step": 3109
-    },
-    {
-      "epoch": 0.5302501531821936,
-      "grad_norm": 0.964966893196106,
-      "learning_rate": 4.697357203751066e-06,
-      "loss": 0.1214,
-      "step": 3110
-    },
-    {
-      "epoch": 0.5304206516237312,
-      "grad_norm": 0.827178418636322,
-      "learning_rate": 4.695652173913044e-06,
-      "loss": 0.0768,
-      "step": 3111
-    },
-    {
-      "epoch": 0.5305911500652689,
-      "grad_norm": 1.1814799308776855,
-      "learning_rate": 4.693947144075022e-06,
-      "loss": 0.1719,
-      "step": 3112
-    },
-    {
-      "epoch": 0.5307616485068066,
-      "grad_norm": 1.0025556087493896,
-      "learning_rate": 4.692242114237e-06,
-      "loss": 0.1409,
-      "step": 3113
-    },
-    {
-      "epoch": 0.5309321469483443,
-      "grad_norm": 0.6644530892372131,
-      "learning_rate": 4.690537084398978e-06,
-      "loss": 0.063,
-      "step": 3114
-    },
-    {
-      "epoch": 0.5311026453898819,
-      "grad_norm": 1.0593664646148682,
-      "learning_rate": 4.688832054560955e-06,
-      "loss": 0.1329,
-      "step": 3115
-    },
-    {
-      "epoch": 0.5312731438314197,
-      "grad_norm": 1.1395896673202515,
-      "learning_rate": 4.687127024722933e-06,
-      "loss": 0.114,
-      "step": 3116
-    },
-    {
-      "epoch": 0.5314436422729574,
-      "grad_norm": 1.1389448642730713,
-      "learning_rate": 4.6854219948849114e-06,
-      "loss": 0.0903,
-      "step": 3117
-    },
-    {
-      "epoch": 0.5316141407144951,
-      "grad_norm": 0.9166980981826782,
-      "learning_rate": 4.6837169650468885e-06,
-      "loss": 0.1301,
-      "step": 3118
-    },
-    {
-      "epoch": 0.5317846391560327,
-      "grad_norm": 1.4888975620269775,
-      "learning_rate": 4.6820119352088665e-06,
-      "loss": 0.1246,
-      "step": 3119
-    },
-    {
-      "epoch": 0.5319551375975704,
-      "grad_norm": 0.9784233570098877,
-      "learning_rate": 4.6803069053708444e-06,
-      "loss": 0.074,
-      "step": 3120
-    },
-    {
-      "epoch": 0.5321256360391081,
-      "grad_norm": 1.2400338649749756,
-      "learning_rate": 4.678601875532822e-06,
-      "loss": 0.1186,
-      "step": 3121
-    },
-    {
-      "epoch": 0.5322961344806457,
-      "grad_norm": 1.0570576190948486,
-      "learning_rate": 4.6768968456948e-06,
-      "loss": 0.1189,
-      "step": 3122
-    },
-    {
-      "epoch": 0.5324666329221834,
-      "grad_norm": 0.9685819745063782,
-      "learning_rate": 4.675191815856777e-06,
-      "loss": 0.0836,
-      "step": 3123
-    },
-    {
-      "epoch": 0.5326371313637212,
-      "grad_norm": 0.8460286259651184,
-      "learning_rate": 4.673486786018755e-06,
-      "loss": 0.0361,
-      "step": 3124
-    },
-    {
-      "epoch": 0.5328076298052589,
-      "grad_norm": 1.683355689048767,
-      "learning_rate": 4.671781756180734e-06,
-      "loss": 0.1801,
-      "step": 3125
-    },
-    {
-      "epoch": 0.5329781282467965,
-      "grad_norm": 0.8989179134368896,
-      "learning_rate": 4.670076726342711e-06,
-      "loss": 0.0931,
-      "step": 3126
-    },
-    {
-      "epoch": 0.5331486266883342,
-      "grad_norm": 1.4232580661773682,
-      "learning_rate": 4.668371696504689e-06,
-      "loss": 0.1033,
-      "step": 3127
-    },
-    {
-      "epoch": 0.5333191251298719,
-      "grad_norm": 1.0907427072525024,
-      "learning_rate": 4.666666666666667e-06,
-      "loss": 0.0894,
-      "step": 3128
-    },
-    {
-      "epoch": 0.5334896235714095,
-      "grad_norm": 0.8971290588378906,
-      "learning_rate": 4.664961636828645e-06,
-      "loss": 0.1184,
-      "step": 3129
-    },
-    {
-      "epoch": 0.5336601220129472,
-      "grad_norm": 1.0233826637268066,
-      "learning_rate": 4.663256606990623e-06,
-      "loss": 0.0814,
-      "step": 3130
-    },
-    {
-      "epoch": 0.5338306204544849,
-      "grad_norm": 2.420344591140747,
-      "learning_rate": 4.6615515771526e-06,
-      "loss": 0.2647,
-      "step": 3131
-    },
-    {
-      "epoch": 0.5340011188960226,
-      "grad_norm": 1.550612211227417,
-      "learning_rate": 4.659846547314578e-06,
-      "loss": 0.1122,
-      "step": 3132
-    },
-    {
-      "epoch": 0.5341716173375602,
-      "grad_norm": 1.053017258644104,
-      "learning_rate": 4.658141517476556e-06,
-      "loss": 0.0879,
-      "step": 3133
-    },
-    {
-      "epoch": 0.534342115779098,
-      "grad_norm": 0.7617818117141724,
-      "learning_rate": 4.656436487638534e-06,
-      "loss": 0.0668,
-      "step": 3134
-    },
-    {
-      "epoch": 0.5345126142206357,
-      "grad_norm": 2.0507922172546387,
-      "learning_rate": 4.654731457800512e-06,
-      "loss": 0.2682,
-      "step": 3135
-    },
-    {
-      "epoch": 0.5346831126621733,
-      "grad_norm": 1.3223862648010254,
-      "learning_rate": 4.65302642796249e-06,
-      "loss": 0.1969,
-      "step": 3136
-    },
-    {
-      "epoch": 0.534853611103711,
-      "grad_norm": 0.9410365223884583,
-      "learning_rate": 4.651321398124468e-06,
-      "loss": 0.0857,
-      "step": 3137
-    },
-    {
-      "epoch": 0.5350241095452487,
-      "grad_norm": 1.1620447635650635,
-      "learning_rate": 4.649616368286445e-06,
-      "loss": 0.1836,
-      "step": 3138
-    },
-    {
-      "epoch": 0.5351946079867864,
-      "grad_norm": 1.2213563919067383,
-      "learning_rate": 4.647911338448423e-06,
-      "loss": 0.1202,
-      "step": 3139
-    },
-    {
-      "epoch": 0.535365106428324,
-      "grad_norm": 1.7266759872436523,
-      "learning_rate": 4.646206308610401e-06,
-      "loss": 0.2072,
-      "step": 3140
-    },
-    {
-      "epoch": 0.5355356048698617,
-      "grad_norm": 1.4056097269058228,
-      "learning_rate": 4.644501278772379e-06,
-      "loss": 0.2147,
-      "step": 3141
-    },
-    {
-      "epoch": 0.5357061033113995,
-      "grad_norm": 1.0043752193450928,
-      "learning_rate": 4.642796248934357e-06,
-      "loss": 0.1131,
-      "step": 3142
-    },
-    {
-      "epoch": 0.535876601752937,
-      "grad_norm": 0.7852779030799866,
-      "learning_rate": 4.641091219096335e-06,
-      "loss": 0.0757,
-      "step": 3143
-    },
-    {
-      "epoch": 0.5360471001944748,
-      "grad_norm": 1.6596177816390991,
-      "learning_rate": 4.639386189258313e-06,
-      "loss": 0.2113,
-      "step": 3144
-    },
-    {
-      "epoch": 0.5362175986360125,
-      "grad_norm": 1.0587300062179565,
-      "learning_rate": 4.637681159420291e-06,
-      "loss": 0.1213,
-      "step": 3145
-    },
-    {
-      "epoch": 0.5363880970775502,
-      "grad_norm": 1.4061554670333862,
-      "learning_rate": 4.635976129582268e-06,
-      "loss": 0.0821,
-      "step": 3146
-    },
-    {
-      "epoch": 0.5365585955190878,
-      "grad_norm": 1.5772520303726196,
-      "learning_rate": 4.634271099744246e-06,
-      "loss": 0.2397,
-      "step": 3147
-    },
-    {
-      "epoch": 0.5367290939606255,
-      "grad_norm": 1.2124773263931274,
-      "learning_rate": 4.6325660699062236e-06,
-      "loss": 0.208,
-      "step": 3148
-    },
-    {
-      "epoch": 0.5368995924021632,
-      "grad_norm": 1.1358237266540527,
-      "learning_rate": 4.6308610400682015e-06,
-      "loss": 0.1777,
-      "step": 3149
-    },
-    {
-      "epoch": 0.5370700908437008,
-      "grad_norm": 0.6872425675392151,
-      "learning_rate": 4.6291560102301795e-06,
-      "loss": 0.0927,
-      "step": 3150
-    },
-    {
-      "epoch": 0.5372405892852385,
-      "grad_norm": 1.0333954095840454,
-      "learning_rate": 4.627450980392157e-06,
-      "loss": 0.046,
-      "step": 3151
-    },
-    {
-      "epoch": 0.5374110877267763,
-      "grad_norm": 1.1037591695785522,
-      "learning_rate": 4.625745950554135e-06,
-      "loss": 0.1501,
-      "step": 3152
-    },
-    {
-      "epoch": 0.537581586168314,
-      "grad_norm": 0.85722416639328,
-      "learning_rate": 4.624040920716113e-06,
-      "loss": 0.1153,
-      "step": 3153
-    },
-    {
-      "epoch": 0.5377520846098516,
-      "grad_norm": 1.087350606918335,
-      "learning_rate": 4.62233589087809e-06,
-      "loss": 0.1529,
-      "step": 3154
-    },
-    {
-      "epoch": 0.5379225830513893,
-      "grad_norm": 0.8361851572990417,
-      "learning_rate": 4.620630861040068e-06,
-      "loss": 0.0732,
-      "step": 3155
-    },
-    {
-      "epoch": 0.538093081492927,
-      "grad_norm": 1.204031229019165,
-      "learning_rate": 4.618925831202046e-06,
-      "loss": 0.1681,
-      "step": 3156
-    },
-    {
-      "epoch": 0.5382635799344646,
-      "grad_norm": 0.6141670346260071,
-      "learning_rate": 4.617220801364024e-06,
-      "loss": 0.0487,
-      "step": 3157
-    },
-    {
-      "epoch": 0.5384340783760023,
-      "grad_norm": 0.8045641779899597,
-      "learning_rate": 4.615515771526002e-06,
-      "loss": 0.0821,
-      "step": 3158
-    },
-    {
-      "epoch": 0.53860457681754,
-      "grad_norm": 1.0021306276321411,
-      "learning_rate": 4.61381074168798e-06,
-      "loss": 0.1144,
-      "step": 3159
-    },
-    {
-      "epoch": 0.5387750752590778,
-      "grad_norm": 1.6075087785720825,
-      "learning_rate": 4.612105711849958e-06,
-      "loss": 0.1496,
-      "step": 3160
-    },
-    {
-      "epoch": 0.5389455737006154,
-      "grad_norm": 0.6130344271659851,
-      "learning_rate": 4.610400682011936e-06,
-      "loss": 0.0743,
-      "step": 3161
-    },
-    {
-      "epoch": 0.5391160721421531,
-      "grad_norm": 0.8822925090789795,
-      "learning_rate": 4.608695652173913e-06,
-      "loss": 0.1014,
-      "step": 3162
-    },
-    {
-      "epoch": 0.5392865705836908,
-      "grad_norm": 1.3176257610321045,
-      "learning_rate": 4.606990622335891e-06,
-      "loss": 0.1362,
-      "step": 3163
-    },
-    {
-      "epoch": 0.5394570690252284,
-      "grad_norm": 1.2960896492004395,
-      "learning_rate": 4.605285592497869e-06,
-      "loss": 0.1509,
-      "step": 3164
-    },
-    {
-      "epoch": 0.5396275674667661,
-      "grad_norm": 1.0878572463989258,
-      "learning_rate": 4.603580562659847e-06,
-      "loss": 0.1354,
-      "step": 3165
-    },
-    {
-      "epoch": 0.5397980659083038,
-      "grad_norm": 1.2113642692565918,
-      "learning_rate": 4.601875532821825e-06,
-      "loss": 0.1624,
-      "step": 3166
-    },
-    {
-      "epoch": 0.5399685643498415,
-      "grad_norm": 0.9236718416213989,
-      "learning_rate": 4.600170502983803e-06,
-      "loss": 0.108,
-      "step": 3167
-    },
-    {
-      "epoch": 0.5401390627913791,
-      "grad_norm": 1.6693158149719238,
-      "learning_rate": 4.598465473145781e-06,
-      "loss": 0.1436,
-      "step": 3168
-    },
-    {
-      "epoch": 0.5403095612329168,
-      "grad_norm": 1.2242648601531982,
-      "learning_rate": 4.596760443307759e-06,
-      "loss": 0.1203,
-      "step": 3169
-    },
-    {
-      "epoch": 0.5404800596744546,
-      "grad_norm": 1.18451988697052,
-      "learning_rate": 4.595055413469736e-06,
-      "loss": 0.1556,
-      "step": 3170
-    },
-    {
-      "epoch": 0.5406505581159923,
-      "grad_norm": 1.312850832939148,
-      "learning_rate": 4.593350383631714e-06,
-      "loss": 0.1569,
-      "step": 3171
-    },
-    {
-      "epoch": 0.5408210565575299,
-      "grad_norm": 0.8044204115867615,
-      "learning_rate": 4.591645353793692e-06,
-      "loss": 0.1119,
-      "step": 3172
-    },
-    {
-      "epoch": 0.5409915549990676,
-      "grad_norm": 1.1342873573303223,
-      "learning_rate": 4.58994032395567e-06,
-      "loss": 0.0958,
-      "step": 3173
-    },
-    {
-      "epoch": 0.5411620534406053,
-      "grad_norm": 0.992530107498169,
-      "learning_rate": 4.588235294117647e-06,
-      "loss": 0.0664,
-      "step": 3174
-    },
-    {
-      "epoch": 0.5413325518821429,
-      "grad_norm": 1.0137033462524414,
-      "learning_rate": 4.586530264279625e-06,
-      "loss": 0.0971,
-      "step": 3175
-    },
-    {
-      "epoch": 0.5415030503236806,
-      "grad_norm": 1.353960633277893,
-      "learning_rate": 4.5848252344416036e-06,
-      "loss": 0.1803,
-      "step": 3176
-    },
-    {
-      "epoch": 0.5416735487652183,
-      "grad_norm": 0.8238753080368042,
-      "learning_rate": 4.583120204603581e-06,
-      "loss": 0.0998,
-      "step": 3177
-    },
-    {
-      "epoch": 0.541844047206756,
-      "grad_norm": 1.0195826292037964,
-      "learning_rate": 4.581415174765559e-06,
-      "loss": 0.0941,
-      "step": 3178
-    },
-    {
-      "epoch": 0.5420145456482937,
-      "grad_norm": 1.0271581411361694,
-      "learning_rate": 4.5797101449275366e-06,
-      "loss": 0.1519,
-      "step": 3179
-    },
-    {
-      "epoch": 0.5421850440898314,
-      "grad_norm": 1.2750300168991089,
-      "learning_rate": 4.5780051150895145e-06,
-      "loss": 0.1654,
-      "step": 3180
-    },
-    {
-      "epoch": 0.5423555425313691,
-      "grad_norm": 1.4126577377319336,
-      "learning_rate": 4.5763000852514925e-06,
-      "loss": 0.244,
-      "step": 3181
-    },
-    {
-      "epoch": 0.5425260409729067,
-      "grad_norm": 1.528464913368225,
-      "learning_rate": 4.5745950554134696e-06,
-      "loss": 0.1097,
-      "step": 3182
-    },
-    {
-      "epoch": 0.5426965394144444,
-      "grad_norm": 1.406211018562317,
-      "learning_rate": 4.5728900255754475e-06,
-      "loss": 0.1076,
-      "step": 3183
-    },
-    {
-      "epoch": 0.5428670378559821,
-      "grad_norm": 1.0801411867141724,
-      "learning_rate": 4.571184995737426e-06,
-      "loss": 0.1189,
-      "step": 3184
-    },
-    {
-      "epoch": 0.5430375362975198,
-      "grad_norm": 0.8011865019798279,
-      "learning_rate": 4.569479965899403e-06,
-      "loss": 0.1224,
-      "step": 3185
-    },
-    {
-      "epoch": 0.5432080347390574,
-      "grad_norm": 1.7127665281295776,
-      "learning_rate": 4.567774936061381e-06,
-      "loss": 0.2387,
-      "step": 3186
-    },
-    {
-      "epoch": 0.5433785331805951,
-      "grad_norm": 1.7253592014312744,
-      "learning_rate": 4.566069906223359e-06,
-      "loss": 0.2265,
-      "step": 3187
-    },
-    {
-      "epoch": 0.5435490316221329,
-      "grad_norm": 1.0014303922653198,
-      "learning_rate": 4.564364876385337e-06,
-      "loss": 0.1081,
-      "step": 3188
-    },
-    {
-      "epoch": 0.5437195300636705,
-      "grad_norm": 0.881578803062439,
-      "learning_rate": 4.562659846547315e-06,
-      "loss": 0.079,
-      "step": 3189
-    },
-    {
-      "epoch": 0.5438900285052082,
-      "grad_norm": 1.217948079109192,
-      "learning_rate": 4.560954816709292e-06,
-      "loss": 0.1245,
-      "step": 3190
-    },
-    {
-      "epoch": 0.5440605269467459,
-      "grad_norm": 1.711934208869934,
-      "learning_rate": 4.55924978687127e-06,
-      "loss": 0.2053,
-      "step": 3191
-    },
-    {
-      "epoch": 0.5442310253882836,
-      "grad_norm": 1.4666311740875244,
-      "learning_rate": 4.557544757033249e-06,
-      "loss": 0.1972,
-      "step": 3192
-    },
-    {
-      "epoch": 0.5444015238298212,
-      "grad_norm": 0.7553595900535583,
-      "learning_rate": 4.555839727195226e-06,
-      "loss": 0.0775,
-      "step": 3193
-    },
-    {
-      "epoch": 0.5445720222713589,
-      "grad_norm": 1.167011022567749,
-      "learning_rate": 4.554134697357204e-06,
-      "loss": 0.1076,
-      "step": 3194
-    },
-    {
-      "epoch": 0.5447425207128966,
-      "grad_norm": 1.2989052534103394,
-      "learning_rate": 4.552429667519182e-06,
-      "loss": 0.2399,
-      "step": 3195
-    },
-    {
-      "epoch": 0.5449130191544342,
-      "grad_norm": 1.0434139966964722,
-      "learning_rate": 4.55072463768116e-06,
-      "loss": 0.0606,
-      "step": 3196
-    },
-    {
-      "epoch": 0.545083517595972,
-      "grad_norm": 0.8419189453125,
-      "learning_rate": 4.549019607843138e-06,
-      "loss": 0.0666,
-      "step": 3197
-    },
-    {
-      "epoch": 0.5452540160375097,
-      "grad_norm": 1.121074914932251,
-      "learning_rate": 4.547314578005115e-06,
-      "loss": 0.1093,
-      "step": 3198
-    },
-    {
-      "epoch": 0.5454245144790474,
-      "grad_norm": 0.7801600098609924,
-      "learning_rate": 4.545609548167093e-06,
-      "loss": 0.1091,
-      "step": 3199
-    },
-    {
-      "epoch": 0.545595012920585,
-      "grad_norm": 1.791938066482544,
-      "learning_rate": 4.543904518329071e-06,
-      "loss": 0.118,
-      "step": 3200
-    },
-    {
-      "epoch": 0.545595012920585,
-      "eval_f1_score": 0.35294117647058826,
-      "eval_loss": 0.13997682929039001,
-      "eval_runtime": 182.7002,
-      "eval_samples_per_second": 54.734,
-      "eval_steps_per_second": 3.421,
-      "step": 3200
-    },
-    {
-      "epoch": 0.5457655113621227,
-      "grad_norm": 0.9591838121414185,
-      "learning_rate": 4.542199488491049e-06,
-      "loss": 0.1401,
-      "step": 3201
-    },
-    {
-      "epoch": 0.5459360098036604,
-      "grad_norm": 0.9607254266738892,
-      "learning_rate": 4.540494458653027e-06,
-      "loss": 0.1178,
-      "step": 3202
-    },
-    {
-      "epoch": 0.546106508245198,
-      "grad_norm": 1.1044963598251343,
-      "learning_rate": 4.538789428815005e-06,
-      "loss": 0.1426,
-      "step": 3203
-    },
-    {
-      "epoch": 0.5462770066867357,
-      "grad_norm": 0.8686749935150146,
-      "learning_rate": 4.537084398976983e-06,
-      "loss": 0.098,
-      "step": 3204
-    },
-    {
-      "epoch": 0.5464475051282734,
-      "grad_norm": 1.3227715492248535,
-      "learning_rate": 4.535379369138961e-06,
-      "loss": 0.1531,
-      "step": 3205
-    },
-    {
-      "epoch": 0.5466180035698112,
-      "grad_norm": 1.3910294771194458,
-      "learning_rate": 4.533674339300938e-06,
-      "loss": 0.1657,
-      "step": 3206
-    },
-    {
-      "epoch": 0.5467885020113488,
-      "grad_norm": 0.758912205696106,
-      "learning_rate": 4.531969309462916e-06,
-      "loss": 0.0972,
-      "step": 3207
-    },
-    {
-      "epoch": 0.5469590004528865,
-      "grad_norm": 1.038732647895813,
-      "learning_rate": 4.530264279624894e-06,
-      "loss": 0.1369,
-      "step": 3208
-    },
-    {
-      "epoch": 0.5471294988944242,
-      "grad_norm": 1.2257394790649414,
-      "learning_rate": 4.528559249786872e-06,
-      "loss": 0.1925,
-      "step": 3209
-    },
-    {
-      "epoch": 0.5472999973359618,
-      "grad_norm": 0.6908015012741089,
-      "learning_rate": 4.5268542199488496e-06,
-      "loss": 0.1089,
-      "step": 3210
-    },
-    {
-      "epoch": 0.5474704957774995,
-      "grad_norm": 2.2734169960021973,
-      "learning_rate": 4.5251491901108275e-06,
-      "loss": 0.24,
-      "step": 3211
-    },
-    {
-      "epoch": 0.5476409942190372,
-      "grad_norm": 0.8731111884117126,
-      "learning_rate": 4.5234441602728055e-06,
-      "loss": 0.0902,
-      "step": 3212
-    },
-    {
-      "epoch": 0.5478114926605749,
-      "grad_norm": 1.2739516496658325,
-      "learning_rate": 4.5217391304347826e-06,
-      "loss": 0.1523,
-      "step": 3213
-    },
-    {
-      "epoch": 0.5479819911021125,
-      "grad_norm": 1.1364442110061646,
-      "learning_rate": 4.5200341005967605e-06,
-      "loss": 0.112,
-      "step": 3214
-    },
-    {
-      "epoch": 0.5481524895436503,
-      "grad_norm": 1.0354875326156616,
-      "learning_rate": 4.5183290707587384e-06,
-      "loss": 0.1174,
-      "step": 3215
-    },
-    {
-      "epoch": 0.548322987985188,
-      "grad_norm": 0.9430590271949768,
-      "learning_rate": 4.516624040920716e-06,
-      "loss": 0.1376,
-      "step": 3216
-    },
-    {
-      "epoch": 0.5484934864267257,
-      "grad_norm": 1.5261237621307373,
-      "learning_rate": 4.514919011082694e-06,
-      "loss": 0.1735,
-      "step": 3217
-    },
-    {
-      "epoch": 0.5486639848682633,
-      "grad_norm": 1.2810159921646118,
-      "learning_rate": 4.513213981244672e-06,
-      "loss": 0.1195,
-      "step": 3218
-    },
-    {
-      "epoch": 0.548834483309801,
-      "grad_norm": 1.7963886260986328,
-      "learning_rate": 4.51150895140665e-06,
-      "loss": 0.1406,
-      "step": 3219
-    },
-    {
-      "epoch": 0.5490049817513387,
-      "grad_norm": 1.2858192920684814,
-      "learning_rate": 4.509803921568628e-06,
-      "loss": 0.1395,
-      "step": 3220
-    },
-    {
-      "epoch": 0.5491754801928763,
-      "grad_norm": 0.7192848324775696,
-      "learning_rate": 4.508098891730605e-06,
-      "loss": 0.0846,
-      "step": 3221
-    },
-    {
-      "epoch": 0.549345978634414,
-      "grad_norm": 1.022417426109314,
-      "learning_rate": 4.506393861892583e-06,
-      "loss": 0.0808,
-      "step": 3222
-    },
-    {
-      "epoch": 0.5495164770759517,
-      "grad_norm": 1.14131760597229,
-      "learning_rate": 4.504688832054561e-06,
-      "loss": 0.142,
-      "step": 3223
-    },
-    {
-      "epoch": 0.5496869755174895,
-      "grad_norm": 0.8375114798545837,
-      "learning_rate": 4.502983802216539e-06,
-      "loss": 0.0939,
-      "step": 3224
-    },
-    {
-      "epoch": 0.5498574739590271,
-      "grad_norm": 1.2288278341293335,
-      "learning_rate": 4.501278772378517e-06,
-      "loss": 0.1465,
-      "step": 3225
-    },
-    {
-      "epoch": 0.5500279724005648,
-      "grad_norm": 1.3353421688079834,
-      "learning_rate": 4.499573742540495e-06,
-      "loss": 0.1193,
-      "step": 3226
-    },
-    {
-      "epoch": 0.5501984708421025,
-      "grad_norm": 1.305950403213501,
-      "learning_rate": 4.497868712702473e-06,
-      "loss": 0.1002,
-      "step": 3227
-    },
-    {
-      "epoch": 0.5503689692836401,
-      "grad_norm": 1.2217447757720947,
-      "learning_rate": 4.496163682864451e-06,
-      "loss": 0.1215,
-      "step": 3228
-    },
-    {
-      "epoch": 0.5505394677251778,
-      "grad_norm": 1.4431780576705933,
-      "learning_rate": 4.494458653026428e-06,
-      "loss": 0.1579,
-      "step": 3229
-    },
-    {
-      "epoch": 0.5507099661667155,
-      "grad_norm": 0.9493196606636047,
-      "learning_rate": 4.492753623188406e-06,
-      "loss": 0.0546,
-      "step": 3230
-    },
-    {
-      "epoch": 0.5508804646082532,
-      "grad_norm": 1.1164307594299316,
-      "learning_rate": 4.491048593350384e-06,
-      "loss": 0.1236,
-      "step": 3231
-    },
-    {
-      "epoch": 0.5510509630497908,
-      "grad_norm": 1.8813072443008423,
-      "learning_rate": 4.489343563512362e-06,
-      "loss": 0.1998,
-      "step": 3232
-    },
-    {
-      "epoch": 0.5512214614913286,
-      "grad_norm": 0.7001495957374573,
-      "learning_rate": 4.48763853367434e-06,
-      "loss": 0.1295,
-      "step": 3233
-    },
-    {
-      "epoch": 0.5513919599328663,
-      "grad_norm": 0.6423521041870117,
-      "learning_rate": 4.485933503836318e-06,
-      "loss": 0.0689,
-      "step": 3234
-    },
-    {
-      "epoch": 0.5515624583744039,
-      "grad_norm": 0.7978529930114746,
-      "learning_rate": 4.484228473998296e-06,
-      "loss": 0.074,
-      "step": 3235
-    },
-    {
-      "epoch": 0.5517329568159416,
-      "grad_norm": 0.6903978586196899,
-      "learning_rate": 4.482523444160274e-06,
-      "loss": 0.0555,
-      "step": 3236
-    },
-    {
-      "epoch": 0.5519034552574793,
-      "grad_norm": 1.7434931993484497,
-      "learning_rate": 4.480818414322251e-06,
-      "loss": 0.1858,
-      "step": 3237
-    },
-    {
-      "epoch": 0.552073953699017,
-      "grad_norm": 0.8836773037910461,
-      "learning_rate": 4.479113384484229e-06,
-      "loss": 0.1221,
-      "step": 3238
-    },
-    {
-      "epoch": 0.5522444521405546,
-      "grad_norm": 1.0610016584396362,
-      "learning_rate": 4.477408354646207e-06,
-      "loss": 0.1436,
-      "step": 3239
-    },
-    {
-      "epoch": 0.5524149505820923,
-      "grad_norm": 1.3645976781845093,
-      "learning_rate": 4.475703324808185e-06,
-      "loss": 0.1772,
-      "step": 3240
-    },
-    {
-      "epoch": 0.55258544902363,
-      "grad_norm": 2.251723527908325,
-      "learning_rate": 4.4739982949701625e-06,
-      "loss": 0.2335,
-      "step": 3241
-    },
-    {
-      "epoch": 0.5527559474651677,
-      "grad_norm": 1.4421985149383545,
-      "learning_rate": 4.47229326513214e-06,
-      "loss": 0.1223,
-      "step": 3242
-    },
-    {
-      "epoch": 0.5529264459067054,
-      "grad_norm": 1.719007968902588,
-      "learning_rate": 4.4705882352941184e-06,
-      "loss": 0.1606,
-      "step": 3243
-    },
-    {
-      "epoch": 0.5530969443482431,
-      "grad_norm": 0.9367814064025879,
-      "learning_rate": 4.468883205456096e-06,
-      "loss": 0.0743,
-      "step": 3244
-    },
-    {
-      "epoch": 0.5532674427897808,
-      "grad_norm": 1.8511252403259277,
-      "learning_rate": 4.4671781756180735e-06,
-      "loss": 0.2017,
-      "step": 3245
-    },
-    {
-      "epoch": 0.5534379412313184,
-      "grad_norm": 0.7525725364685059,
-      "learning_rate": 4.4654731457800514e-06,
-      "loss": 0.0691,
-      "step": 3246
-    },
-    {
-      "epoch": 0.5536084396728561,
-      "grad_norm": 0.9518800377845764,
-      "learning_rate": 4.463768115942029e-06,
-      "loss": 0.1091,
-      "step": 3247
-    },
-    {
-      "epoch": 0.5537789381143938,
-      "grad_norm": 1.3286054134368896,
-      "learning_rate": 4.462063086104007e-06,
-      "loss": 0.1454,
-      "step": 3248
-    },
-    {
-      "epoch": 0.5539494365559314,
-      "grad_norm": 1.1900709867477417,
-      "learning_rate": 4.4603580562659844e-06,
-      "loss": 0.0533,
-      "step": 3249
-    },
-    {
-      "epoch": 0.5541199349974691,
-      "grad_norm": 1.1784849166870117,
-      "learning_rate": 4.458653026427962e-06,
-      "loss": 0.1506,
-      "step": 3250
-    },
-    {
-      "epoch": 0.5542904334390069,
-      "grad_norm": 1.5197845697402954,
-      "learning_rate": 4.456947996589941e-06,
-      "loss": 0.1212,
-      "step": 3251
-    },
-    {
-      "epoch": 0.5544609318805446,
-      "grad_norm": 1.4505349397659302,
-      "learning_rate": 4.455242966751918e-06,
-      "loss": 0.163,
-      "step": 3252
-    },
-    {
-      "epoch": 0.5546314303220822,
-      "grad_norm": 1.64879310131073,
-      "learning_rate": 4.453537936913896e-06,
-      "loss": 0.1799,
-      "step": 3253
-    },
-    {
-      "epoch": 0.5548019287636199,
-      "grad_norm": 1.1190886497497559,
-      "learning_rate": 4.451832907075874e-06,
-      "loss": 0.1426,
-      "step": 3254
-    },
-    {
-      "epoch": 0.5549724272051576,
-      "grad_norm": 0.9179021120071411,
-      "learning_rate": 4.450127877237852e-06,
-      "loss": 0.1091,
-      "step": 3255
-    },
-    {
-      "epoch": 0.5551429256466952,
-      "grad_norm": 1.9500393867492676,
-      "learning_rate": 4.44842284739983e-06,
-      "loss": 0.1813,
-      "step": 3256
-    },
-    {
-      "epoch": 0.5553134240882329,
-      "grad_norm": 1.007033109664917,
-      "learning_rate": 4.446717817561807e-06,
-      "loss": 0.078,
-      "step": 3257
-    },
-    {
-      "epoch": 0.5554839225297706,
-      "grad_norm": 1.0274510383605957,
-      "learning_rate": 4.445012787723785e-06,
-      "loss": 0.0701,
-      "step": 3258
-    },
-    {
-      "epoch": 0.5556544209713083,
-      "grad_norm": 1.0544816255569458,
-      "learning_rate": 4.443307757885764e-06,
-      "loss": 0.1107,
-      "step": 3259
-    },
-    {
-      "epoch": 0.555824919412846,
-      "grad_norm": 1.1101003885269165,
-      "learning_rate": 4.441602728047741e-06,
-      "loss": 0.1783,
-      "step": 3260
-    },
-    {
-      "epoch": 0.5559954178543837,
-      "grad_norm": 1.0594007968902588,
-      "learning_rate": 4.439897698209719e-06,
-      "loss": 0.1132,
-      "step": 3261
-    },
-    {
-      "epoch": 0.5561659162959214,
-      "grad_norm": 1.0110628604888916,
-      "learning_rate": 4.438192668371697e-06,
-      "loss": 0.1181,
-      "step": 3262
-    },
-    {
-      "epoch": 0.5563364147374591,
-      "grad_norm": 1.122523546218872,
-      "learning_rate": 4.436487638533675e-06,
-      "loss": 0.1384,
-      "step": 3263
-    },
-    {
-      "epoch": 0.5565069131789967,
-      "grad_norm": 1.0796430110931396,
-      "learning_rate": 4.434782608695653e-06,
-      "loss": 0.118,
-      "step": 3264
-    },
-    {
-      "epoch": 0.5566774116205344,
-      "grad_norm": 1.0949740409851074,
-      "learning_rate": 4.43307757885763e-06,
-      "loss": 0.0754,
-      "step": 3265
-    },
-    {
-      "epoch": 0.5568479100620721,
-      "grad_norm": 1.1344374418258667,
-      "learning_rate": 4.431372549019608e-06,
-      "loss": 0.1355,
-      "step": 3266
-    },
-    {
-      "epoch": 0.5570184085036097,
-      "grad_norm": 0.9445719718933105,
-      "learning_rate": 4.429667519181586e-06,
-      "loss": 0.0758,
-      "step": 3267
-    },
-    {
-      "epoch": 0.5571889069451474,
-      "grad_norm": 1.1814863681793213,
-      "learning_rate": 4.427962489343564e-06,
-      "loss": 0.157,
-      "step": 3268
-    },
-    {
-      "epoch": 0.5573594053866852,
-      "grad_norm": 1.7171623706817627,
-      "learning_rate": 4.426257459505542e-06,
-      "loss": 0.2329,
-      "step": 3269
-    },
-    {
-      "epoch": 0.5575299038282229,
-      "grad_norm": 2.123250722885132,
-      "learning_rate": 4.42455242966752e-06,
-      "loss": 0.2147,
-      "step": 3270
-    },
-    {
-      "epoch": 0.5577004022697605,
-      "grad_norm": 0.832039475440979,
-      "learning_rate": 4.422847399829498e-06,
-      "loss": 0.1344,
-      "step": 3271
-    },
-    {
-      "epoch": 0.5578709007112982,
-      "grad_norm": 1.1542187929153442,
-      "learning_rate": 4.4211423699914755e-06,
-      "loss": 0.1422,
-      "step": 3272
-    },
-    {
-      "epoch": 0.5580413991528359,
-      "grad_norm": 1.8834720849990845,
-      "learning_rate": 4.419437340153453e-06,
-      "loss": 0.1717,
-      "step": 3273
-    },
-    {
-      "epoch": 0.5582118975943735,
-      "grad_norm": 0.9512043595314026,
-      "learning_rate": 4.417732310315431e-06,
-      "loss": 0.1079,
-      "step": 3274
-    },
-    {
-      "epoch": 0.5583823960359112,
-      "grad_norm": 1.4551664590835571,
-      "learning_rate": 4.4160272804774085e-06,
-      "loss": 0.1559,
-      "step": 3275
-    },
-    {
-      "epoch": 0.5585528944774489,
-      "grad_norm": 1.529512882232666,
-      "learning_rate": 4.4143222506393865e-06,
-      "loss": 0.1045,
-      "step": 3276
-    },
-    {
-      "epoch": 0.5587233929189866,
-      "grad_norm": 1.2470623254776,
-      "learning_rate": 4.4126172208013644e-06,
-      "loss": 0.1676,
-      "step": 3277
-    },
-    {
-      "epoch": 0.5588938913605243,
-      "grad_norm": 1.3405524492263794,
-      "learning_rate": 4.410912190963342e-06,
-      "loss": 0.1573,
-      "step": 3278
-    },
-    {
-      "epoch": 0.559064389802062,
-      "grad_norm": 0.9211791157722473,
-      "learning_rate": 4.40920716112532e-06,
-      "loss": 0.1034,
-      "step": 3279
-    },
-    {
-      "epoch": 0.5592348882435997,
-      "grad_norm": 0.8946516513824463,
-      "learning_rate": 4.407502131287298e-06,
-      "loss": 0.0788,
-      "step": 3280
-    },
-    {
-      "epoch": 0.5594053866851373,
-      "grad_norm": 1.298832893371582,
-      "learning_rate": 4.405797101449275e-06,
-      "loss": 0.1464,
-      "step": 3281
-    },
-    {
-      "epoch": 0.559575885126675,
-      "grad_norm": 0.7750890254974365,
-      "learning_rate": 4.404092071611253e-06,
-      "loss": 0.0968,
-      "step": 3282
-    },
-    {
-      "epoch": 0.5597463835682127,
-      "grad_norm": 1.0168732404708862,
-      "learning_rate": 4.402387041773231e-06,
-      "loss": 0.1289,
-      "step": 3283
-    },
-    {
-      "epoch": 0.5599168820097504,
-      "grad_norm": 0.8322630524635315,
-      "learning_rate": 4.400682011935209e-06,
-      "loss": 0.0385,
-      "step": 3284
-    },
-    {
-      "epoch": 0.560087380451288,
-      "grad_norm": 0.9391579627990723,
-      "learning_rate": 4.398976982097187e-06,
-      "loss": 0.0804,
-      "step": 3285
-    },
-    {
-      "epoch": 0.5602578788928257,
-      "grad_norm": 1.2026317119598389,
-      "learning_rate": 4.397271952259165e-06,
-      "loss": 0.1053,
-      "step": 3286
-    },
-    {
-      "epoch": 0.5604283773343635,
-      "grad_norm": 0.9412278532981873,
-      "learning_rate": 4.395566922421143e-06,
-      "loss": 0.0869,
-      "step": 3287
-    },
-    {
-      "epoch": 0.5605988757759011,
-      "grad_norm": 0.8000788688659668,
-      "learning_rate": 4.39386189258312e-06,
-      "loss": 0.0791,
-      "step": 3288
-    },
-    {
-      "epoch": 0.5607693742174388,
-      "grad_norm": 1.013322353363037,
-      "learning_rate": 4.392156862745098e-06,
-      "loss": 0.1263,
-      "step": 3289
-    },
-    {
-      "epoch": 0.5609398726589765,
-      "grad_norm": 1.9154322147369385,
-      "learning_rate": 4.390451832907076e-06,
-      "loss": 0.1904,
-      "step": 3290
-    },
-    {
-      "epoch": 0.5611103711005142,
-      "grad_norm": 0.8865220546722412,
-      "learning_rate": 4.388746803069054e-06,
-      "loss": 0.0427,
-      "step": 3291
-    },
-    {
-      "epoch": 0.5612808695420518,
-      "grad_norm": 1.1697574853897095,
-      "learning_rate": 4.387041773231032e-06,
-      "loss": 0.0839,
-      "step": 3292
-    },
-    {
-      "epoch": 0.5614513679835895,
-      "grad_norm": 0.6758675575256348,
-      "learning_rate": 4.38533674339301e-06,
-      "loss": 0.0544,
-      "step": 3293
-    },
-    {
-      "epoch": 0.5616218664251272,
-      "grad_norm": 1.0683929920196533,
-      "learning_rate": 4.383631713554988e-06,
-      "loss": 0.1041,
-      "step": 3294
-    },
-    {
-      "epoch": 0.5617923648666648,
-      "grad_norm": 0.6282282471656799,
-      "learning_rate": 4.381926683716966e-06,
-      "loss": 0.06,
-      "step": 3295
-    },
-    {
-      "epoch": 0.5619628633082026,
-      "grad_norm": 1.2010581493377686,
-      "learning_rate": 4.380221653878943e-06,
-      "loss": 0.101,
-      "step": 3296
-    },
-    {
-      "epoch": 0.5621333617497403,
-      "grad_norm": 0.8598968386650085,
-      "learning_rate": 4.378516624040921e-06,
-      "loss": 0.0899,
-      "step": 3297
-    },
-    {
-      "epoch": 0.562303860191278,
-      "grad_norm": 1.1526156663894653,
-      "learning_rate": 4.376811594202899e-06,
-      "loss": 0.1155,
-      "step": 3298
-    },
-    {
-      "epoch": 0.5624743586328156,
-      "grad_norm": 1.0849175453186035,
-      "learning_rate": 4.375106564364877e-06,
-      "loss": 0.1122,
-      "step": 3299
-    },
-    {
-      "epoch": 0.5626448570743533,
-      "grad_norm": 1.1330934762954712,
-      "learning_rate": 4.373401534526855e-06,
-      "loss": 0.1131,
-      "step": 3300
-    },
-    {
-      "epoch": 0.5626448570743533,
-      "eval_f1_score": 0.3415977961432507,
-      "eval_loss": 0.14648857712745667,
-      "eval_runtime": 182.7055,
-      "eval_samples_per_second": 54.733,
-      "eval_steps_per_second": 3.421,
-      "step": 3300
-    },
-    {
-      "epoch": 0.562815355515891,
-      "grad_norm": 1.0012074708938599,
-      "learning_rate": 4.371696504688833e-06,
-      "loss": 0.1121,
-      "step": 3301
-    },
-    {
-      "epoch": 0.5629858539574286,
-      "grad_norm": 1.2949241399765015,
-      "learning_rate": 4.369991474850811e-06,
-      "loss": 0.0782,
-      "step": 3302
-    },
-    {
-      "epoch": 0.5631563523989663,
-      "grad_norm": 1.4209520816802979,
-      "learning_rate": 4.3682864450127885e-06,
-      "loss": 0.1423,
-      "step": 3303
-    },
-    {
-      "epoch": 0.563326850840504,
-      "grad_norm": 1.8024206161499023,
-      "learning_rate": 4.366581415174766e-06,
-      "loss": 0.1967,
-      "step": 3304
-    },
-    {
-      "epoch": 0.5634973492820418,
-      "grad_norm": 1.5164883136749268,
-      "learning_rate": 4.3648763853367436e-06,
-      "loss": 0.1384,
-      "step": 3305
-    },
-    {
-      "epoch": 0.5636678477235794,
-      "grad_norm": 0.7025049924850464,
-      "learning_rate": 4.3631713554987215e-06,
-      "loss": 0.0665,
-      "step": 3306
-    },
-    {
-      "epoch": 0.5638383461651171,
-      "grad_norm": 0.8478127717971802,
-      "learning_rate": 4.3614663256606995e-06,
-      "loss": 0.0797,
-      "step": 3307
-    },
-    {
-      "epoch": 0.5640088446066548,
-      "grad_norm": 1.3572134971618652,
-      "learning_rate": 4.359761295822677e-06,
-      "loss": 0.1691,
-      "step": 3308
-    },
-    {
-      "epoch": 0.5641793430481925,
-      "grad_norm": 1.7278790473937988,
-      "learning_rate": 4.358056265984655e-06,
-      "loss": 0.0971,
-      "step": 3309
-    },
-    {
-      "epoch": 0.5643498414897301,
-      "grad_norm": 1.7452386617660522,
-      "learning_rate": 4.356351236146633e-06,
-      "loss": 0.0885,
-      "step": 3310
-    },
-    {
-      "epoch": 0.5645203399312678,
-      "grad_norm": 1.2867670059204102,
-      "learning_rate": 4.354646206308611e-06,
-      "loss": 0.138,
-      "step": 3311
-    },
-    {
-      "epoch": 0.5646908383728055,
-      "grad_norm": 1.348374843597412,
-      "learning_rate": 4.352941176470588e-06,
-      "loss": 0.204,
-      "step": 3312
-    },
-    {
-      "epoch": 0.5648613368143431,
-      "grad_norm": 1.2316983938217163,
-      "learning_rate": 4.351236146632566e-06,
-      "loss": 0.1296,
-      "step": 3313
-    },
-    {
-      "epoch": 0.5650318352558809,
-      "grad_norm": 0.9647215008735657,
-      "learning_rate": 4.349531116794544e-06,
-      "loss": 0.077,
-      "step": 3314
-    },
-    {
-      "epoch": 0.5652023336974186,
-      "grad_norm": 1.275435209274292,
-      "learning_rate": 4.347826086956522e-06,
-      "loss": 0.1548,
-      "step": 3315
-    },
-    {
-      "epoch": 0.5653728321389563,
-      "grad_norm": 0.8637387752532959,
-      "learning_rate": 4.3461210571185e-06,
-      "loss": 0.082,
-      "step": 3316
-    },
-    {
-      "epoch": 0.5655433305804939,
-      "grad_norm": 0.9807612299919128,
-      "learning_rate": 4.344416027280477e-06,
-      "loss": 0.1143,
-      "step": 3317
-    },
-    {
-      "epoch": 0.5657138290220316,
-      "grad_norm": 1.227411150932312,
-      "learning_rate": 4.342710997442456e-06,
-      "loss": 0.0981,
-      "step": 3318
-    },
-    {
-      "epoch": 0.5658843274635693,
-      "grad_norm": 1.0344433784484863,
-      "learning_rate": 4.341005967604434e-06,
-      "loss": 0.0503,
-      "step": 3319
-    },
-    {
-      "epoch": 0.5660548259051069,
-      "grad_norm": 1.2785519361495972,
-      "learning_rate": 4.339300937766411e-06,
-      "loss": 0.1693,
-      "step": 3320
-    },
-    {
-      "epoch": 0.5662253243466446,
-      "grad_norm": 1.272032618522644,
-      "learning_rate": 4.337595907928389e-06,
-      "loss": 0.1297,
-      "step": 3321
-    },
-    {
-      "epoch": 0.5663958227881823,
-      "grad_norm": 0.9052433967590332,
-      "learning_rate": 4.335890878090367e-06,
-      "loss": 0.0809,
-      "step": 3322
-    },
-    {
-      "epoch": 0.5665663212297201,
-      "grad_norm": 1.420202612876892,
-      "learning_rate": 4.334185848252345e-06,
-      "loss": 0.1847,
-      "step": 3323
-    },
-    {
-      "epoch": 0.5667368196712577,
-      "grad_norm": 1.4241760969161987,
-      "learning_rate": 4.332480818414322e-06,
-      "loss": 0.1444,
-      "step": 3324
-    },
-    {
-      "epoch": 0.5669073181127954,
-      "grad_norm": 1.4881341457366943,
-      "learning_rate": 4.3307757885763e-06,
-      "loss": 0.134,
-      "step": 3325
-    },
-    {
-      "epoch": 0.5670778165543331,
-      "grad_norm": 0.7717912197113037,
-      "learning_rate": 4.329070758738279e-06,
-      "loss": 0.0363,
-      "step": 3326
-    },
-    {
-      "epoch": 0.5672483149958707,
-      "grad_norm": 0.9192156195640564,
-      "learning_rate": 4.327365728900256e-06,
-      "loss": 0.1181,
-      "step": 3327
-    },
-    {
-      "epoch": 0.5674188134374084,
-      "grad_norm": 1.1816763877868652,
-      "learning_rate": 4.325660699062234e-06,
-      "loss": 0.1295,
-      "step": 3328
-    },
-    {
-      "epoch": 0.5675893118789461,
-      "grad_norm": 2.1257288455963135,
-      "learning_rate": 4.323955669224212e-06,
-      "loss": 0.1418,
-      "step": 3329
-    },
-    {
-      "epoch": 0.5677598103204838,
-      "grad_norm": 1.3970263004302979,
-      "learning_rate": 4.32225063938619e-06,
-      "loss": 0.1446,
-      "step": 3330
-    },
-    {
-      "epoch": 0.5679303087620214,
-      "grad_norm": 1.0706558227539062,
-      "learning_rate": 4.320545609548168e-06,
-      "loss": 0.1379,
-      "step": 3331
-    },
-    {
-      "epoch": 0.5681008072035592,
-      "grad_norm": 1.8220258951187134,
-      "learning_rate": 4.318840579710145e-06,
-      "loss": 0.0634,
-      "step": 3332
-    },
-    {
-      "epoch": 0.5682713056450969,
-      "grad_norm": 0.7084428071975708,
-      "learning_rate": 4.317135549872123e-06,
-      "loss": 0.0661,
-      "step": 3333
-    },
-    {
-      "epoch": 0.5684418040866345,
-      "grad_norm": 1.0817866325378418,
-      "learning_rate": 4.315430520034101e-06,
-      "loss": 0.1151,
-      "step": 3334
-    },
-    {
-      "epoch": 0.5686123025281722,
-      "grad_norm": 0.9813879728317261,
-      "learning_rate": 4.313725490196079e-06,
-      "loss": 0.1207,
-      "step": 3335
-    },
-    {
-      "epoch": 0.5687828009697099,
-      "grad_norm": 1.0461468696594238,
-      "learning_rate": 4.3120204603580566e-06,
-      "loss": 0.0888,
-      "step": 3336
-    },
-    {
-      "epoch": 0.5689532994112476,
-      "grad_norm": 1.514338731765747,
-      "learning_rate": 4.3103154305200345e-06,
-      "loss": 0.1898,
-      "step": 3337
-    },
-    {
-      "epoch": 0.5691237978527852,
-      "grad_norm": 1.2786973714828491,
-      "learning_rate": 4.3086104006820125e-06,
-      "loss": 0.1808,
-      "step": 3338
-    },
-    {
-      "epoch": 0.5692942962943229,
-      "grad_norm": 1.331338882446289,
-      "learning_rate": 4.30690537084399e-06,
-      "loss": 0.1076,
-      "step": 3339
-    },
-    {
-      "epoch": 0.5694647947358606,
-      "grad_norm": 1.214442491531372,
-      "learning_rate": 4.3052003410059675e-06,
-      "loss": 0.1187,
-      "step": 3340
-    },
-    {
-      "epoch": 0.5696352931773982,
-      "grad_norm": 1.40016770362854,
-      "learning_rate": 4.3034953111679455e-06,
-      "loss": 0.1684,
-      "step": 3341
-    },
-    {
-      "epoch": 0.569805791618936,
-      "grad_norm": 1.2763607501983643,
-      "learning_rate": 4.301790281329923e-06,
-      "loss": 0.1473,
-      "step": 3342
-    },
-    {
-      "epoch": 0.5699762900604737,
-      "grad_norm": 1.1418249607086182,
-      "learning_rate": 4.300085251491901e-06,
-      "loss": 0.1407,
-      "step": 3343
-    },
-    {
-      "epoch": 0.5701467885020114,
-      "grad_norm": 0.975607693195343,
-      "learning_rate": 4.298380221653879e-06,
-      "loss": 0.1556,
-      "step": 3344
-    },
-    {
-      "epoch": 0.570317286943549,
-      "grad_norm": 0.8215646743774414,
-      "learning_rate": 4.296675191815857e-06,
-      "loss": 0.0585,
-      "step": 3345
-    },
-    {
-      "epoch": 0.5704877853850867,
-      "grad_norm": 1.139883279800415,
-      "learning_rate": 4.294970161977835e-06,
-      "loss": 0.1284,
-      "step": 3346
-    },
-    {
-      "epoch": 0.5706582838266244,
-      "grad_norm": 1.4383643865585327,
-      "learning_rate": 4.293265132139813e-06,
-      "loss": 0.1621,
-      "step": 3347
-    },
-    {
-      "epoch": 0.570828782268162,
-      "grad_norm": 0.9598284959793091,
-      "learning_rate": 4.29156010230179e-06,
-      "loss": 0.0739,
-      "step": 3348
-    },
-    {
-      "epoch": 0.5709992807096997,
-      "grad_norm": 1.20563805103302,
-      "learning_rate": 4.289855072463768e-06,
-      "loss": 0.0754,
-      "step": 3349
-    },
-    {
-      "epoch": 0.5711697791512375,
-      "grad_norm": 1.4110171794891357,
-      "learning_rate": 4.288150042625746e-06,
-      "loss": 0.112,
-      "step": 3350
-    },
-    {
-      "epoch": 0.5713402775927752,
-      "grad_norm": 0.9314337968826294,
-      "learning_rate": 4.286445012787724e-06,
-      "loss": 0.1166,
-      "step": 3351
-    },
-    {
-      "epoch": 0.5715107760343128,
-      "grad_norm": 1.6237143278121948,
-      "learning_rate": 4.284739982949702e-06,
-      "loss": 0.178,
-      "step": 3352
-    },
-    {
-      "epoch": 0.5716812744758505,
-      "grad_norm": 1.2615511417388916,
-      "learning_rate": 4.28303495311168e-06,
-      "loss": 0.1454,
-      "step": 3353
-    },
-    {
-      "epoch": 0.5718517729173882,
-      "grad_norm": 0.9481834173202515,
-      "learning_rate": 4.281329923273658e-06,
-      "loss": 0.0704,
-      "step": 3354
-    },
-    {
-      "epoch": 0.5720222713589258,
-      "grad_norm": 0.8312090039253235,
-      "learning_rate": 4.279624893435636e-06,
-      "loss": 0.0806,
-      "step": 3355
-    },
-    {
-      "epoch": 0.5721927698004635,
-      "grad_norm": 1.0366708040237427,
-      "learning_rate": 4.277919863597613e-06,
-      "loss": 0.1035,
-      "step": 3356
-    },
-    {
-      "epoch": 0.5723632682420012,
-      "grad_norm": 1.049806833267212,
-      "learning_rate": 4.276214833759591e-06,
-      "loss": 0.1438,
-      "step": 3357
-    },
-    {
-      "epoch": 0.572533766683539,
-      "grad_norm": 1.5999555587768555,
-      "learning_rate": 4.274509803921569e-06,
-      "loss": 0.1176,
-      "step": 3358
-    },
-    {
-      "epoch": 0.5727042651250766,
-      "grad_norm": 1.3076897859573364,
-      "learning_rate": 4.272804774083547e-06,
-      "loss": 0.1431,
-      "step": 3359
-    },
-    {
-      "epoch": 0.5728747635666143,
-      "grad_norm": 1.069909930229187,
-      "learning_rate": 4.271099744245525e-06,
-      "loss": 0.0521,
-      "step": 3360
-    },
-    {
-      "epoch": 0.573045262008152,
-      "grad_norm": 1.748558759689331,
-      "learning_rate": 4.269394714407503e-06,
-      "loss": 0.186,
-      "step": 3361
-    },
-    {
-      "epoch": 0.5732157604496897,
-      "grad_norm": 2.2109155654907227,
-      "learning_rate": 4.267689684569481e-06,
-      "loss": 0.1384,
-      "step": 3362
-    },
-    {
-      "epoch": 0.5733862588912273,
-      "grad_norm": 2.6417386531829834,
-      "learning_rate": 4.265984654731458e-06,
-      "loss": 0.2327,
-      "step": 3363
-    },
-    {
-      "epoch": 0.573556757332765,
-      "grad_norm": 1.4780867099761963,
-      "learning_rate": 4.264279624893436e-06,
-      "loss": 0.168,
-      "step": 3364
-    },
-    {
-      "epoch": 0.5737272557743027,
-      "grad_norm": 1.3667103052139282,
-      "learning_rate": 4.262574595055414e-06,
-      "loss": 0.1058,
-      "step": 3365
-    },
-    {
-      "epoch": 0.5738977542158403,
-      "grad_norm": 1.0596626996994019,
-      "learning_rate": 4.260869565217392e-06,
-      "loss": 0.086,
-      "step": 3366
-    },
-    {
-      "epoch": 0.574068252657378,
-      "grad_norm": 1.2773600816726685,
-      "learning_rate": 4.2591645353793696e-06,
-      "loss": 0.2099,
-      "step": 3367
-    },
-    {
-      "epoch": 0.5742387510989158,
-      "grad_norm": 1.5703792572021484,
-      "learning_rate": 4.2574595055413475e-06,
-      "loss": 0.1702,
-      "step": 3368
-    },
-    {
-      "epoch": 0.5744092495404535,
-      "grad_norm": 1.1637566089630127,
-      "learning_rate": 4.2557544757033255e-06,
-      "loss": 0.1827,
-      "step": 3369
-    },
-    {
-      "epoch": 0.5745797479819911,
-      "grad_norm": 0.9364039897918701,
-      "learning_rate": 4.254049445865303e-06,
-      "loss": 0.0797,
-      "step": 3370
-    },
-    {
-      "epoch": 0.5747502464235288,
-      "grad_norm": 1.0836312770843506,
-      "learning_rate": 4.2523444160272805e-06,
-      "loss": 0.0998,
-      "step": 3371
-    },
-    {
-      "epoch": 0.5749207448650665,
-      "grad_norm": 1.0185374021530151,
-      "learning_rate": 4.2506393861892584e-06,
-      "loss": 0.1401,
-      "step": 3372
-    },
-    {
-      "epoch": 0.5750912433066041,
-      "grad_norm": 1.0646388530731201,
-      "learning_rate": 4.248934356351236e-06,
-      "loss": 0.1173,
-      "step": 3373
-    },
-    {
-      "epoch": 0.5752617417481418,
-      "grad_norm": 1.3602221012115479,
-      "learning_rate": 4.247229326513214e-06,
-      "loss": 0.154,
-      "step": 3374
-    },
-    {
-      "epoch": 0.5754322401896795,
-      "grad_norm": 0.8349620699882507,
-      "learning_rate": 4.245524296675192e-06,
-      "loss": 0.0912,
-      "step": 3375
-    },
-    {
-      "epoch": 0.5756027386312172,
-      "grad_norm": 1.0016189813613892,
-      "learning_rate": 4.24381926683717e-06,
-      "loss": 0.1697,
-      "step": 3376
-    },
-    {
-      "epoch": 0.5757732370727549,
-      "grad_norm": 1.3852578401565552,
-      "learning_rate": 4.242114236999148e-06,
-      "loss": 0.171,
-      "step": 3377
-    },
-    {
-      "epoch": 0.5759437355142926,
-      "grad_norm": 0.9213678240776062,
-      "learning_rate": 4.240409207161126e-06,
-      "loss": 0.069,
-      "step": 3378
-    },
-    {
-      "epoch": 0.5761142339558303,
-      "grad_norm": 2.2820992469787598,
-      "learning_rate": 4.238704177323103e-06,
-      "loss": 0.1907,
-      "step": 3379
-    },
-    {
-      "epoch": 0.5762847323973679,
-      "grad_norm": 0.8198938965797424,
-      "learning_rate": 4.236999147485081e-06,
-      "loss": 0.0571,
-      "step": 3380
-    },
-    {
-      "epoch": 0.5764552308389056,
-      "grad_norm": 0.998900830745697,
-      "learning_rate": 4.235294117647059e-06,
-      "loss": 0.1048,
-      "step": 3381
-    },
-    {
-      "epoch": 0.5766257292804433,
-      "grad_norm": 1.1199216842651367,
-      "learning_rate": 4.233589087809037e-06,
-      "loss": 0.13,
-      "step": 3382
-    },
-    {
-      "epoch": 0.576796227721981,
-      "grad_norm": 0.9903812408447266,
-      "learning_rate": 4.231884057971015e-06,
-      "loss": 0.0964,
-      "step": 3383
-    },
-    {
-      "epoch": 0.5769667261635186,
-      "grad_norm": 1.2220855951309204,
-      "learning_rate": 4.230179028132992e-06,
-      "loss": 0.1361,
-      "step": 3384
-    },
-    {
-      "epoch": 0.5771372246050563,
-      "grad_norm": 1.296735405921936,
-      "learning_rate": 4.228473998294971e-06,
-      "loss": 0.1406,
-      "step": 3385
-    },
-    {
-      "epoch": 0.5773077230465941,
-      "grad_norm": 1.0935611724853516,
-      "learning_rate": 4.226768968456949e-06,
-      "loss": 0.1243,
-      "step": 3386
-    },
-    {
-      "epoch": 0.5774782214881317,
-      "grad_norm": 1.115261197090149,
-      "learning_rate": 4.225063938618926e-06,
-      "loss": 0.1335,
-      "step": 3387
-    },
-    {
-      "epoch": 0.5776487199296694,
-      "grad_norm": 1.1657440662384033,
-      "learning_rate": 4.223358908780904e-06,
-      "loss": 0.1457,
-      "step": 3388
-    },
-    {
-      "epoch": 0.5778192183712071,
-      "grad_norm": 0.8687394857406616,
-      "learning_rate": 4.221653878942882e-06,
-      "loss": 0.0648,
-      "step": 3389
-    },
-    {
-      "epoch": 0.5779897168127448,
-      "grad_norm": 1.134837031364441,
-      "learning_rate": 4.21994884910486e-06,
-      "loss": 0.1265,
-      "step": 3390
-    },
-    {
-      "epoch": 0.5781602152542824,
-      "grad_norm": 1.6776621341705322,
-      "learning_rate": 4.218243819266838e-06,
-      "loss": 0.2094,
-      "step": 3391
-    },
-    {
-      "epoch": 0.5783307136958201,
-      "grad_norm": 1.5459142923355103,
-      "learning_rate": 4.216538789428815e-06,
-      "loss": 0.1759,
-      "step": 3392
-    },
-    {
-      "epoch": 0.5785012121373578,
-      "grad_norm": 1.0331432819366455,
-      "learning_rate": 4.214833759590794e-06,
-      "loss": 0.1374,
-      "step": 3393
-    },
-    {
-      "epoch": 0.5786717105788954,
-      "grad_norm": 0.9535524845123291,
-      "learning_rate": 4.213128729752772e-06,
-      "loss": 0.0827,
-      "step": 3394
-    },
-    {
-      "epoch": 0.5788422090204332,
-      "grad_norm": 1.3507156372070312,
-      "learning_rate": 4.211423699914749e-06,
-      "loss": 0.1657,
-      "step": 3395
-    },
-    {
-      "epoch": 0.5790127074619709,
-      "grad_norm": 0.9514930248260498,
-      "learning_rate": 4.209718670076727e-06,
-      "loss": 0.1019,
-      "step": 3396
-    },
-    {
-      "epoch": 0.5791832059035086,
-      "grad_norm": 1.091905951499939,
-      "learning_rate": 4.208013640238705e-06,
-      "loss": 0.1319,
-      "step": 3397
-    },
-    {
-      "epoch": 0.5793537043450462,
-      "grad_norm": 0.9134871363639832,
-      "learning_rate": 4.2063086104006825e-06,
-      "loss": 0.0862,
-      "step": 3398
-    },
-    {
-      "epoch": 0.5795242027865839,
-      "grad_norm": 1.7938615083694458,
-      "learning_rate": 4.20460358056266e-06,
-      "loss": 0.1689,
-      "step": 3399
-    },
-    {
-      "epoch": 0.5796947012281216,
-      "grad_norm": 2.3345704078674316,
-      "learning_rate": 4.202898550724638e-06,
-      "loss": 0.2431,
-      "step": 3400
-    },
-    {
-      "epoch": 0.5796947012281216,
-      "eval_f1_score": 0.36548223350253806,
-      "eval_loss": 0.1384306102991104,
-      "eval_runtime": 182.6782,
-      "eval_samples_per_second": 54.741,
-      "eval_steps_per_second": 3.421,
-      "step": 3400
-    },
-    {
-      "epoch": 0.5798651996696592,
-      "grad_norm": 1.92391037940979,
-      "learning_rate": 4.2011935208866155e-06,
-      "loss": 0.1983,
-      "step": 3401
-    },
-    {
-      "epoch": 0.5800356981111969,
-      "grad_norm": 2.0211410522460938,
-      "learning_rate": 4.1994884910485935e-06,
-      "loss": 0.2481,
-      "step": 3402
-    },
-    {
-      "epoch": 0.5802061965527346,
-      "grad_norm": 0.8273227214813232,
-      "learning_rate": 4.1977834612105714e-06,
-      "loss": 0.0832,
-      "step": 3403
-    },
-    {
-      "epoch": 0.5803766949942724,
-      "grad_norm": 1.275863766670227,
-      "learning_rate": 4.196078431372549e-06,
-      "loss": 0.1543,
-      "step": 3404
-    },
-    {
-      "epoch": 0.58054719343581,
-      "grad_norm": 0.9006935358047485,
-      "learning_rate": 4.194373401534527e-06,
-      "loss": 0.1069,
-      "step": 3405
-    },
-    {
-      "epoch": 0.5807176918773477,
-      "grad_norm": 1.3350001573562622,
-      "learning_rate": 4.192668371696505e-06,
-      "loss": 0.1434,
-      "step": 3406
-    },
-    {
-      "epoch": 0.5808881903188854,
-      "grad_norm": 0.8021498322486877,
-      "learning_rate": 4.190963341858482e-06,
-      "loss": 0.1094,
-      "step": 3407
-    },
-    {
-      "epoch": 0.5810586887604231,
-      "grad_norm": 0.9336419105529785,
-      "learning_rate": 4.18925831202046e-06,
-      "loss": 0.1128,
-      "step": 3408
-    },
-    {
-      "epoch": 0.5812291872019607,
-      "grad_norm": 0.8901069760322571,
-      "learning_rate": 4.187553282182438e-06,
-      "loss": 0.1042,
-      "step": 3409
-    },
-    {
-      "epoch": 0.5813996856434984,
-      "grad_norm": 1.0303782224655151,
-      "learning_rate": 4.185848252344416e-06,
-      "loss": 0.0964,
-      "step": 3410
-    },
-    {
-      "epoch": 0.5815701840850361,
-      "grad_norm": 1.1685336828231812,
-      "learning_rate": 4.184143222506394e-06,
-      "loss": 0.1435,
-      "step": 3411
-    },
-    {
-      "epoch": 0.5817406825265737,
-      "grad_norm": 1.0988837480545044,
-      "learning_rate": 4.182438192668372e-06,
-      "loss": 0.1413,
-      "step": 3412
-    },
-    {
-      "epoch": 0.5819111809681115,
-      "grad_norm": 1.4995113611221313,
-      "learning_rate": 4.18073316283035e-06,
-      "loss": 0.1951,
-      "step": 3413
-    },
-    {
-      "epoch": 0.5820816794096492,
-      "grad_norm": 1.328747272491455,
-      "learning_rate": 4.179028132992328e-06,
-      "loss": 0.1393,
-      "step": 3414
-    },
-    {
-      "epoch": 0.5822521778511869,
-      "grad_norm": 1.1531280279159546,
-      "learning_rate": 4.177323103154305e-06,
-      "loss": 0.068,
-      "step": 3415
-    },
-    {
-      "epoch": 0.5824226762927245,
-      "grad_norm": 0.9878232479095459,
-      "learning_rate": 4.175618073316283e-06,
-      "loss": 0.0741,
-      "step": 3416
-    },
-    {
-      "epoch": 0.5825931747342622,
-      "grad_norm": 1.3367799520492554,
-      "learning_rate": 4.173913043478261e-06,
-      "loss": 0.1205,
-      "step": 3417
-    },
-    {
-      "epoch": 0.5827636731757999,
-      "grad_norm": 1.0414282083511353,
-      "learning_rate": 4.172208013640239e-06,
-      "loss": 0.1037,
-      "step": 3418
-    },
-    {
-      "epoch": 0.5829341716173375,
-      "grad_norm": 0.865533173084259,
-      "learning_rate": 4.170502983802217e-06,
-      "loss": 0.1143,
-      "step": 3419
-    },
-    {
-      "epoch": 0.5831046700588752,
-      "grad_norm": 1.3106995820999146,
-      "learning_rate": 4.168797953964195e-06,
-      "loss": 0.1218,
-      "step": 3420
-    },
-    {
-      "epoch": 0.5832751685004129,
-      "grad_norm": 0.8746435046195984,
-      "learning_rate": 4.167092924126173e-06,
-      "loss": 0.1248,
-      "step": 3421
-    },
-    {
-      "epoch": 0.5834456669419507,
-      "grad_norm": 0.6834039092063904,
-      "learning_rate": 4.165387894288151e-06,
-      "loss": 0.0722,
-      "step": 3422
-    },
-    {
-      "epoch": 0.5836161653834883,
-      "grad_norm": 1.1980594396591187,
-      "learning_rate": 4.163682864450128e-06,
-      "loss": 0.1023,
-      "step": 3423
-    },
-    {
-      "epoch": 0.583786663825026,
-      "grad_norm": 0.9332726001739502,
-      "learning_rate": 4.161977834612106e-06,
-      "loss": 0.0713,
-      "step": 3424
-    },
-    {
-      "epoch": 0.5839571622665637,
-      "grad_norm": 1.0442097187042236,
-      "learning_rate": 4.160272804774084e-06,
-      "loss": 0.0868,
-      "step": 3425
-    },
-    {
-      "epoch": 0.5841276607081013,
-      "grad_norm": 1.9792109727859497,
-      "learning_rate": 4.158567774936062e-06,
-      "loss": 0.1933,
-      "step": 3426
-    },
-    {
-      "epoch": 0.584298159149639,
-      "grad_norm": 0.6008835434913635,
-      "learning_rate": 4.15686274509804e-06,
-      "loss": 0.0551,
-      "step": 3427
-    },
-    {
-      "epoch": 0.5844686575911767,
-      "grad_norm": 1.496996283531189,
-      "learning_rate": 4.155157715260018e-06,
-      "loss": 0.1422,
-      "step": 3428
-    },
-    {
-      "epoch": 0.5846391560327144,
-      "grad_norm": 0.8596488833427429,
-      "learning_rate": 4.1534526854219955e-06,
-      "loss": 0.0683,
-      "step": 3429
-    },
-    {
-      "epoch": 0.584809654474252,
-      "grad_norm": 0.5586788654327393,
-      "learning_rate": 4.1517476555839735e-06,
-      "loss": 0.1202,
-      "step": 3430
-    },
-    {
-      "epoch": 0.5849801529157898,
-      "grad_norm": 1.0073561668395996,
-      "learning_rate": 4.150042625745951e-06,
-      "loss": 0.0933,
-      "step": 3431
-    },
-    {
-      "epoch": 0.5851506513573275,
-      "grad_norm": 1.4064346551895142,
-      "learning_rate": 4.1483375959079285e-06,
-      "loss": 0.1591,
-      "step": 3432
-    },
-    {
-      "epoch": 0.5853211497988651,
-      "grad_norm": 1.2082147598266602,
-      "learning_rate": 4.1466325660699065e-06,
-      "loss": 0.1669,
-      "step": 3433
-    },
-    {
-      "epoch": 0.5854916482404028,
-      "grad_norm": 1.391356348991394,
-      "learning_rate": 4.1449275362318844e-06,
-      "loss": 0.1946,
-      "step": 3434
-    },
-    {
-      "epoch": 0.5856621466819405,
-      "grad_norm": 0.9328740835189819,
-      "learning_rate": 4.143222506393862e-06,
-      "loss": 0.0658,
-      "step": 3435
-    },
-    {
-      "epoch": 0.5858326451234782,
-      "grad_norm": 1.4556068181991577,
-      "learning_rate": 4.14151747655584e-06,
-      "loss": 0.1474,
-      "step": 3436
-    },
-    {
-      "epoch": 0.5860031435650158,
-      "grad_norm": 1.2817412614822388,
-      "learning_rate": 4.139812446717818e-06,
-      "loss": 0.1583,
-      "step": 3437
-    },
-    {
-      "epoch": 0.5861736420065535,
-      "grad_norm": 1.0311654806137085,
-      "learning_rate": 4.138107416879795e-06,
-      "loss": 0.1175,
-      "step": 3438
-    },
-    {
-      "epoch": 0.5863441404480912,
-      "grad_norm": 1.2356747388839722,
-      "learning_rate": 4.136402387041773e-06,
-      "loss": 0.1375,
-      "step": 3439
-    },
-    {
-      "epoch": 0.5865146388896288,
-      "grad_norm": 1.3934600353240967,
-      "learning_rate": 4.134697357203751e-06,
-      "loss": 0.1493,
-      "step": 3440
-    },
-    {
-      "epoch": 0.5866851373311666,
-      "grad_norm": 0.7090312838554382,
-      "learning_rate": 4.132992327365729e-06,
-      "loss": 0.082,
-      "step": 3441
-    },
-    {
-      "epoch": 0.5868556357727043,
-      "grad_norm": 1.185271978378296,
-      "learning_rate": 4.131287297527707e-06,
-      "loss": 0.1339,
-      "step": 3442
-    },
-    {
-      "epoch": 0.587026134214242,
-      "grad_norm": 1.2944258451461792,
-      "learning_rate": 4.129582267689685e-06,
-      "loss": 0.1124,
-      "step": 3443
-    },
-    {
-      "epoch": 0.5871966326557796,
-      "grad_norm": 1.0491424798965454,
-      "learning_rate": 4.127877237851663e-06,
-      "loss": 0.1042,
-      "step": 3444
-    },
-    {
-      "epoch": 0.5873671310973173,
-      "grad_norm": 1.1843968629837036,
-      "learning_rate": 4.126172208013641e-06,
-      "loss": 0.1105,
-      "step": 3445
-    },
-    {
-      "epoch": 0.587537629538855,
-      "grad_norm": 1.6263676881790161,
-      "learning_rate": 4.124467178175618e-06,
-      "loss": 0.1593,
-      "step": 3446
-    },
-    {
-      "epoch": 0.5877081279803926,
-      "grad_norm": 0.8808202743530273,
-      "learning_rate": 4.122762148337596e-06,
-      "loss": 0.1022,
-      "step": 3447
-    },
-    {
-      "epoch": 0.5878786264219303,
-      "grad_norm": 0.7366436123847961,
-      "learning_rate": 4.121057118499574e-06,
-      "loss": 0.0866,
-      "step": 3448
-    },
-    {
-      "epoch": 0.588049124863468,
-      "grad_norm": 1.401879072189331,
-      "learning_rate": 4.119352088661552e-06,
-      "loss": 0.2235,
-      "step": 3449
-    },
-    {
-      "epoch": 0.5882196233050058,
-      "grad_norm": 0.8640059232711792,
-      "learning_rate": 4.11764705882353e-06,
-      "loss": 0.0559,
-      "step": 3450
-    },
-    {
-      "epoch": 0.5883901217465434,
-      "grad_norm": 1.500951886177063,
-      "learning_rate": 4.115942028985507e-06,
-      "loss": 0.2105,
-      "step": 3451
-    },
-    {
-      "epoch": 0.5885606201880811,
-      "grad_norm": 0.9160787463188171,
-      "learning_rate": 4.114236999147486e-06,
-      "loss": 0.1177,
-      "step": 3452
-    },
-    {
-      "epoch": 0.5887311186296188,
-      "grad_norm": 1.2144598960876465,
-      "learning_rate": 4.112531969309464e-06,
-      "loss": 0.1491,
-      "step": 3453
-    },
-    {
-      "epoch": 0.5889016170711565,
-      "grad_norm": 1.3946925401687622,
-      "learning_rate": 4.110826939471441e-06,
-      "loss": 0.1667,
-      "step": 3454
-    },
-    {
-      "epoch": 0.5890721155126941,
-      "grad_norm": 1.3554019927978516,
-      "learning_rate": 4.109121909633419e-06,
-      "loss": 0.1413,
-      "step": 3455
-    },
-    {
-      "epoch": 0.5892426139542318,
-      "grad_norm": 1.6783015727996826,
-      "learning_rate": 4.107416879795397e-06,
-      "loss": 0.1542,
-      "step": 3456
-    },
-    {
-      "epoch": 0.5894131123957695,
-      "grad_norm": 1.0624831914901733,
-      "learning_rate": 4.105711849957375e-06,
-      "loss": 0.1066,
-      "step": 3457
-    },
-    {
-      "epoch": 0.5895836108373071,
-      "grad_norm": 1.2082974910736084,
-      "learning_rate": 4.104006820119353e-06,
-      "loss": 0.1176,
-      "step": 3458
-    },
-    {
-      "epoch": 0.5897541092788449,
-      "grad_norm": 0.7654959559440613,
-      "learning_rate": 4.10230179028133e-06,
-      "loss": 0.0742,
-      "step": 3459
-    },
-    {
-      "epoch": 0.5899246077203826,
-      "grad_norm": 1.610687255859375,
-      "learning_rate": 4.1005967604433085e-06,
-      "loss": 0.1725,
-      "step": 3460
-    },
-    {
-      "epoch": 0.5900951061619203,
-      "grad_norm": 0.9428104162216187,
-      "learning_rate": 4.0988917306052865e-06,
-      "loss": 0.1361,
-      "step": 3461
-    },
-    {
-      "epoch": 0.5902656046034579,
-      "grad_norm": 1.117551565170288,
-      "learning_rate": 4.0971867007672636e-06,
-      "loss": 0.0913,
-      "step": 3462
-    },
-    {
-      "epoch": 0.5904361030449956,
-      "grad_norm": 1.391464114189148,
-      "learning_rate": 4.0954816709292415e-06,
-      "loss": 0.1721,
-      "step": 3463
-    },
-    {
-      "epoch": 0.5906066014865333,
-      "grad_norm": 0.8190073370933533,
-      "learning_rate": 4.0937766410912195e-06,
-      "loss": 0.0675,
-      "step": 3464
-    },
-    {
-      "epoch": 0.5907770999280709,
-      "grad_norm": 1.8234949111938477,
-      "learning_rate": 4.092071611253197e-06,
-      "loss": 0.1997,
-      "step": 3465
-    },
-    {
-      "epoch": 0.5909475983696086,
-      "grad_norm": 1.2955584526062012,
-      "learning_rate": 4.090366581415175e-06,
-      "loss": 0.1551,
-      "step": 3466
-    },
-    {
-      "epoch": 0.5911180968111464,
-      "grad_norm": 1.6789677143096924,
-      "learning_rate": 4.0886615515771525e-06,
-      "loss": 0.1606,
-      "step": 3467
-    },
-    {
-      "epoch": 0.5912885952526841,
-      "grad_norm": 0.9646121859550476,
-      "learning_rate": 4.086956521739131e-06,
-      "loss": 0.0902,
-      "step": 3468
-    },
-    {
-      "epoch": 0.5914590936942217,
-      "grad_norm": 0.8161157369613647,
-      "learning_rate": 4.085251491901109e-06,
-      "loss": 0.0834,
-      "step": 3469
-    },
-    {
-      "epoch": 0.5916295921357594,
-      "grad_norm": 1.4160490036010742,
-      "learning_rate": 4.083546462063086e-06,
-      "loss": 0.1711,
-      "step": 3470
-    },
-    {
-      "epoch": 0.5918000905772971,
-      "grad_norm": 1.1606277227401733,
-      "learning_rate": 4.081841432225064e-06,
-      "loss": 0.1086,
-      "step": 3471
-    },
-    {
-      "epoch": 0.5919705890188347,
-      "grad_norm": 1.933803677558899,
-      "learning_rate": 4.080136402387042e-06,
-      "loss": 0.1797,
-      "step": 3472
-    },
-    {
-      "epoch": 0.5921410874603724,
-      "grad_norm": 0.8342519998550415,
-      "learning_rate": 4.07843137254902e-06,
-      "loss": 0.0939,
-      "step": 3473
-    },
-    {
-      "epoch": 0.5923115859019101,
-      "grad_norm": 1.1106159687042236,
-      "learning_rate": 4.076726342710997e-06,
-      "loss": 0.1578,
-      "step": 3474
-    },
-    {
-      "epoch": 0.5924820843434478,
-      "grad_norm": 1.416045904159546,
-      "learning_rate": 4.075021312872975e-06,
-      "loss": 0.1282,
-      "step": 3475
-    },
-    {
-      "epoch": 0.5926525827849854,
-      "grad_norm": 0.9018409252166748,
-      "learning_rate": 4.073316283034953e-06,
-      "loss": 0.0761,
-      "step": 3476
-    },
-    {
-      "epoch": 0.5928230812265232,
-      "grad_norm": 0.9520737528800964,
-      "learning_rate": 4.071611253196931e-06,
-      "loss": 0.1169,
-      "step": 3477
-    },
-    {
-      "epoch": 0.5929935796680609,
-      "grad_norm": 1.7352948188781738,
-      "learning_rate": 4.069906223358909e-06,
-      "loss": 0.1989,
-      "step": 3478
-    },
-    {
-      "epoch": 0.5931640781095985,
-      "grad_norm": 1.0018260478973389,
-      "learning_rate": 4.068201193520887e-06,
-      "loss": 0.0931,
-      "step": 3479
-    },
-    {
-      "epoch": 0.5933345765511362,
-      "grad_norm": 1.7456133365631104,
-      "learning_rate": 4.066496163682865e-06,
-      "loss": 0.1758,
-      "step": 3480
-    },
-    {
-      "epoch": 0.5935050749926739,
-      "grad_norm": 1.323087215423584,
-      "learning_rate": 4.064791133844843e-06,
-      "loss": 0.1263,
-      "step": 3481
-    },
-    {
-      "epoch": 0.5936755734342116,
-      "grad_norm": 1.2114142179489136,
-      "learning_rate": 4.06308610400682e-06,
-      "loss": 0.1273,
-      "step": 3482
-    },
-    {
-      "epoch": 0.5938460718757492,
-      "grad_norm": 0.7798213958740234,
-      "learning_rate": 4.061381074168798e-06,
-      "loss": 0.0675,
-      "step": 3483
-    },
-    {
-      "epoch": 0.5940165703172869,
-      "grad_norm": 1.5562611818313599,
-      "learning_rate": 4.059676044330776e-06,
-      "loss": 0.181,
-      "step": 3484
-    },
-    {
-      "epoch": 0.5941870687588247,
-      "grad_norm": 0.8951976299285889,
-      "learning_rate": 4.057971014492754e-06,
-      "loss": 0.0977,
-      "step": 3485
-    },
-    {
-      "epoch": 0.5943575672003623,
-      "grad_norm": 1.1890310049057007,
-      "learning_rate": 4.056265984654732e-06,
-      "loss": 0.1344,
-      "step": 3486
-    },
-    {
-      "epoch": 0.5945280656419,
-      "grad_norm": 1.9643477201461792,
-      "learning_rate": 4.05456095481671e-06,
-      "loss": 0.2614,
-      "step": 3487
-    },
-    {
-      "epoch": 0.5946985640834377,
-      "grad_norm": 1.0449378490447998,
-      "learning_rate": 4.052855924978688e-06,
-      "loss": 0.0596,
-      "step": 3488
-    },
-    {
-      "epoch": 0.5948690625249754,
-      "grad_norm": 1.2042781114578247,
-      "learning_rate": 4.051150895140666e-06,
-      "loss": 0.1624,
-      "step": 3489
-    },
-    {
-      "epoch": 0.595039560966513,
-      "grad_norm": 0.9169061779975891,
-      "learning_rate": 4.049445865302643e-06,
-      "loss": 0.1545,
-      "step": 3490
-    },
-    {
-      "epoch": 0.5952100594080507,
-      "grad_norm": 1.092411756515503,
-      "learning_rate": 4.047740835464621e-06,
-      "loss": 0.123,
-      "step": 3491
-    },
-    {
-      "epoch": 0.5953805578495884,
-      "grad_norm": 2.1680240631103516,
-      "learning_rate": 4.046035805626599e-06,
-      "loss": 0.2052,
-      "step": 3492
-    },
-    {
-      "epoch": 0.595551056291126,
-      "grad_norm": 1.5322246551513672,
-      "learning_rate": 4.0443307757885766e-06,
-      "loss": 0.1582,
-      "step": 3493
-    },
-    {
-      "epoch": 0.5957215547326637,
-      "grad_norm": 1.3674644231796265,
-      "learning_rate": 4.0426257459505545e-06,
-      "loss": 0.12,
-      "step": 3494
-    },
-    {
-      "epoch": 0.5958920531742015,
-      "grad_norm": 1.8770805597305298,
-      "learning_rate": 4.0409207161125325e-06,
-      "loss": 0.1854,
-      "step": 3495
-    },
-    {
-      "epoch": 0.5960625516157392,
-      "grad_norm": 1.1395719051361084,
-      "learning_rate": 4.03921568627451e-06,
-      "loss": 0.1042,
-      "step": 3496
-    },
-    {
-      "epoch": 0.5962330500572768,
-      "grad_norm": 1.364762544631958,
-      "learning_rate": 4.037510656436488e-06,
-      "loss": 0.1567,
-      "step": 3497
-    },
-    {
-      "epoch": 0.5964035484988145,
-      "grad_norm": 0.9239014387130737,
-      "learning_rate": 4.0358056265984655e-06,
-      "loss": 0.127,
-      "step": 3498
-    },
-    {
-      "epoch": 0.5965740469403522,
-      "grad_norm": 1.2227081060409546,
-      "learning_rate": 4.034100596760443e-06,
-      "loss": 0.123,
-      "step": 3499
-    },
-    {
-      "epoch": 0.5967445453818899,
-      "grad_norm": 1.3842462301254272,
-      "learning_rate": 4.032395566922421e-06,
-      "loss": 0.1772,
-      "step": 3500
-    },
-    {
-      "epoch": 0.5967445453818899,
-      "eval_f1_score": 0.3755656108597285,
-      "eval_loss": 0.13983075320720673,
-      "eval_runtime": 182.7252,
-      "eval_samples_per_second": 54.727,
-      "eval_steps_per_second": 3.42,
-      "step": 3500
-    },
-    {
-      "epoch": 0.5969150438234275,
-      "grad_norm": 1.9180755615234375,
-      "learning_rate": 4.030690537084399e-06,
-      "loss": 0.1808,
-      "step": 3501
-    },
-    {
-      "epoch": 0.5970855422649652,
-      "grad_norm": 1.2909221649169922,
-      "learning_rate": 4.028985507246377e-06,
-      "loss": 0.0999,
-      "step": 3502
-    },
-    {
-      "epoch": 0.597256040706503,
-      "grad_norm": 1.1232291460037231,
-      "learning_rate": 4.027280477408355e-06,
-      "loss": 0.1751,
-      "step": 3503
-    },
-    {
-      "epoch": 0.5974265391480406,
-      "grad_norm": 1.0960628986358643,
-      "learning_rate": 4.025575447570333e-06,
-      "loss": 0.0742,
-      "step": 3504
-    },
-    {
-      "epoch": 0.5975970375895783,
-      "grad_norm": 1.700650691986084,
-      "learning_rate": 4.023870417732311e-06,
-      "loss": 0.1958,
-      "step": 3505
-    },
-    {
-      "epoch": 0.597767536031116,
-      "grad_norm": 1.1192090511322021,
-      "learning_rate": 4.022165387894288e-06,
-      "loss": 0.084,
-      "step": 3506
-    },
-    {
-      "epoch": 0.5979380344726537,
-      "grad_norm": 0.9199531078338623,
-      "learning_rate": 4.020460358056266e-06,
-      "loss": 0.0806,
-      "step": 3507
-    },
-    {
-      "epoch": 0.5981085329141913,
-      "grad_norm": 0.8182160258293152,
-      "learning_rate": 4.018755328218244e-06,
-      "loss": 0.0819,
-      "step": 3508
-    },
-    {
-      "epoch": 0.598279031355729,
-      "grad_norm": 1.237420916557312,
-      "learning_rate": 4.017050298380222e-06,
-      "loss": 0.1106,
-      "step": 3509
-    },
-    {
-      "epoch": 0.5984495297972667,
-      "grad_norm": 0.9624441266059875,
-      "learning_rate": 4.0153452685422e-06,
-      "loss": 0.0933,
-      "step": 3510
-    },
-    {
-      "epoch": 0.5986200282388043,
-      "grad_norm": 0.7896148562431335,
-      "learning_rate": 4.013640238704178e-06,
-      "loss": 0.0938,
-      "step": 3511
-    },
-    {
-      "epoch": 0.598790526680342,
-      "grad_norm": 1.687626838684082,
-      "learning_rate": 4.011935208866156e-06,
-      "loss": 0.2776,
-      "step": 3512
-    },
-    {
-      "epoch": 0.5989610251218798,
-      "grad_norm": 0.9339796304702759,
-      "learning_rate": 4.010230179028133e-06,
-      "loss": 0.1304,
-      "step": 3513
-    },
-    {
-      "epoch": 0.5991315235634175,
-      "grad_norm": 0.9085601568222046,
-      "learning_rate": 4.008525149190111e-06,
-      "loss": 0.0723,
-      "step": 3514
-    },
-    {
-      "epoch": 0.5993020220049551,
-      "grad_norm": 1.3926743268966675,
-      "learning_rate": 4.006820119352089e-06,
-      "loss": 0.1761,
-      "step": 3515
-    },
-    {
-      "epoch": 0.5994725204464928,
-      "grad_norm": 0.9031898379325867,
-      "learning_rate": 4.005115089514067e-06,
-      "loss": 0.0809,
-      "step": 3516
-    },
-    {
-      "epoch": 0.5996430188880305,
-      "grad_norm": 0.839905321598053,
-      "learning_rate": 4.003410059676045e-06,
-      "loss": 0.0704,
-      "step": 3517
-    },
-    {
-      "epoch": 0.5998135173295681,
-      "grad_norm": 1.688146710395813,
-      "learning_rate": 4.001705029838022e-06,
-      "loss": 0.1111,
-      "step": 3518
-    },
-    {
-      "epoch": 0.5999840157711058,
-      "grad_norm": 1.9072600603103638,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 0.1669,
-      "step": 3519
-    },
-    {
-      "epoch": 0.6001545142126435,
-      "grad_norm": 1.4360688924789429,
-      "learning_rate": 3.998294970161979e-06,
-      "loss": 0.1837,
-      "step": 3520
-    },
-    {
-      "epoch": 0.6003250126541813,
-      "grad_norm": 1.0013833045959473,
-      "learning_rate": 3.996589940323956e-06,
-      "loss": 0.0866,
-      "step": 3521
-    },
-    {
-      "epoch": 0.6004955110957189,
-      "grad_norm": 2.101608991622925,
-      "learning_rate": 3.994884910485934e-06,
-      "loss": 0.1842,
-      "step": 3522
-    },
-    {
-      "epoch": 0.6006660095372566,
-      "grad_norm": 1.478049874305725,
-      "learning_rate": 3.993179880647912e-06,
-      "loss": 0.1477,
-      "step": 3523
-    },
-    {
-      "epoch": 0.6008365079787943,
-      "grad_norm": 1.2308162450790405,
-      "learning_rate": 3.9914748508098896e-06,
-      "loss": 0.1252,
-      "step": 3524
-    },
-    {
-      "epoch": 0.6010070064203319,
-      "grad_norm": 1.1021760702133179,
-      "learning_rate": 3.9897698209718675e-06,
-      "loss": 0.0901,
-      "step": 3525
-    },
-    {
-      "epoch": 0.6011775048618696,
-      "grad_norm": 1.6949081420898438,
-      "learning_rate": 3.988064791133845e-06,
-      "loss": 0.2353,
-      "step": 3526
-    },
-    {
-      "epoch": 0.6013480033034073,
-      "grad_norm": 1.4775804281234741,
-      "learning_rate": 3.986359761295823e-06,
-      "loss": 0.2189,
-      "step": 3527
-    },
-    {
-      "epoch": 0.601518501744945,
-      "grad_norm": 1.1562957763671875,
-      "learning_rate": 3.984654731457801e-06,
-      "loss": 0.1427,
-      "step": 3528
-    },
-    {
-      "epoch": 0.6016890001864826,
-      "grad_norm": 1.0964555740356445,
-      "learning_rate": 3.9829497016197784e-06,
-      "loss": 0.1065,
-      "step": 3529
-    },
-    {
-      "epoch": 0.6018594986280204,
-      "grad_norm": 0.8958936929702759,
-      "learning_rate": 3.981244671781756e-06,
-      "loss": 0.0687,
-      "step": 3530
-    },
-    {
-      "epoch": 0.6020299970695581,
-      "grad_norm": 1.092345118522644,
-      "learning_rate": 3.979539641943734e-06,
-      "loss": 0.105,
-      "step": 3531
-    },
-    {
-      "epoch": 0.6022004955110957,
-      "grad_norm": 0.8545036911964417,
-      "learning_rate": 3.977834612105712e-06,
-      "loss": 0.0719,
-      "step": 3532
-    },
-    {
-      "epoch": 0.6023709939526334,
-      "grad_norm": 1.1633192300796509,
-      "learning_rate": 3.97612958226769e-06,
-      "loss": 0.084,
-      "step": 3533
-    },
-    {
-      "epoch": 0.6025414923941711,
-      "grad_norm": 1.2050695419311523,
-      "learning_rate": 3.974424552429667e-06,
-      "loss": 0.1153,
-      "step": 3534
-    },
-    {
-      "epoch": 0.6027119908357088,
-      "grad_norm": 1.1858649253845215,
-      "learning_rate": 3.972719522591646e-06,
-      "loss": 0.1457,
-      "step": 3535
-    },
-    {
-      "epoch": 0.6028824892772464,
-      "grad_norm": 1.2307145595550537,
-      "learning_rate": 3.971014492753624e-06,
-      "loss": 0.1005,
-      "step": 3536
-    },
-    {
-      "epoch": 0.6030529877187841,
-      "grad_norm": 1.0503499507904053,
-      "learning_rate": 3.969309462915601e-06,
-      "loss": 0.1343,
-      "step": 3537
-    },
-    {
-      "epoch": 0.6032234861603218,
-      "grad_norm": 1.2630836963653564,
-      "learning_rate": 3.967604433077579e-06,
-      "loss": 0.0734,
-      "step": 3538
-    },
-    {
-      "epoch": 0.6033939846018594,
-      "grad_norm": 1.0602471828460693,
-      "learning_rate": 3.965899403239557e-06,
-      "loss": 0.1298,
-      "step": 3539
-    },
-    {
-      "epoch": 0.6035644830433972,
-      "grad_norm": 1.1123082637786865,
-      "learning_rate": 3.964194373401535e-06,
-      "loss": 0.1645,
-      "step": 3540
-    },
-    {
-      "epoch": 0.6037349814849349,
-      "grad_norm": 0.9318695068359375,
-      "learning_rate": 3.962489343563513e-06,
-      "loss": 0.1145,
-      "step": 3541
-    },
-    {
-      "epoch": 0.6039054799264726,
-      "grad_norm": 1.139296054840088,
-      "learning_rate": 3.96078431372549e-06,
-      "loss": 0.1015,
-      "step": 3542
-    },
-    {
-      "epoch": 0.6040759783680102,
-      "grad_norm": 1.1052963733673096,
-      "learning_rate": 3.959079283887468e-06,
-      "loss": 0.0879,
-      "step": 3543
-    },
-    {
-      "epoch": 0.6042464768095479,
-      "grad_norm": 1.128821611404419,
-      "learning_rate": 3.957374254049447e-06,
-      "loss": 0.1126,
-      "step": 3544
-    },
-    {
-      "epoch": 0.6044169752510856,
-      "grad_norm": 0.9472864866256714,
-      "learning_rate": 3.955669224211424e-06,
-      "loss": 0.072,
-      "step": 3545
-    },
-    {
-      "epoch": 0.6045874736926232,
-      "grad_norm": 0.9522503018379211,
-      "learning_rate": 3.953964194373402e-06,
-      "loss": 0.0858,
-      "step": 3546
-    },
-    {
-      "epoch": 0.6047579721341609,
-      "grad_norm": 1.2889169454574585,
-      "learning_rate": 3.95225916453538e-06,
-      "loss": 0.163,
-      "step": 3547
-    },
-    {
-      "epoch": 0.6049284705756987,
-      "grad_norm": 1.1692979335784912,
-      "learning_rate": 3.950554134697358e-06,
-      "loss": 0.1005,
-      "step": 3548
-    },
-    {
-      "epoch": 0.6050989690172364,
-      "grad_norm": 0.6599539518356323,
-      "learning_rate": 3.948849104859335e-06,
-      "loss": 0.0532,
-      "step": 3549
-    },
-    {
-      "epoch": 0.605269467458774,
-      "grad_norm": 1.4034006595611572,
-      "learning_rate": 3.947144075021313e-06,
-      "loss": 0.0865,
-      "step": 3550
-    },
-    {
-      "epoch": 0.6054399659003117,
-      "grad_norm": 0.9632552862167358,
-      "learning_rate": 3.945439045183291e-06,
-      "loss": 0.0768,
-      "step": 3551
-    },
-    {
-      "epoch": 0.6056104643418494,
-      "grad_norm": 1.3031929731369019,
-      "learning_rate": 3.943734015345269e-06,
-      "loss": 0.1316,
-      "step": 3552
-    },
-    {
-      "epoch": 0.6057809627833871,
-      "grad_norm": 0.8928091526031494,
-      "learning_rate": 3.942028985507247e-06,
-      "loss": 0.0637,
-      "step": 3553
-    },
-    {
-      "epoch": 0.6059514612249247,
-      "grad_norm": 1.0940797328948975,
-      "learning_rate": 3.940323955669225e-06,
-      "loss": 0.1166,
-      "step": 3554
-    },
-    {
-      "epoch": 0.6061219596664624,
-      "grad_norm": 0.6712972521781921,
-      "learning_rate": 3.9386189258312025e-06,
-      "loss": 0.0324,
-      "step": 3555
-    },
-    {
-      "epoch": 0.6062924581080001,
-      "grad_norm": 1.1135120391845703,
-      "learning_rate": 3.9369138959931805e-06,
-      "loss": 0.1124,
-      "step": 3556
-    },
-    {
-      "epoch": 0.6064629565495377,
-      "grad_norm": 1.8652747869491577,
-      "learning_rate": 3.935208866155158e-06,
-      "loss": 0.1985,
-      "step": 3557
-    },
-    {
-      "epoch": 0.6066334549910755,
-      "grad_norm": 1.6062350273132324,
-      "learning_rate": 3.9335038363171355e-06,
-      "loss": 0.1151,
-      "step": 3558
-    },
-    {
-      "epoch": 0.6068039534326132,
-      "grad_norm": 0.5442900657653809,
-      "learning_rate": 3.9317988064791135e-06,
-      "loss": 0.0433,
-      "step": 3559
-    },
-    {
-      "epoch": 0.6069744518741509,
-      "grad_norm": 1.7723592519760132,
-      "learning_rate": 3.9300937766410914e-06,
-      "loss": 0.2031,
-      "step": 3560
-    },
-    {
-      "epoch": 0.6071449503156885,
-      "grad_norm": 1.5068206787109375,
-      "learning_rate": 3.928388746803069e-06,
-      "loss": 0.1334,
-      "step": 3561
-    },
-    {
-      "epoch": 0.6073154487572262,
-      "grad_norm": 1.8842564821243286,
-      "learning_rate": 3.926683716965047e-06,
-      "loss": 0.2207,
-      "step": 3562
-    },
-    {
-      "epoch": 0.6074859471987639,
-      "grad_norm": 1.6659555435180664,
-      "learning_rate": 3.924978687127025e-06,
-      "loss": 0.1166,
-      "step": 3563
-    },
-    {
-      "epoch": 0.6076564456403015,
-      "grad_norm": 1.5390290021896362,
-      "learning_rate": 3.923273657289003e-06,
-      "loss": 0.2027,
-      "step": 3564
-    },
-    {
-      "epoch": 0.6078269440818392,
-      "grad_norm": 1.0417344570159912,
-      "learning_rate": 3.92156862745098e-06,
-      "loss": 0.0851,
-      "step": 3565
-    },
-    {
-      "epoch": 0.607997442523377,
-      "grad_norm": 1.357947587966919,
-      "learning_rate": 3.919863597612958e-06,
-      "loss": 0.1115,
-      "step": 3566
-    },
-    {
-      "epoch": 0.6081679409649147,
-      "grad_norm": 1.5044528245925903,
-      "learning_rate": 3.918158567774936e-06,
-      "loss": 0.1992,
-      "step": 3567
-    },
-    {
-      "epoch": 0.6083384394064523,
-      "grad_norm": 0.89848792552948,
-      "learning_rate": 3.916453537936914e-06,
-      "loss": 0.1002,
-      "step": 3568
-    },
-    {
-      "epoch": 0.60850893784799,
-      "grad_norm": 0.9809942245483398,
-      "learning_rate": 3.914748508098892e-06,
-      "loss": 0.1006,
-      "step": 3569
-    },
-    {
-      "epoch": 0.6086794362895277,
-      "grad_norm": 1.763635516166687,
-      "learning_rate": 3.91304347826087e-06,
-      "loss": 0.1184,
-      "step": 3570
-    },
-    {
-      "epoch": 0.6088499347310653,
-      "grad_norm": 1.5588911771774292,
-      "learning_rate": 3.911338448422848e-06,
-      "loss": 0.2385,
-      "step": 3571
-    },
-    {
-      "epoch": 0.609020433172603,
-      "grad_norm": 0.902176022529602,
-      "learning_rate": 3.909633418584826e-06,
-      "loss": 0.0772,
-      "step": 3572
-    },
-    {
-      "epoch": 0.6091909316141407,
-      "grad_norm": 1.4685441255569458,
-      "learning_rate": 3.907928388746803e-06,
-      "loss": 0.2082,
-      "step": 3573
-    },
-    {
-      "epoch": 0.6093614300556784,
-      "grad_norm": 1.2932947874069214,
-      "learning_rate": 3.906223358908781e-06,
-      "loss": 0.1485,
-      "step": 3574
-    },
-    {
-      "epoch": 0.609531928497216,
-      "grad_norm": 1.1713253259658813,
-      "learning_rate": 3.904518329070759e-06,
-      "loss": 0.101,
-      "step": 3575
-    },
-    {
-      "epoch": 0.6097024269387538,
-      "grad_norm": 1.101948618888855,
-      "learning_rate": 3.902813299232737e-06,
-      "loss": 0.0719,
-      "step": 3576
-    },
-    {
-      "epoch": 0.6098729253802915,
-      "grad_norm": 1.0822521448135376,
-      "learning_rate": 3.901108269394715e-06,
-      "loss": 0.1476,
-      "step": 3577
-    },
-    {
-      "epoch": 0.6100434238218291,
-      "grad_norm": 1.064866304397583,
-      "learning_rate": 3.899403239556693e-06,
-      "loss": 0.1732,
-      "step": 3578
-    },
-    {
-      "epoch": 0.6102139222633668,
-      "grad_norm": 1.401872992515564,
-      "learning_rate": 3.897698209718671e-06,
-      "loss": 0.1713,
-      "step": 3579
-    },
-    {
-      "epoch": 0.6103844207049045,
-      "grad_norm": 0.7649058103561401,
-      "learning_rate": 3.895993179880649e-06,
-      "loss": 0.079,
-      "step": 3580
-    },
-    {
-      "epoch": 0.6105549191464422,
-      "grad_norm": 0.8753597736358643,
-      "learning_rate": 3.894288150042626e-06,
-      "loss": 0.1105,
-      "step": 3581
-    },
-    {
-      "epoch": 0.6107254175879798,
-      "grad_norm": 1.8067176342010498,
-      "learning_rate": 3.892583120204604e-06,
-      "loss": 0.1334,
-      "step": 3582
-    },
-    {
-      "epoch": 0.6108959160295175,
-      "grad_norm": 1.6385232210159302,
-      "learning_rate": 3.890878090366582e-06,
-      "loss": 0.2134,
-      "step": 3583
-    },
-    {
-      "epoch": 0.6110664144710553,
-      "grad_norm": 1.6331698894500732,
-      "learning_rate": 3.88917306052856e-06,
-      "loss": 0.1875,
-      "step": 3584
-    },
-    {
-      "epoch": 0.6112369129125929,
-      "grad_norm": 1.0041797161102295,
-      "learning_rate": 3.887468030690537e-06,
-      "loss": 0.1336,
-      "step": 3585
-    },
-    {
-      "epoch": 0.6114074113541306,
-      "grad_norm": 1.4989748001098633,
-      "learning_rate": 3.8857630008525155e-06,
-      "loss": 0.1608,
-      "step": 3586
-    },
-    {
-      "epoch": 0.6115779097956683,
-      "grad_norm": 0.7531823515892029,
-      "learning_rate": 3.8840579710144935e-06,
-      "loss": 0.0711,
-      "step": 3587
-    },
-    {
-      "epoch": 0.611748408237206,
-      "grad_norm": 1.2867484092712402,
-      "learning_rate": 3.882352941176471e-06,
-      "loss": 0.1445,
-      "step": 3588
-    },
-    {
-      "epoch": 0.6119189066787436,
-      "grad_norm": 1.0655556917190552,
-      "learning_rate": 3.8806479113384485e-06,
-      "loss": 0.1091,
-      "step": 3589
-    },
-    {
-      "epoch": 0.6120894051202813,
-      "grad_norm": 1.1923772096633911,
-      "learning_rate": 3.8789428815004265e-06,
-      "loss": 0.1657,
-      "step": 3590
-    },
-    {
-      "epoch": 0.612259903561819,
-      "grad_norm": 0.8422601222991943,
-      "learning_rate": 3.8772378516624044e-06,
-      "loss": 0.1081,
-      "step": 3591
-    },
-    {
-      "epoch": 0.6124304020033566,
-      "grad_norm": 0.8882838487625122,
-      "learning_rate": 3.875532821824382e-06,
-      "loss": 0.1266,
-      "step": 3592
-    },
-    {
-      "epoch": 0.6126009004448943,
-      "grad_norm": 0.9292901158332825,
-      "learning_rate": 3.8738277919863595e-06,
-      "loss": 0.097,
-      "step": 3593
-    },
-    {
-      "epoch": 0.6127713988864321,
-      "grad_norm": 1.0757999420166016,
-      "learning_rate": 3.872122762148338e-06,
-      "loss": 0.1103,
-      "step": 3594
-    },
-    {
-      "epoch": 0.6129418973279698,
-      "grad_norm": 0.8698549866676331,
-      "learning_rate": 3.870417732310316e-06,
-      "loss": 0.0912,
-      "step": 3595
-    },
-    {
-      "epoch": 0.6131123957695074,
-      "grad_norm": 1.0098843574523926,
-      "learning_rate": 3.868712702472293e-06,
-      "loss": 0.1067,
-      "step": 3596
-    },
-    {
-      "epoch": 0.6132828942110451,
-      "grad_norm": 0.9158385992050171,
-      "learning_rate": 3.867007672634271e-06,
-      "loss": 0.1195,
-      "step": 3597
-    },
-    {
-      "epoch": 0.6134533926525828,
-      "grad_norm": 1.382043480873108,
-      "learning_rate": 3.865302642796249e-06,
-      "loss": 0.1285,
-      "step": 3598
-    },
-    {
-      "epoch": 0.6136238910941205,
-      "grad_norm": 1.214117407798767,
-      "learning_rate": 3.863597612958227e-06,
-      "loss": 0.1821,
-      "step": 3599
-    },
-    {
-      "epoch": 0.6137943895356581,
-      "grad_norm": 1.2158795595169067,
-      "learning_rate": 3.861892583120205e-06,
-      "loss": 0.0981,
-      "step": 3600
-    },
-    {
-      "epoch": 0.6137943895356581,
-      "eval_f1_score": 0.37468354430379747,
-      "eval_loss": 0.13799934089183807,
-      "eval_runtime": 182.6656,
-      "eval_samples_per_second": 54.745,
-      "eval_steps_per_second": 3.422,
-      "step": 3600
-    },
-    {
-      "epoch": 0.6139648879771958,
-      "grad_norm": 1.1329641342163086,
-      "learning_rate": 3.860187553282182e-06,
-      "loss": 0.1213,
-      "step": 3601
-    },
-    {
-      "epoch": 0.6141353864187336,
-      "grad_norm": 1.107250452041626,
-      "learning_rate": 3.858482523444161e-06,
-      "loss": 0.1135,
-      "step": 3602
-    },
-    {
-      "epoch": 0.6143058848602712,
-      "grad_norm": 1.2173339128494263,
-      "learning_rate": 3.856777493606139e-06,
-      "loss": 0.1582,
-      "step": 3603
-    },
-    {
-      "epoch": 0.6144763833018089,
-      "grad_norm": 0.8356215953826904,
-      "learning_rate": 3.855072463768116e-06,
-      "loss": 0.0714,
-      "step": 3604
-    },
-    {
-      "epoch": 0.6146468817433466,
-      "grad_norm": 1.1072112321853638,
-      "learning_rate": 3.853367433930094e-06,
-      "loss": 0.1626,
-      "step": 3605
-    },
-    {
-      "epoch": 0.6148173801848843,
-      "grad_norm": 1.1094462871551514,
-      "learning_rate": 3.851662404092072e-06,
-      "loss": 0.1109,
-      "step": 3606
-    },
-    {
-      "epoch": 0.6149878786264219,
-      "grad_norm": 0.8917166590690613,
-      "learning_rate": 3.84995737425405e-06,
-      "loss": 0.0976,
-      "step": 3607
-    },
-    {
-      "epoch": 0.6151583770679596,
-      "grad_norm": 0.9127071499824524,
-      "learning_rate": 3.848252344416028e-06,
-      "loss": 0.0789,
-      "step": 3608
-    },
-    {
-      "epoch": 0.6153288755094973,
-      "grad_norm": 0.988960325717926,
-      "learning_rate": 3.846547314578005e-06,
-      "loss": 0.1172,
-      "step": 3609
-    },
-    {
-      "epoch": 0.6154993739510349,
-      "grad_norm": 1.2566884756088257,
-      "learning_rate": 3.844842284739983e-06,
-      "loss": 0.1606,
-      "step": 3610
-    },
-    {
-      "epoch": 0.6156698723925726,
-      "grad_norm": 0.613823413848877,
-      "learning_rate": 3.843137254901962e-06,
-      "loss": 0.0685,
-      "step": 3611
-    },
-    {
-      "epoch": 0.6158403708341104,
-      "grad_norm": 1.1060439348220825,
-      "learning_rate": 3.841432225063939e-06,
-      "loss": 0.0973,
-      "step": 3612
-    },
-    {
-      "epoch": 0.6160108692756481,
-      "grad_norm": 1.327811360359192,
-      "learning_rate": 3.839727195225917e-06,
-      "loss": 0.1621,
-      "step": 3613
-    },
-    {
-      "epoch": 0.6161813677171857,
-      "grad_norm": 0.819416344165802,
-      "learning_rate": 3.838022165387895e-06,
-      "loss": 0.0598,
-      "step": 3614
-    },
-    {
-      "epoch": 0.6163518661587234,
-      "grad_norm": 0.9425433278083801,
-      "learning_rate": 3.836317135549873e-06,
-      "loss": 0.0993,
-      "step": 3615
-    },
-    {
-      "epoch": 0.6165223646002611,
-      "grad_norm": 0.7881563305854797,
-      "learning_rate": 3.834612105711851e-06,
-      "loss": 0.0835,
-      "step": 3616
-    },
-    {
-      "epoch": 0.6166928630417987,
-      "grad_norm": 1.2177718877792358,
-      "learning_rate": 3.832907075873828e-06,
-      "loss": 0.1378,
-      "step": 3617
-    },
-    {
-      "epoch": 0.6168633614833364,
-      "grad_norm": 1.150881290435791,
-      "learning_rate": 3.831202046035806e-06,
-      "loss": 0.1218,
-      "step": 3618
-    },
-    {
-      "epoch": 0.6170338599248741,
-      "grad_norm": 0.942000150680542,
-      "learning_rate": 3.8294970161977844e-06,
-      "loss": 0.1079,
-      "step": 3619
-    },
-    {
-      "epoch": 0.6172043583664119,
-      "grad_norm": 1.0197066068649292,
-      "learning_rate": 3.8277919863597615e-06,
-      "loss": 0.1312,
-      "step": 3620
-    },
-    {
-      "epoch": 0.6173748568079495,
-      "grad_norm": 1.200119137763977,
-      "learning_rate": 3.8260869565217395e-06,
-      "loss": 0.1345,
-      "step": 3621
-    },
-    {
-      "epoch": 0.6175453552494872,
-      "grad_norm": 0.7664061188697815,
-      "learning_rate": 3.824381926683717e-06,
-      "loss": 0.0989,
-      "step": 3622
-    },
-    {
-      "epoch": 0.6177158536910249,
-      "grad_norm": 1.594568133354187,
-      "learning_rate": 3.822676896845695e-06,
-      "loss": 0.1371,
-      "step": 3623
-    },
-    {
-      "epoch": 0.6178863521325625,
-      "grad_norm": 0.9898309111595154,
-      "learning_rate": 3.8209718670076725e-06,
-      "loss": 0.1059,
-      "step": 3624
-    },
-    {
-      "epoch": 0.6180568505741002,
-      "grad_norm": 1.4782969951629639,
-      "learning_rate": 3.81926683716965e-06,
-      "loss": 0.117,
-      "step": 3625
-    },
-    {
-      "epoch": 0.6182273490156379,
-      "grad_norm": 1.10325288772583,
-      "learning_rate": 3.817561807331628e-06,
-      "loss": 0.1597,
-      "step": 3626
-    },
-    {
-      "epoch": 0.6183978474571756,
-      "grad_norm": 0.9282763600349426,
-      "learning_rate": 3.815856777493606e-06,
-      "loss": 0.0651,
-      "step": 3627
-    },
-    {
-      "epoch": 0.6185683458987132,
-      "grad_norm": 1.0336819887161255,
-      "learning_rate": 3.8141517476555847e-06,
-      "loss": 0.063,
-      "step": 3628
-    },
-    {
-      "epoch": 0.618738844340251,
-      "grad_norm": 0.8412141799926758,
-      "learning_rate": 3.812446717817562e-06,
-      "loss": 0.0629,
-      "step": 3629
-    },
-    {
-      "epoch": 0.6189093427817887,
-      "grad_norm": 0.7322096228599548,
-      "learning_rate": 3.81074168797954e-06,
-      "loss": 0.0489,
-      "step": 3630
-    },
-    {
-      "epoch": 0.6190798412233263,
-      "grad_norm": 1.0511646270751953,
-      "learning_rate": 3.8090366581415177e-06,
-      "loss": 0.0878,
-      "step": 3631
-    },
-    {
-      "epoch": 0.619250339664864,
-      "grad_norm": 1.8431638479232788,
-      "learning_rate": 3.8073316283034956e-06,
-      "loss": 0.1911,
-      "step": 3632
-    },
-    {
-      "epoch": 0.6194208381064017,
-      "grad_norm": 1.5251615047454834,
-      "learning_rate": 3.805626598465473e-06,
-      "loss": 0.0968,
-      "step": 3633
-    },
-    {
-      "epoch": 0.6195913365479394,
-      "grad_norm": 0.9593001008033752,
-      "learning_rate": 3.803921568627451e-06,
-      "loss": 0.1134,
-      "step": 3634
-    },
-    {
-      "epoch": 0.619761834989477,
-      "grad_norm": 1.0317368507385254,
-      "learning_rate": 3.802216538789429e-06,
-      "loss": 0.0664,
-      "step": 3635
-    },
-    {
-      "epoch": 0.6199323334310147,
-      "grad_norm": 1.1496648788452148,
-      "learning_rate": 3.800511508951407e-06,
-      "loss": 0.1018,
-      "step": 3636
-    },
-    {
-      "epoch": 0.6201028318725524,
-      "grad_norm": 1.7094920873641968,
-      "learning_rate": 3.798806479113385e-06,
-      "loss": 0.1388,
-      "step": 3637
-    },
-    {
-      "epoch": 0.62027333031409,
-      "grad_norm": 1.065308928489685,
-      "learning_rate": 3.797101449275363e-06,
-      "loss": 0.1489,
-      "step": 3638
-    },
-    {
-      "epoch": 0.6204438287556278,
-      "grad_norm": 0.7095522880554199,
-      "learning_rate": 3.7953964194373404e-06,
-      "loss": 0.0547,
-      "step": 3639
-    },
-    {
-      "epoch": 0.6206143271971655,
-      "grad_norm": 1.4230934381484985,
-      "learning_rate": 3.7936913895993184e-06,
-      "loss": 0.1406,
-      "step": 3640
-    },
-    {
-      "epoch": 0.6207848256387032,
-      "grad_norm": 0.6358377933502197,
-      "learning_rate": 3.791986359761296e-06,
-      "loss": 0.0872,
-      "step": 3641
-    },
-    {
-      "epoch": 0.6209553240802408,
-      "grad_norm": 1.4998244047164917,
-      "learning_rate": 3.790281329923274e-06,
-      "loss": 0.1274,
-      "step": 3642
-    },
-    {
-      "epoch": 0.6211258225217785,
-      "grad_norm": 1.0013731718063354,
-      "learning_rate": 3.7885763000852514e-06,
-      "loss": 0.0714,
-      "step": 3643
-    },
-    {
-      "epoch": 0.6212963209633162,
-      "grad_norm": 1.1653109788894653,
-      "learning_rate": 3.7868712702472297e-06,
-      "loss": 0.0909,
-      "step": 3644
-    },
-    {
-      "epoch": 0.6214668194048539,
-      "grad_norm": 0.8749520182609558,
-      "learning_rate": 3.7851662404092077e-06,
-      "loss": 0.0743,
-      "step": 3645
-    },
-    {
-      "epoch": 0.6216373178463915,
-      "grad_norm": 1.5417793989181519,
-      "learning_rate": 3.7834612105711856e-06,
-      "loss": 0.1736,
-      "step": 3646
-    },
-    {
-      "epoch": 0.6218078162879292,
-      "grad_norm": 1.2029529809951782,
-      "learning_rate": 3.781756180733163e-06,
-      "loss": 0.1049,
-      "step": 3647
-    },
-    {
-      "epoch": 0.621978314729467,
-      "grad_norm": 1.119659423828125,
-      "learning_rate": 3.780051150895141e-06,
-      "loss": 0.0916,
-      "step": 3648
-    },
-    {
-      "epoch": 0.6221488131710046,
-      "grad_norm": 1.0468127727508545,
-      "learning_rate": 3.7783461210571186e-06,
-      "loss": 0.111,
-      "step": 3649
-    },
-    {
-      "epoch": 0.6223193116125423,
-      "grad_norm": 1.7078468799591064,
-      "learning_rate": 3.7766410912190966e-06,
-      "loss": 0.1596,
-      "step": 3650
-    },
-    {
-      "epoch": 0.62248981005408,
-      "grad_norm": 1.146774172782898,
-      "learning_rate": 3.774936061381074e-06,
-      "loss": 0.1152,
-      "step": 3651
-    },
-    {
-      "epoch": 0.6226603084956177,
-      "grad_norm": 0.7872032523155212,
-      "learning_rate": 3.773231031543052e-06,
-      "loss": 0.0634,
-      "step": 3652
-    },
-    {
-      "epoch": 0.6228308069371553,
-      "grad_norm": 1.1197142601013184,
-      "learning_rate": 3.7715260017050304e-06,
-      "loss": 0.1147,
-      "step": 3653
-    },
-    {
-      "epoch": 0.623001305378693,
-      "grad_norm": 0.7918756604194641,
-      "learning_rate": 3.769820971867008e-06,
-      "loss": 0.078,
-      "step": 3654
-    },
-    {
-      "epoch": 0.6231718038202307,
-      "grad_norm": 1.0666242837905884,
-      "learning_rate": 3.768115942028986e-06,
-      "loss": 0.152,
-      "step": 3655
-    },
-    {
-      "epoch": 0.6233423022617683,
-      "grad_norm": 2.1137020587921143,
-      "learning_rate": 3.766410912190964e-06,
-      "loss": 0.2122,
-      "step": 3656
-    },
-    {
-      "epoch": 0.6235128007033061,
-      "grad_norm": 0.9335868954658508,
-      "learning_rate": 3.7647058823529414e-06,
-      "loss": 0.1008,
-      "step": 3657
-    },
-    {
-      "epoch": 0.6236832991448438,
-      "grad_norm": 1.6401317119598389,
-      "learning_rate": 3.7630008525149193e-06,
-      "loss": 0.1642,
-      "step": 3658
-    },
-    {
-      "epoch": 0.6238537975863815,
-      "grad_norm": 1.146842122077942,
-      "learning_rate": 3.761295822676897e-06,
-      "loss": 0.1184,
-      "step": 3659
-    },
-    {
-      "epoch": 0.6240242960279191,
-      "grad_norm": 0.8928496241569519,
-      "learning_rate": 3.7595907928388748e-06,
-      "loss": 0.0952,
-      "step": 3660
-    },
-    {
-      "epoch": 0.6241947944694568,
-      "grad_norm": 1.258764386177063,
-      "learning_rate": 3.757885763000853e-06,
-      "loss": 0.156,
-      "step": 3661
-    },
-    {
-      "epoch": 0.6243652929109945,
-      "grad_norm": 1.6915022134780884,
-      "learning_rate": 3.7561807331628307e-06,
-      "loss": 0.1303,
-      "step": 3662
-    },
-    {
-      "epoch": 0.6245357913525321,
-      "grad_norm": 0.8792423605918884,
-      "learning_rate": 3.7544757033248086e-06,
-      "loss": 0.0861,
-      "step": 3663
-    },
-    {
-      "epoch": 0.6247062897940698,
-      "grad_norm": 1.505020022392273,
-      "learning_rate": 3.7527706734867866e-06,
-      "loss": 0.1456,
-      "step": 3664
-    },
-    {
-      "epoch": 0.6248767882356075,
-      "grad_norm": 0.7560902833938599,
-      "learning_rate": 3.751065643648764e-06,
-      "loss": 0.1045,
-      "step": 3665
-    },
-    {
-      "epoch": 0.6250472866771453,
-      "grad_norm": 1.0680619478225708,
-      "learning_rate": 3.749360613810742e-06,
-      "loss": 0.1004,
-      "step": 3666
-    },
-    {
-      "epoch": 0.6252177851186829,
-      "grad_norm": 1.1420097351074219,
-      "learning_rate": 3.7476555839727196e-06,
-      "loss": 0.1339,
-      "step": 3667
-    },
-    {
-      "epoch": 0.6253882835602206,
-      "grad_norm": 1.1699634790420532,
-      "learning_rate": 3.7459505541346975e-06,
-      "loss": 0.1435,
-      "step": 3668
-    },
-    {
-      "epoch": 0.6255587820017583,
-      "grad_norm": 1.75743567943573,
-      "learning_rate": 3.744245524296676e-06,
-      "loss": 0.1742,
-      "step": 3669
-    },
-    {
-      "epoch": 0.6257292804432959,
-      "grad_norm": 0.8941011428833008,
-      "learning_rate": 3.7425404944586534e-06,
-      "loss": 0.0935,
-      "step": 3670
-    },
-    {
-      "epoch": 0.6258997788848336,
-      "grad_norm": 1.1393169164657593,
-      "learning_rate": 3.7408354646206314e-06,
-      "loss": 0.1109,
-      "step": 3671
-    },
-    {
-      "epoch": 0.6260702773263713,
-      "grad_norm": 0.9778454303741455,
-      "learning_rate": 3.739130434782609e-06,
-      "loss": 0.0932,
-      "step": 3672
-    },
-    {
-      "epoch": 0.626240775767909,
-      "grad_norm": 0.9243488311767578,
-      "learning_rate": 3.737425404944587e-06,
-      "loss": 0.1189,
-      "step": 3673
-    },
-    {
-      "epoch": 0.6264112742094466,
-      "grad_norm": 1.424014687538147,
-      "learning_rate": 3.7357203751065648e-06,
-      "loss": 0.1866,
-      "step": 3674
-    },
-    {
-      "epoch": 0.6265817726509844,
-      "grad_norm": 1.0741225481033325,
-      "learning_rate": 3.7340153452685423e-06,
-      "loss": 0.1011,
-      "step": 3675
-    },
-    {
-      "epoch": 0.6267522710925221,
-      "grad_norm": 0.9581297636032104,
-      "learning_rate": 3.7323103154305202e-06,
-      "loss": 0.0813,
-      "step": 3676
-    },
-    {
-      "epoch": 0.6269227695340597,
-      "grad_norm": 1.1146941184997559,
-      "learning_rate": 3.7306052855924978e-06,
-      "loss": 0.0959,
-      "step": 3677
-    },
-    {
-      "epoch": 0.6270932679755974,
-      "grad_norm": 1.449524998664856,
-      "learning_rate": 3.728900255754476e-06,
-      "loss": 0.1614,
-      "step": 3678
-    },
-    {
-      "epoch": 0.6272637664171351,
-      "grad_norm": 1.197462558746338,
-      "learning_rate": 3.727195225916454e-06,
-      "loss": 0.1207,
-      "step": 3679
-    },
-    {
-      "epoch": 0.6274342648586728,
-      "grad_norm": 1.4450664520263672,
-      "learning_rate": 3.7254901960784316e-06,
-      "loss": 0.1771,
-      "step": 3680
-    },
-    {
-      "epoch": 0.6276047633002104,
-      "grad_norm": 0.942268967628479,
-      "learning_rate": 3.7237851662404096e-06,
-      "loss": 0.1028,
-      "step": 3681
-    },
-    {
-      "epoch": 0.6277752617417481,
-      "grad_norm": 0.8110191226005554,
-      "learning_rate": 3.7220801364023875e-06,
-      "loss": 0.0894,
-      "step": 3682
-    },
-    {
-      "epoch": 0.6279457601832859,
-      "grad_norm": 0.8219690322875977,
-      "learning_rate": 3.720375106564365e-06,
-      "loss": 0.0416,
-      "step": 3683
-    },
-    {
-      "epoch": 0.6281162586248235,
-      "grad_norm": 1.3544223308563232,
-      "learning_rate": 3.718670076726343e-06,
-      "loss": 0.1467,
-      "step": 3684
-    },
-    {
-      "epoch": 0.6282867570663612,
-      "grad_norm": 0.9889111518859863,
-      "learning_rate": 3.7169650468883205e-06,
-      "loss": 0.0893,
-      "step": 3685
-    },
-    {
-      "epoch": 0.6284572555078989,
-      "grad_norm": 1.080906629562378,
-      "learning_rate": 3.715260017050299e-06,
-      "loss": 0.1105,
-      "step": 3686
-    },
-    {
-      "epoch": 0.6286277539494366,
-      "grad_norm": 1.4223904609680176,
-      "learning_rate": 3.713554987212277e-06,
-      "loss": 0.1193,
-      "step": 3687
-    },
-    {
-      "epoch": 0.6287982523909742,
-      "grad_norm": 1.2540326118469238,
-      "learning_rate": 3.7118499573742543e-06,
-      "loss": 0.1006,
-      "step": 3688
-    },
-    {
-      "epoch": 0.6289687508325119,
-      "grad_norm": 1.8358228206634521,
-      "learning_rate": 3.7101449275362323e-06,
-      "loss": 0.2074,
-      "step": 3689
-    },
-    {
-      "epoch": 0.6291392492740496,
-      "grad_norm": 0.9577386975288391,
-      "learning_rate": 3.70843989769821e-06,
-      "loss": 0.1069,
-      "step": 3690
-    },
-    {
-      "epoch": 0.6293097477155873,
-      "grad_norm": 1.1612111330032349,
-      "learning_rate": 3.7067348678601878e-06,
-      "loss": 0.1156,
-      "step": 3691
-    },
-    {
-      "epoch": 0.629480246157125,
-      "grad_norm": 0.8670212626457214,
-      "learning_rate": 3.7050298380221657e-06,
-      "loss": 0.0882,
-      "step": 3692
-    },
-    {
-      "epoch": 0.6296507445986627,
-      "grad_norm": 1.4416753053665161,
-      "learning_rate": 3.7033248081841432e-06,
-      "loss": 0.2159,
-      "step": 3693
-    },
-    {
-      "epoch": 0.6298212430402004,
-      "grad_norm": 1.1963956356048584,
-      "learning_rate": 3.7016197783461216e-06,
-      "loss": 0.1619,
-      "step": 3694
-    },
-    {
-      "epoch": 0.629991741481738,
-      "grad_norm": 1.2075142860412598,
-      "learning_rate": 3.6999147485080996e-06,
-      "loss": 0.144,
-      "step": 3695
-    },
-    {
-      "epoch": 0.6301622399232757,
-      "grad_norm": 1.5625414848327637,
-      "learning_rate": 3.698209718670077e-06,
-      "loss": 0.1359,
-      "step": 3696
-    },
-    {
-      "epoch": 0.6303327383648134,
-      "grad_norm": 0.7365015149116516,
-      "learning_rate": 3.696504688832055e-06,
-      "loss": 0.0789,
-      "step": 3697
-    },
-    {
-      "epoch": 0.6305032368063511,
-      "grad_norm": 1.8156697750091553,
-      "learning_rate": 3.6947996589940326e-06,
-      "loss": 0.1729,
-      "step": 3698
-    },
-    {
-      "epoch": 0.6306737352478887,
-      "grad_norm": 1.6419410705566406,
-      "learning_rate": 3.6930946291560105e-06,
-      "loss": 0.1807,
-      "step": 3699
-    },
-    {
-      "epoch": 0.6308442336894264,
-      "grad_norm": 1.471413016319275,
-      "learning_rate": 3.6913895993179884e-06,
-      "loss": 0.1174,
-      "step": 3700
-    },
-    {
-      "epoch": 0.6308442336894264,
-      "eval_f1_score": 0.39243498817966904,
-      "eval_loss": 0.13921166956424713,
-      "eval_runtime": 182.6385,
-      "eval_samples_per_second": 54.753,
-      "eval_steps_per_second": 3.422,
-      "step": 3700
-    },
-    {
-      "epoch": 0.6310147321309642,
-      "grad_norm": 1.068755030632019,
-      "learning_rate": 3.689684569479966e-06,
-      "loss": 0.1144,
-      "step": 3701
-    },
-    {
-      "epoch": 0.6311852305725018,
-      "grad_norm": 1.3145835399627686,
-      "learning_rate": 3.687979539641944e-06,
-      "loss": 0.1531,
-      "step": 3702
-    },
-    {
-      "epoch": 0.6313557290140395,
-      "grad_norm": 0.9875397086143494,
-      "learning_rate": 3.6862745098039223e-06,
-      "loss": 0.0833,
-      "step": 3703
-    },
-    {
-      "epoch": 0.6315262274555772,
-      "grad_norm": 1.2603150606155396,
-      "learning_rate": 3.6845694799659e-06,
-      "loss": 0.1177,
-      "step": 3704
-    },
-    {
-      "epoch": 0.6316967258971149,
-      "grad_norm": 1.634809970855713,
-      "learning_rate": 3.6828644501278778e-06,
-      "loss": 0.1531,
-      "step": 3705
-    },
-    {
-      "epoch": 0.6318672243386525,
-      "grad_norm": 1.2602183818817139,
-      "learning_rate": 3.6811594202898553e-06,
-      "loss": 0.087,
-      "step": 3706
-    },
-    {
-      "epoch": 0.6320377227801902,
-      "grad_norm": 1.5184547901153564,
-      "learning_rate": 3.6794543904518332e-06,
-      "loss": 0.1241,
-      "step": 3707
-    },
-    {
-      "epoch": 0.6322082212217279,
-      "grad_norm": 1.2422740459442139,
-      "learning_rate": 3.6777493606138108e-06,
-      "loss": 0.1015,
-      "step": 3708
-    },
-    {
-      "epoch": 0.6323787196632655,
-      "grad_norm": 1.2954447269439697,
-      "learning_rate": 3.6760443307757887e-06,
-      "loss": 0.1166,
-      "step": 3709
-    },
-    {
-      "epoch": 0.6325492181048032,
-      "grad_norm": 1.390432596206665,
-      "learning_rate": 3.6743393009377667e-06,
-      "loss": 0.0809,
-      "step": 3710
-    },
-    {
-      "epoch": 0.632719716546341,
-      "grad_norm": 1.4333117008209229,
-      "learning_rate": 3.6726342710997446e-06,
-      "loss": 0.1123,
-      "step": 3711
-    },
-    {
-      "epoch": 0.6328902149878787,
-      "grad_norm": 1.1851670742034912,
-      "learning_rate": 3.6709292412617225e-06,
-      "loss": 0.1495,
-      "step": 3712
-    },
-    {
-      "epoch": 0.6330607134294163,
-      "grad_norm": 1.3659263849258423,
-      "learning_rate": 3.6692242114237005e-06,
-      "loss": 0.1568,
-      "step": 3713
-    },
-    {
-      "epoch": 0.633231211870954,
-      "grad_norm": 0.888971745967865,
-      "learning_rate": 3.667519181585678e-06,
-      "loss": 0.1037,
-      "step": 3714
-    },
-    {
-      "epoch": 0.6334017103124917,
-      "grad_norm": 0.7728138566017151,
-      "learning_rate": 3.665814151747656e-06,
-      "loss": 0.0762,
-      "step": 3715
-    },
-    {
-      "epoch": 0.6335722087540293,
-      "grad_norm": 0.8941513895988464,
-      "learning_rate": 3.6641091219096335e-06,
-      "loss": 0.0907,
-      "step": 3716
-    },
-    {
-      "epoch": 0.633742707195567,
-      "grad_norm": 1.4324272871017456,
-      "learning_rate": 3.6624040920716114e-06,
-      "loss": 0.1573,
-      "step": 3717
-    },
-    {
-      "epoch": 0.6339132056371047,
-      "grad_norm": 1.6339035034179688,
-      "learning_rate": 3.660699062233589e-06,
-      "loss": 0.1796,
-      "step": 3718
-    },
-    {
-      "epoch": 0.6340837040786425,
-      "grad_norm": 1.3584506511688232,
-      "learning_rate": 3.6589940323955673e-06,
-      "loss": 0.0951,
-      "step": 3719
-    },
-    {
-      "epoch": 0.63425420252018,
-      "grad_norm": 1.2877280712127686,
-      "learning_rate": 3.6572890025575453e-06,
-      "loss": 0.1783,
-      "step": 3720
-    },
-    {
-      "epoch": 0.6344247009617178,
-      "grad_norm": 1.242985725402832,
-      "learning_rate": 3.6555839727195232e-06,
-      "loss": 0.1605,
-      "step": 3721
-    },
-    {
-      "epoch": 0.6345951994032555,
-      "grad_norm": 0.9977898597717285,
-      "learning_rate": 3.6538789428815008e-06,
-      "loss": 0.0877,
-      "step": 3722
-    },
-    {
-      "epoch": 0.6347656978447931,
-      "grad_norm": 1.6359201669692993,
-      "learning_rate": 3.6521739130434787e-06,
-      "loss": 0.1524,
-      "step": 3723
-    },
-    {
-      "epoch": 0.6349361962863308,
-      "grad_norm": 1.876238465309143,
-      "learning_rate": 3.6504688832054562e-06,
-      "loss": 0.23,
-      "step": 3724
-    },
-    {
-      "epoch": 0.6351066947278685,
-      "grad_norm": 0.8664019107818604,
-      "learning_rate": 3.648763853367434e-06,
-      "loss": 0.0898,
-      "step": 3725
-    },
-    {
-      "epoch": 0.6352771931694062,
-      "grad_norm": 1.3366501331329346,
-      "learning_rate": 3.6470588235294117e-06,
-      "loss": 0.1431,
-      "step": 3726
-    },
-    {
-      "epoch": 0.6354476916109438,
-      "grad_norm": 1.0310951471328735,
-      "learning_rate": 3.6453537936913896e-06,
-      "loss": 0.1006,
-      "step": 3727
-    },
-    {
-      "epoch": 0.6356181900524815,
-      "grad_norm": 1.419967770576477,
-      "learning_rate": 3.643648763853368e-06,
-      "loss": 0.1024,
-      "step": 3728
-    },
-    {
-      "epoch": 0.6357886884940193,
-      "grad_norm": 1.6248990297317505,
-      "learning_rate": 3.6419437340153455e-06,
-      "loss": 0.1434,
-      "step": 3729
-    },
-    {
-      "epoch": 0.6359591869355569,
-      "grad_norm": 0.9425191879272461,
-      "learning_rate": 3.6402387041773235e-06,
-      "loss": 0.1084,
-      "step": 3730
-    },
-    {
-      "epoch": 0.6361296853770946,
-      "grad_norm": 0.997732400894165,
-      "learning_rate": 3.6385336743393014e-06,
-      "loss": 0.1066,
-      "step": 3731
-    },
-    {
-      "epoch": 0.6363001838186323,
-      "grad_norm": 1.0537697076797485,
-      "learning_rate": 3.636828644501279e-06,
-      "loss": 0.095,
-      "step": 3732
-    },
-    {
-      "epoch": 0.63647068226017,
-      "grad_norm": 1.1263371706008911,
-      "learning_rate": 3.635123614663257e-06,
-      "loss": 0.1279,
-      "step": 3733
-    },
-    {
-      "epoch": 0.6366411807017076,
-      "grad_norm": 1.06753408908844,
-      "learning_rate": 3.6334185848252344e-06,
-      "loss": 0.1181,
-      "step": 3734
-    },
-    {
-      "epoch": 0.6368116791432453,
-      "grad_norm": 1.4207340478897095,
-      "learning_rate": 3.6317135549872124e-06,
-      "loss": 0.1478,
-      "step": 3735
-    },
-    {
-      "epoch": 0.636982177584783,
-      "grad_norm": 0.9397812485694885,
-      "learning_rate": 3.6300085251491908e-06,
-      "loss": 0.1224,
-      "step": 3736
-    },
-    {
-      "epoch": 0.6371526760263208,
-      "grad_norm": 1.3355766534805298,
-      "learning_rate": 3.6283034953111683e-06,
-      "loss": 0.1126,
-      "step": 3737
-    },
-    {
-      "epoch": 0.6373231744678584,
-      "grad_norm": 1.1690781116485596,
-      "learning_rate": 3.6265984654731462e-06,
-      "loss": 0.083,
-      "step": 3738
-    },
-    {
-      "epoch": 0.6374936729093961,
-      "grad_norm": 1.7492491006851196,
-      "learning_rate": 3.624893435635124e-06,
-      "loss": 0.0986,
-      "step": 3739
-    },
-    {
-      "epoch": 0.6376641713509338,
-      "grad_norm": 1.1186761856079102,
-      "learning_rate": 3.6231884057971017e-06,
-      "loss": 0.0829,
-      "step": 3740
-    },
-    {
-      "epoch": 0.6378346697924714,
-      "grad_norm": 1.0030421018600464,
-      "learning_rate": 3.6214833759590796e-06,
-      "loss": 0.1317,
-      "step": 3741
-    },
-    {
-      "epoch": 0.6380051682340091,
-      "grad_norm": 1.014840006828308,
-      "learning_rate": 3.619778346121057e-06,
-      "loss": 0.0665,
-      "step": 3742
-    },
-    {
-      "epoch": 0.6381756666755468,
-      "grad_norm": 0.9337121844291687,
-      "learning_rate": 3.618073316283035e-06,
-      "loss": 0.0969,
-      "step": 3743
-    },
-    {
-      "epoch": 0.6383461651170845,
-      "grad_norm": 1.4615178108215332,
-      "learning_rate": 3.6163682864450126e-06,
-      "loss": 0.1539,
-      "step": 3744
-    },
-    {
-      "epoch": 0.6385166635586221,
-      "grad_norm": 1.2855770587921143,
-      "learning_rate": 3.614663256606991e-06,
-      "loss": 0.0709,
-      "step": 3745
-    },
-    {
-      "epoch": 0.6386871620001598,
-      "grad_norm": 0.9363089203834534,
-      "learning_rate": 3.612958226768969e-06,
-      "loss": 0.0952,
-      "step": 3746
-    },
-    {
-      "epoch": 0.6388576604416976,
-      "grad_norm": 1.3684135675430298,
-      "learning_rate": 3.6112531969309465e-06,
-      "loss": 0.1522,
-      "step": 3747
-    },
-    {
-      "epoch": 0.6390281588832352,
-      "grad_norm": 1.185192346572876,
-      "learning_rate": 3.6095481670929244e-06,
-      "loss": 0.1127,
-      "step": 3748
-    },
-    {
-      "epoch": 0.6391986573247729,
-      "grad_norm": 0.8936169147491455,
-      "learning_rate": 3.6078431372549024e-06,
-      "loss": 0.0814,
-      "step": 3749
-    },
-    {
-      "epoch": 0.6393691557663106,
-      "grad_norm": 1.8236827850341797,
-      "learning_rate": 3.60613810741688e-06,
-      "loss": 0.216,
-      "step": 3750
-    },
-    {
-      "epoch": 0.6395396542078483,
-      "grad_norm": 0.8242852687835693,
-      "learning_rate": 3.604433077578858e-06,
-      "loss": 0.0427,
-      "step": 3751
-    },
-    {
-      "epoch": 0.6397101526493859,
-      "grad_norm": 0.960161566734314,
-      "learning_rate": 3.6027280477408354e-06,
-      "loss": 0.1096,
-      "step": 3752
-    },
-    {
-      "epoch": 0.6398806510909236,
-      "grad_norm": 1.8381279706954956,
-      "learning_rate": 3.6010230179028137e-06,
-      "loss": 0.1023,
-      "step": 3753
-    },
-    {
-      "epoch": 0.6400511495324613,
-      "grad_norm": 1.9192173480987549,
-      "learning_rate": 3.5993179880647917e-06,
-      "loss": 0.0642,
-      "step": 3754
-    },
-    {
-      "epoch": 0.6402216479739989,
-      "grad_norm": 1.2048285007476807,
-      "learning_rate": 3.5976129582267692e-06,
-      "loss": 0.1275,
-      "step": 3755
-    },
-    {
-      "epoch": 0.6403921464155367,
-      "grad_norm": 0.9745681285858154,
-      "learning_rate": 3.595907928388747e-06,
-      "loss": 0.0782,
-      "step": 3756
-    },
-    {
-      "epoch": 0.6405626448570744,
-      "grad_norm": 1.9655649662017822,
-      "learning_rate": 3.594202898550725e-06,
-      "loss": 0.1831,
-      "step": 3757
-    },
-    {
-      "epoch": 0.6407331432986121,
-      "grad_norm": 2.731715440750122,
-      "learning_rate": 3.5924978687127026e-06,
-      "loss": 0.2583,
-      "step": 3758
-    },
-    {
-      "epoch": 0.6409036417401497,
-      "grad_norm": 1.749211072921753,
-      "learning_rate": 3.5907928388746806e-06,
-      "loss": 0.1881,
-      "step": 3759
-    },
-    {
-      "epoch": 0.6410741401816874,
-      "grad_norm": 2.5945215225219727,
-      "learning_rate": 3.589087809036658e-06,
-      "loss": 0.2383,
-      "step": 3760
-    },
-    {
-      "epoch": 0.6412446386232251,
-      "grad_norm": 1.4661320447921753,
-      "learning_rate": 3.5873827791986365e-06,
-      "loss": 0.1612,
-      "step": 3761
-    },
-    {
-      "epoch": 0.6414151370647627,
-      "grad_norm": 1.5145949125289917,
-      "learning_rate": 3.5856777493606144e-06,
-      "loss": 0.1546,
-      "step": 3762
-    },
-    {
-      "epoch": 0.6415856355063004,
-      "grad_norm": 1.1292080879211426,
-      "learning_rate": 3.583972719522592e-06,
-      "loss": 0.1296,
-      "step": 3763
-    },
-    {
-      "epoch": 0.6417561339478381,
-      "grad_norm": 1.090998888015747,
-      "learning_rate": 3.58226768968457e-06,
-      "loss": 0.1035,
-      "step": 3764
-    },
-    {
-      "epoch": 0.6419266323893759,
-      "grad_norm": 2.3366146087646484,
-      "learning_rate": 3.5805626598465474e-06,
-      "loss": 0.2259,
-      "step": 3765
-    },
-    {
-      "epoch": 0.6420971308309135,
-      "grad_norm": 1.2206898927688599,
-      "learning_rate": 3.5788576300085254e-06,
-      "loss": 0.1429,
-      "step": 3766
-    },
-    {
-      "epoch": 0.6422676292724512,
-      "grad_norm": 1.031091332435608,
-      "learning_rate": 3.5771526001705033e-06,
-      "loss": 0.1154,
-      "step": 3767
-    },
-    {
-      "epoch": 0.6424381277139889,
-      "grad_norm": 1.4148261547088623,
-      "learning_rate": 3.575447570332481e-06,
-      "loss": 0.214,
-      "step": 3768
-    },
-    {
-      "epoch": 0.6426086261555265,
-      "grad_norm": 1.4270838499069214,
-      "learning_rate": 3.573742540494459e-06,
-      "loss": 0.113,
-      "step": 3769
-    },
-    {
-      "epoch": 0.6427791245970642,
-      "grad_norm": 1.338118314743042,
-      "learning_rate": 3.572037510656437e-06,
-      "loss": 0.0562,
-      "step": 3770
-    },
-    {
-      "epoch": 0.6429496230386019,
-      "grad_norm": 1.1331686973571777,
-      "learning_rate": 3.5703324808184147e-06,
-      "loss": 0.1756,
-      "step": 3771
-    },
-    {
-      "epoch": 0.6431201214801396,
-      "grad_norm": 1.1654812097549438,
-      "learning_rate": 3.5686274509803926e-06,
-      "loss": 0.147,
-      "step": 3772
-    },
-    {
-      "epoch": 0.6432906199216772,
-      "grad_norm": 1.7656182050704956,
-      "learning_rate": 3.56692242114237e-06,
-      "loss": 0.2249,
-      "step": 3773
-    },
-    {
-      "epoch": 0.643461118363215,
-      "grad_norm": 1.0476933717727661,
-      "learning_rate": 3.565217391304348e-06,
-      "loss": 0.1162,
-      "step": 3774
-    },
-    {
-      "epoch": 0.6436316168047527,
-      "grad_norm": 1.039013385772705,
-      "learning_rate": 3.5635123614663256e-06,
-      "loss": 0.1064,
-      "step": 3775
-    },
-    {
-      "epoch": 0.6438021152462903,
-      "grad_norm": 0.9786112904548645,
-      "learning_rate": 3.5618073316283036e-06,
-      "loss": 0.0754,
-      "step": 3776
-    },
-    {
-      "epoch": 0.643972613687828,
-      "grad_norm": 1.4536173343658447,
-      "learning_rate": 3.5601023017902815e-06,
-      "loss": 0.1466,
-      "step": 3777
-    },
-    {
-      "epoch": 0.6441431121293657,
-      "grad_norm": 0.7579110860824585,
-      "learning_rate": 3.55839727195226e-06,
-      "loss": 0.0687,
-      "step": 3778
-    },
-    {
-      "epoch": 0.6443136105709034,
-      "grad_norm": 0.8089442849159241,
-      "learning_rate": 3.5566922421142374e-06,
-      "loss": 0.106,
-      "step": 3779
-    },
-    {
-      "epoch": 0.644484109012441,
-      "grad_norm": 1.1764006614685059,
-      "learning_rate": 3.5549872122762154e-06,
-      "loss": 0.1383,
-      "step": 3780
-    },
-    {
-      "epoch": 0.6446546074539787,
-      "grad_norm": 1.211492657661438,
-      "learning_rate": 3.553282182438193e-06,
-      "loss": 0.145,
-      "step": 3781
-    },
-    {
-      "epoch": 0.6448251058955164,
-      "grad_norm": 1.0914140939712524,
-      "learning_rate": 3.551577152600171e-06,
-      "loss": 0.1325,
-      "step": 3782
-    },
-    {
-      "epoch": 0.644995604337054,
-      "grad_norm": 1.4939686059951782,
-      "learning_rate": 3.5498721227621484e-06,
-      "loss": 0.1502,
-      "step": 3783
-    },
-    {
-      "epoch": 0.6451661027785918,
-      "grad_norm": 0.7499918341636658,
-      "learning_rate": 3.5481670929241263e-06,
-      "loss": 0.0849,
-      "step": 3784
-    },
-    {
-      "epoch": 0.6453366012201295,
-      "grad_norm": 0.8260846734046936,
-      "learning_rate": 3.5464620630861043e-06,
-      "loss": 0.0753,
-      "step": 3785
-    },
-    {
-      "epoch": 0.6455070996616672,
-      "grad_norm": 1.132487177848816,
-      "learning_rate": 3.544757033248082e-06,
-      "loss": 0.141,
-      "step": 3786
-    },
-    {
-      "epoch": 0.6456775981032048,
-      "grad_norm": 0.7627885937690735,
-      "learning_rate": 3.54305200341006e-06,
-      "loss": 0.0792,
-      "step": 3787
-    },
-    {
-      "epoch": 0.6458480965447425,
-      "grad_norm": 1.0610328912734985,
-      "learning_rate": 3.541346973572038e-06,
-      "loss": 0.0832,
-      "step": 3788
-    },
-    {
-      "epoch": 0.6460185949862802,
-      "grad_norm": 1.3593807220458984,
-      "learning_rate": 3.5396419437340156e-06,
-      "loss": 0.1432,
-      "step": 3789
-    },
-    {
-      "epoch": 0.6461890934278179,
-      "grad_norm": 1.1016868352890015,
-      "learning_rate": 3.5379369138959936e-06,
-      "loss": 0.1177,
-      "step": 3790
-    },
-    {
-      "epoch": 0.6463595918693555,
-      "grad_norm": 0.8754037022590637,
-      "learning_rate": 3.536231884057971e-06,
-      "loss": 0.0605,
-      "step": 3791
-    },
-    {
-      "epoch": 0.6465300903108933,
-      "grad_norm": 1.6358007192611694,
-      "learning_rate": 3.534526854219949e-06,
-      "loss": 0.1972,
-      "step": 3792
-    },
-    {
-      "epoch": 0.646700588752431,
-      "grad_norm": 1.2707304954528809,
-      "learning_rate": 3.5328218243819266e-06,
-      "loss": 0.1678,
-      "step": 3793
-    },
-    {
-      "epoch": 0.6468710871939686,
-      "grad_norm": 1.263748288154602,
-      "learning_rate": 3.5311167945439045e-06,
-      "loss": 0.1438,
-      "step": 3794
-    },
-    {
-      "epoch": 0.6470415856355063,
-      "grad_norm": 1.232202172279358,
-      "learning_rate": 3.529411764705883e-06,
-      "loss": 0.1368,
-      "step": 3795
-    },
-    {
-      "epoch": 0.647212084077044,
-      "grad_norm": 2.188258171081543,
-      "learning_rate": 3.527706734867861e-06,
-      "loss": 0.1507,
-      "step": 3796
-    },
-    {
-      "epoch": 0.6473825825185817,
-      "grad_norm": 1.0352145433425903,
-      "learning_rate": 3.5260017050298384e-06,
-      "loss": 0.0969,
-      "step": 3797
-    },
-    {
-      "epoch": 0.6475530809601193,
-      "grad_norm": 1.2449350357055664,
-      "learning_rate": 3.5242966751918163e-06,
-      "loss": 0.1264,
-      "step": 3798
-    },
-    {
-      "epoch": 0.647723579401657,
-      "grad_norm": 0.8735178709030151,
-      "learning_rate": 3.522591645353794e-06,
-      "loss": 0.0992,
-      "step": 3799
-    },
-    {
-      "epoch": 0.6478940778431947,
-      "grad_norm": 1.3349305391311646,
-      "learning_rate": 3.5208866155157718e-06,
-      "loss": 0.0783,
-      "step": 3800
-    },
-    {
-      "epoch": 0.6478940778431947,
-      "eval_f1_score": 0.36507936507936506,
-      "eval_loss": 0.1384362429380417,
-      "eval_runtime": 182.6236,
-      "eval_samples_per_second": 54.757,
-      "eval_steps_per_second": 3.422,
-      "step": 3800
-    },
-    {
-      "epoch": 0.6480645762847324,
-      "grad_norm": 1.3321579694747925,
-      "learning_rate": 3.5191815856777493e-06,
-      "loss": 0.1426,
-      "step": 3801
-    },
-    {
-      "epoch": 0.6482350747262701,
-      "grad_norm": 2.1399898529052734,
-      "learning_rate": 3.5174765558397273e-06,
-      "loss": 0.1892,
-      "step": 3802
-    },
-    {
-      "epoch": 0.6484055731678078,
-      "grad_norm": 1.6246662139892578,
-      "learning_rate": 3.5157715260017056e-06,
-      "loss": 0.1607,
-      "step": 3803
-    },
-    {
-      "epoch": 0.6485760716093455,
-      "grad_norm": 1.098414421081543,
-      "learning_rate": 3.514066496163683e-06,
-      "loss": 0.1288,
-      "step": 3804
-    },
-    {
-      "epoch": 0.6487465700508831,
-      "grad_norm": 1.7618242502212524,
-      "learning_rate": 3.512361466325661e-06,
-      "loss": 0.1852,
-      "step": 3805
-    },
-    {
-      "epoch": 0.6489170684924208,
-      "grad_norm": 2.6852781772613525,
-      "learning_rate": 3.510656436487639e-06,
-      "loss": 0.2276,
-      "step": 3806
-    },
-    {
-      "epoch": 0.6490875669339585,
-      "grad_norm": 0.8381413221359253,
-      "learning_rate": 3.5089514066496166e-06,
-      "loss": 0.0753,
-      "step": 3807
-    },
-    {
-      "epoch": 0.6492580653754961,
-      "grad_norm": 1.3367550373077393,
-      "learning_rate": 3.5072463768115945e-06,
-      "loss": 0.1042,
-      "step": 3808
-    },
-    {
-      "epoch": 0.6494285638170338,
-      "grad_norm": 1.3574849367141724,
-      "learning_rate": 3.505541346973572e-06,
-      "loss": 0.1217,
-      "step": 3809
-    },
-    {
-      "epoch": 0.6495990622585716,
-      "grad_norm": 1.8206461668014526,
-      "learning_rate": 3.50383631713555e-06,
-      "loss": 0.2149,
-      "step": 3810
-    },
-    {
-      "epoch": 0.6497695607001093,
-      "grad_norm": 1.9767436981201172,
-      "learning_rate": 3.5021312872975284e-06,
-      "loss": 0.1617,
-      "step": 3811
-    },
-    {
-      "epoch": 0.6499400591416469,
-      "grad_norm": 1.1481764316558838,
-      "learning_rate": 3.500426257459506e-06,
-      "loss": 0.126,
-      "step": 3812
-    },
-    {
-      "epoch": 0.6501105575831846,
-      "grad_norm": 1.4578038454055786,
-      "learning_rate": 3.498721227621484e-06,
-      "loss": 0.17,
-      "step": 3813
-    },
-    {
-      "epoch": 0.6502810560247223,
-      "grad_norm": 1.791718602180481,
-      "learning_rate": 3.4970161977834618e-06,
-      "loss": 0.1549,
-      "step": 3814
-    },
-    {
-      "epoch": 0.6504515544662599,
-      "grad_norm": 1.4585214853286743,
-      "learning_rate": 3.4953111679454393e-06,
-      "loss": 0.1895,
-      "step": 3815
-    },
-    {
-      "epoch": 0.6506220529077976,
-      "grad_norm": 1.3441143035888672,
-      "learning_rate": 3.4936061381074173e-06,
-      "loss": 0.1039,
-      "step": 3816
-    },
-    {
-      "epoch": 0.6507925513493353,
-      "grad_norm": 1.3681156635284424,
-      "learning_rate": 3.4919011082693948e-06,
-      "loss": 0.1486,
-      "step": 3817
-    },
-    {
-      "epoch": 0.650963049790873,
-      "grad_norm": 1.5697520971298218,
-      "learning_rate": 3.4901960784313727e-06,
-      "loss": 0.1344,
-      "step": 3818
-    },
-    {
-      "epoch": 0.6511335482324107,
-      "grad_norm": 1.6502066850662231,
-      "learning_rate": 3.4884910485933502e-06,
-      "loss": 0.2005,
-      "step": 3819
-    },
-    {
-      "epoch": 0.6513040466739484,
-      "grad_norm": 1.2139993906021118,
-      "learning_rate": 3.4867860187553286e-06,
-      "loss": 0.1447,
-      "step": 3820
-    },
-    {
-      "epoch": 0.6514745451154861,
-      "grad_norm": 1.3323631286621094,
-      "learning_rate": 3.4850809889173066e-06,
-      "loss": 0.099,
-      "step": 3821
-    },
-    {
-      "epoch": 0.6516450435570237,
-      "grad_norm": 2.062469720840454,
-      "learning_rate": 3.483375959079284e-06,
-      "loss": 0.2513,
-      "step": 3822
-    },
-    {
-      "epoch": 0.6518155419985614,
-      "grad_norm": 1.5062077045440674,
-      "learning_rate": 3.481670929241262e-06,
-      "loss": 0.1807,
-      "step": 3823
-    },
-    {
-      "epoch": 0.6519860404400991,
-      "grad_norm": 1.6660444736480713,
-      "learning_rate": 3.47996589940324e-06,
-      "loss": 0.1442,
-      "step": 3824
-    },
-    {
-      "epoch": 0.6521565388816368,
-      "grad_norm": 1.4998832941055298,
-      "learning_rate": 3.4782608695652175e-06,
-      "loss": 0.1212,
-      "step": 3825
-    },
-    {
-      "epoch": 0.6523270373231744,
-      "grad_norm": 0.9821255207061768,
-      "learning_rate": 3.4765558397271955e-06,
-      "loss": 0.1042,
-      "step": 3826
-    },
-    {
-      "epoch": 0.6524975357647121,
-      "grad_norm": 1.3991197347640991,
-      "learning_rate": 3.474850809889173e-06,
-      "loss": 0.1926,
-      "step": 3827
-    },
-    {
-      "epoch": 0.6526680342062499,
-      "grad_norm": 1.1085835695266724,
-      "learning_rate": 3.4731457800511514e-06,
-      "loss": 0.1106,
-      "step": 3828
-    },
-    {
-      "epoch": 0.6528385326477875,
-      "grad_norm": 0.8781197667121887,
-      "learning_rate": 3.4714407502131293e-06,
-      "loss": 0.0754,
-      "step": 3829
-    },
-    {
-      "epoch": 0.6530090310893252,
-      "grad_norm": 1.2629003524780273,
-      "learning_rate": 3.469735720375107e-06,
-      "loss": 0.0979,
-      "step": 3830
-    },
-    {
-      "epoch": 0.6531795295308629,
-      "grad_norm": 1.2773722410202026,
-      "learning_rate": 3.4680306905370848e-06,
-      "loss": 0.16,
-      "step": 3831
-    },
-    {
-      "epoch": 0.6533500279724006,
-      "grad_norm": 1.459194540977478,
-      "learning_rate": 3.4663256606990627e-06,
-      "loss": 0.1644,
-      "step": 3832
-    },
-    {
-      "epoch": 0.6535205264139382,
-      "grad_norm": 1.4942381381988525,
-      "learning_rate": 3.4646206308610402e-06,
-      "loss": 0.1566,
-      "step": 3833
-    },
-    {
-      "epoch": 0.6536910248554759,
-      "grad_norm": 1.6898901462554932,
-      "learning_rate": 3.462915601023018e-06,
-      "loss": 0.1428,
-      "step": 3834
-    },
-    {
-      "epoch": 0.6538615232970136,
-      "grad_norm": 1.0706831216812134,
-      "learning_rate": 3.4612105711849957e-06,
-      "loss": 0.0861,
-      "step": 3835
-    },
-    {
-      "epoch": 0.6540320217385514,
-      "grad_norm": 1.789047360420227,
-      "learning_rate": 3.4595055413469737e-06,
-      "loss": 0.138,
-      "step": 3836
-    },
-    {
-      "epoch": 0.654202520180089,
-      "grad_norm": 1.168027400970459,
-      "learning_rate": 3.457800511508952e-06,
-      "loss": 0.1506,
-      "step": 3837
-    },
-    {
-      "epoch": 0.6543730186216267,
-      "grad_norm": 1.7767421007156372,
-      "learning_rate": 3.4560954816709296e-06,
-      "loss": 0.2074,
-      "step": 3838
-    },
-    {
-      "epoch": 0.6545435170631644,
-      "grad_norm": 0.7228270173072815,
-      "learning_rate": 3.4543904518329075e-06,
-      "loss": 0.086,
-      "step": 3839
-    },
-    {
-      "epoch": 0.654714015504702,
-      "grad_norm": 1.0253210067749023,
-      "learning_rate": 3.452685421994885e-06,
-      "loss": 0.0986,
-      "step": 3840
-    },
-    {
-      "epoch": 0.6548845139462397,
-      "grad_norm": 0.9064176082611084,
-      "learning_rate": 3.450980392156863e-06,
-      "loss": 0.063,
-      "step": 3841
-    },
-    {
-      "epoch": 0.6550550123877774,
-      "grad_norm": 0.9495208859443665,
-      "learning_rate": 3.449275362318841e-06,
-      "loss": 0.0893,
-      "step": 3842
-    },
-    {
-      "epoch": 0.6552255108293151,
-      "grad_norm": 1.4983458518981934,
-      "learning_rate": 3.4475703324808185e-06,
-      "loss": 0.1644,
-      "step": 3843
-    },
-    {
-      "epoch": 0.6553960092708527,
-      "grad_norm": 0.6246259808540344,
-      "learning_rate": 3.4458653026427964e-06,
-      "loss": 0.0532,
-      "step": 3844
-    },
-    {
-      "epoch": 0.6555665077123904,
-      "grad_norm": 1.510249376296997,
-      "learning_rate": 3.4441602728047748e-06,
-      "loss": 0.1764,
-      "step": 3845
-    },
-    {
-      "epoch": 0.6557370061539282,
-      "grad_norm": 0.9134432673454285,
-      "learning_rate": 3.4424552429667523e-06,
-      "loss": 0.0811,
-      "step": 3846
-    },
-    {
-      "epoch": 0.6559075045954658,
-      "grad_norm": 1.1383296251296997,
-      "learning_rate": 3.4407502131287302e-06,
-      "loss": 0.0533,
-      "step": 3847
-    },
-    {
-      "epoch": 0.6560780030370035,
-      "grad_norm": 1.3036679029464722,
-      "learning_rate": 3.4390451832907078e-06,
-      "loss": 0.1011,
-      "step": 3848
-    },
-    {
-      "epoch": 0.6562485014785412,
-      "grad_norm": 1.6370482444763184,
-      "learning_rate": 3.4373401534526857e-06,
-      "loss": 0.1688,
-      "step": 3849
-    },
-    {
-      "epoch": 0.6564189999200789,
-      "grad_norm": 1.7106876373291016,
-      "learning_rate": 3.4356351236146632e-06,
-      "loss": 0.1582,
-      "step": 3850
-    },
-    {
-      "epoch": 0.6565894983616165,
-      "grad_norm": 2.144707441329956,
-      "learning_rate": 3.433930093776641e-06,
-      "loss": 0.1703,
-      "step": 3851
-    },
-    {
-      "epoch": 0.6567599968031542,
-      "grad_norm": 1.745017647743225,
-      "learning_rate": 3.432225063938619e-06,
-      "loss": 0.1742,
-      "step": 3852
-    },
-    {
-      "epoch": 0.6569304952446919,
-      "grad_norm": 1.7782965898513794,
-      "learning_rate": 3.4305200341005975e-06,
-      "loss": 0.1941,
-      "step": 3853
-    },
-    {
-      "epoch": 0.6571009936862295,
-      "grad_norm": 1.200868844985962,
-      "learning_rate": 3.428815004262575e-06,
-      "loss": 0.0851,
-      "step": 3854
-    },
-    {
-      "epoch": 0.6572714921277673,
-      "grad_norm": 1.544946551322937,
-      "learning_rate": 3.427109974424553e-06,
-      "loss": 0.1409,
-      "step": 3855
-    },
-    {
-      "epoch": 0.657441990569305,
-      "grad_norm": 1.847928762435913,
-      "learning_rate": 3.4254049445865305e-06,
-      "loss": 0.2057,
-      "step": 3856
-    },
-    {
-      "epoch": 0.6576124890108427,
-      "grad_norm": 1.2779704332351685,
-      "learning_rate": 3.4236999147485084e-06,
-      "loss": 0.1512,
-      "step": 3857
-    },
-    {
-      "epoch": 0.6577829874523803,
-      "grad_norm": 1.6082793474197388,
-      "learning_rate": 3.421994884910486e-06,
-      "loss": 0.1457,
-      "step": 3858
-    },
-    {
-      "epoch": 0.657953485893918,
-      "grad_norm": 1.6311509609222412,
-      "learning_rate": 3.420289855072464e-06,
-      "loss": 0.1427,
-      "step": 3859
-    },
-    {
-      "epoch": 0.6581239843354557,
-      "grad_norm": 1.1573950052261353,
-      "learning_rate": 3.418584825234442e-06,
-      "loss": 0.0945,
-      "step": 3860
-    },
-    {
-      "epoch": 0.6582944827769933,
-      "grad_norm": 0.9491661190986633,
-      "learning_rate": 3.4168797953964194e-06,
-      "loss": 0.0815,
-      "step": 3861
-    },
-    {
-      "epoch": 0.658464981218531,
-      "grad_norm": 1.9215099811553955,
-      "learning_rate": 3.4151747655583978e-06,
-      "loss": 0.2201,
-      "step": 3862
-    },
-    {
-      "epoch": 0.6586354796600687,
-      "grad_norm": 1.824528455734253,
-      "learning_rate": 3.4134697357203757e-06,
-      "loss": 0.1709,
-      "step": 3863
-    },
-    {
-      "epoch": 0.6588059781016065,
-      "grad_norm": 1.0630437135696411,
-      "learning_rate": 3.4117647058823532e-06,
-      "loss": 0.131,
-      "step": 3864
-    },
-    {
-      "epoch": 0.6589764765431441,
-      "grad_norm": 1.065984845161438,
-      "learning_rate": 3.410059676044331e-06,
-      "loss": 0.1075,
-      "step": 3865
-    },
-    {
-      "epoch": 0.6591469749846818,
-      "grad_norm": 1.3609118461608887,
-      "learning_rate": 3.4083546462063087e-06,
-      "loss": 0.0909,
-      "step": 3866
-    },
-    {
-      "epoch": 0.6593174734262195,
-      "grad_norm": 1.439451813697815,
-      "learning_rate": 3.4066496163682867e-06,
-      "loss": 0.1493,
-      "step": 3867
-    },
-    {
-      "epoch": 0.6594879718677571,
-      "grad_norm": 1.1943893432617188,
-      "learning_rate": 3.404944586530264e-06,
-      "loss": 0.0954,
-      "step": 3868
-    },
-    {
-      "epoch": 0.6596584703092948,
-      "grad_norm": 1.0462403297424316,
-      "learning_rate": 3.403239556692242e-06,
-      "loss": 0.1184,
-      "step": 3869
-    },
-    {
-      "epoch": 0.6598289687508325,
-      "grad_norm": 1.055091142654419,
-      "learning_rate": 3.4015345268542205e-06,
-      "loss": 0.1287,
-      "step": 3870
-    },
-    {
-      "epoch": 0.6599994671923702,
-      "grad_norm": 1.0880324840545654,
-      "learning_rate": 3.3998294970161984e-06,
-      "loss": 0.0777,
-      "step": 3871
-    },
-    {
-      "epoch": 0.6601699656339078,
-      "grad_norm": 2.1348769664764404,
-      "learning_rate": 3.398124467178176e-06,
-      "loss": 0.2215,
-      "step": 3872
-    },
-    {
-      "epoch": 0.6603404640754456,
-      "grad_norm": 0.8190430402755737,
-      "learning_rate": 3.396419437340154e-06,
-      "loss": 0.1036,
-      "step": 3873
-    },
-    {
-      "epoch": 0.6605109625169833,
-      "grad_norm": 1.4419548511505127,
-      "learning_rate": 3.3947144075021314e-06,
-      "loss": 0.108,
-      "step": 3874
-    },
-    {
-      "epoch": 0.6606814609585209,
-      "grad_norm": 1.437876582145691,
-      "learning_rate": 3.3930093776641094e-06,
-      "loss": 0.1422,
-      "step": 3875
-    },
-    {
-      "epoch": 0.6608519594000586,
-      "grad_norm": 1.0468285083770752,
-      "learning_rate": 3.391304347826087e-06,
-      "loss": 0.0848,
-      "step": 3876
-    },
-    {
-      "epoch": 0.6610224578415963,
-      "grad_norm": 0.9778571724891663,
-      "learning_rate": 3.389599317988065e-06,
-      "loss": 0.1,
-      "step": 3877
-    },
-    {
-      "epoch": 0.661192956283134,
-      "grad_norm": 1.0217633247375488,
-      "learning_rate": 3.3878942881500432e-06,
-      "loss": 0.1275,
-      "step": 3878
-    },
-    {
-      "epoch": 0.6613634547246716,
-      "grad_norm": 1.4020358324050903,
-      "learning_rate": 3.3861892583120208e-06,
-      "loss": 0.1543,
-      "step": 3879
-    },
-    {
-      "epoch": 0.6615339531662093,
-      "grad_norm": 1.001997709274292,
-      "learning_rate": 3.3844842284739987e-06,
-      "loss": 0.0907,
-      "step": 3880
-    },
-    {
-      "epoch": 0.661704451607747,
-      "grad_norm": 0.9944140315055847,
-      "learning_rate": 3.3827791986359767e-06,
-      "loss": 0.0432,
-      "step": 3881
-    },
-    {
-      "epoch": 0.6618749500492848,
-      "grad_norm": 0.9731889963150024,
-      "learning_rate": 3.381074168797954e-06,
-      "loss": 0.103,
-      "step": 3882
-    },
-    {
-      "epoch": 0.6620454484908224,
-      "grad_norm": 0.8612279891967773,
-      "learning_rate": 3.379369138959932e-06,
-      "loss": 0.0693,
-      "step": 3883
-    },
-    {
-      "epoch": 0.6622159469323601,
-      "grad_norm": 1.4289515018463135,
-      "learning_rate": 3.3776641091219096e-06,
-      "loss": 0.1183,
-      "step": 3884
-    },
-    {
-      "epoch": 0.6623864453738978,
-      "grad_norm": 1.729993224143982,
-      "learning_rate": 3.3759590792838876e-06,
-      "loss": 0.1182,
-      "step": 3885
-    },
-    {
-      "epoch": 0.6625569438154354,
-      "grad_norm": 0.764060378074646,
-      "learning_rate": 3.374254049445865e-06,
-      "loss": 0.0518,
-      "step": 3886
-    },
-    {
-      "epoch": 0.6627274422569731,
-      "grad_norm": 1.8344427347183228,
-      "learning_rate": 3.3725490196078435e-06,
-      "loss": 0.1916,
-      "step": 3887
-    },
-    {
-      "epoch": 0.6628979406985108,
-      "grad_norm": 1.5775954723358154,
-      "learning_rate": 3.3708439897698214e-06,
-      "loss": 0.1746,
-      "step": 3888
-    },
-    {
-      "epoch": 0.6630684391400485,
-      "grad_norm": 1.4308315515518188,
-      "learning_rate": 3.3691389599317994e-06,
-      "loss": 0.1282,
-      "step": 3889
-    },
-    {
-      "epoch": 0.6632389375815861,
-      "grad_norm": 0.7640116810798645,
-      "learning_rate": 3.367433930093777e-06,
-      "loss": 0.0536,
-      "step": 3890
-    },
-    {
-      "epoch": 0.6634094360231239,
-      "grad_norm": 0.9877718687057495,
-      "learning_rate": 3.365728900255755e-06,
-      "loss": 0.0749,
-      "step": 3891
-    },
-    {
-      "epoch": 0.6635799344646616,
-      "grad_norm": 0.7810732126235962,
-      "learning_rate": 3.3640238704177324e-06,
-      "loss": 0.0778,
-      "step": 3892
-    },
-    {
-      "epoch": 0.6637504329061992,
-      "grad_norm": 1.4315524101257324,
-      "learning_rate": 3.3623188405797103e-06,
-      "loss": 0.1389,
-      "step": 3893
-    },
-    {
-      "epoch": 0.6639209313477369,
-      "grad_norm": 0.85233074426651,
-      "learning_rate": 3.360613810741688e-06,
-      "loss": 0.0658,
-      "step": 3894
-    },
-    {
-      "epoch": 0.6640914297892746,
-      "grad_norm": 1.3665393590927124,
-      "learning_rate": 3.3589087809036662e-06,
-      "loss": 0.1203,
-      "step": 3895
-    },
-    {
-      "epoch": 0.6642619282308123,
-      "grad_norm": 1.551367998123169,
-      "learning_rate": 3.357203751065644e-06,
-      "loss": 0.1798,
-      "step": 3896
-    },
-    {
-      "epoch": 0.6644324266723499,
-      "grad_norm": 1.6179540157318115,
-      "learning_rate": 3.3554987212276217e-06,
-      "loss": 0.1041,
-      "step": 3897
-    },
-    {
-      "epoch": 0.6646029251138876,
-      "grad_norm": 1.2029205560684204,
-      "learning_rate": 3.3537936913895996e-06,
-      "loss": 0.1263,
-      "step": 3898
-    },
-    {
-      "epoch": 0.6647734235554253,
-      "grad_norm": 1.257601022720337,
-      "learning_rate": 3.3520886615515776e-06,
-      "loss": 0.146,
-      "step": 3899
-    },
-    {
-      "epoch": 0.664943921996963,
-      "grad_norm": 1.4511607885360718,
-      "learning_rate": 3.350383631713555e-06,
-      "loss": 0.244,
-      "step": 3900
-    },
-    {
-      "epoch": 0.664943921996963,
-      "eval_f1_score": 0.40294840294840295,
-      "eval_loss": 0.13890038430690765,
-      "eval_runtime": 182.6853,
-      "eval_samples_per_second": 54.739,
-      "eval_steps_per_second": 3.421,
-      "step": 3900
-    },
-    {
-      "epoch": 0.6651144204385007,
-      "grad_norm": 1.7217893600463867,
-      "learning_rate": 3.348678601875533e-06,
-      "loss": 0.2274,
-      "step": 3901
-    },
-    {
-      "epoch": 0.6652849188800384,
-      "grad_norm": 1.937384843826294,
-      "learning_rate": 3.3469735720375106e-06,
-      "loss": 0.2111,
-      "step": 3902
-    },
-    {
-      "epoch": 0.6654554173215761,
-      "grad_norm": 1.0646719932556152,
-      "learning_rate": 3.3452685421994885e-06,
-      "loss": 0.1148,
-      "step": 3903
-    },
-    {
-      "epoch": 0.6656259157631137,
-      "grad_norm": 1.8812726736068726,
-      "learning_rate": 3.343563512361467e-06,
-      "loss": 0.1844,
-      "step": 3904
-    },
-    {
-      "epoch": 0.6657964142046514,
-      "grad_norm": 1.9182530641555786,
-      "learning_rate": 3.3418584825234444e-06,
-      "loss": 0.1732,
-      "step": 3905
-    },
-    {
-      "epoch": 0.6659669126461891,
-      "grad_norm": 1.1144908666610718,
-      "learning_rate": 3.3401534526854224e-06,
-      "loss": 0.106,
-      "step": 3906
-    },
-    {
-      "epoch": 0.6661374110877267,
-      "grad_norm": 1.099799633026123,
-      "learning_rate": 3.3384484228474003e-06,
-      "loss": 0.1331,
-      "step": 3907
-    },
-    {
-      "epoch": 0.6663079095292644,
-      "grad_norm": 1.1822335720062256,
-      "learning_rate": 3.336743393009378e-06,
-      "loss": 0.1259,
-      "step": 3908
-    },
-    {
-      "epoch": 0.6664784079708022,
-      "grad_norm": 1.166798710823059,
-      "learning_rate": 3.335038363171356e-06,
-      "loss": 0.101,
-      "step": 3909
-    },
-    {
-      "epoch": 0.6666489064123399,
-      "grad_norm": 1.1110976934432983,
-      "learning_rate": 3.3333333333333333e-06,
-      "loss": 0.0847,
-      "step": 3910
-    },
-    {
-      "epoch": 0.6668194048538775,
-      "grad_norm": 1.2571207284927368,
-      "learning_rate": 3.3316283034953113e-06,
-      "loss": 0.1258,
-      "step": 3911
-    },
-    {
-      "epoch": 0.6669899032954152,
-      "grad_norm": 0.9218897223472595,
-      "learning_rate": 3.3299232736572896e-06,
-      "loss": 0.1004,
-      "step": 3912
-    },
-    {
-      "epoch": 0.6671604017369529,
-      "grad_norm": 3.1962733268737793,
-      "learning_rate": 3.328218243819267e-06,
-      "loss": 0.1432,
-      "step": 3913
-    },
-    {
-      "epoch": 0.6673309001784905,
-      "grad_norm": 0.97951740026474,
-      "learning_rate": 3.326513213981245e-06,
-      "loss": 0.0763,
-      "step": 3914
-    },
-    {
-      "epoch": 0.6675013986200282,
-      "grad_norm": 1.1191253662109375,
-      "learning_rate": 3.3248081841432226e-06,
-      "loss": 0.1483,
-      "step": 3915
-    },
-    {
-      "epoch": 0.6676718970615659,
-      "grad_norm": 1.0806386470794678,
-      "learning_rate": 3.3231031543052006e-06,
-      "loss": 0.0958,
-      "step": 3916
-    },
-    {
-      "epoch": 0.6678423955031036,
-      "grad_norm": 2.092670440673828,
-      "learning_rate": 3.3213981244671785e-06,
-      "loss": 0.1947,
-      "step": 3917
-    },
-    {
-      "epoch": 0.6680128939446413,
-      "grad_norm": 1.6471540927886963,
-      "learning_rate": 3.319693094629156e-06,
-      "loss": 0.1681,
-      "step": 3918
-    },
-    {
-      "epoch": 0.668183392386179,
-      "grad_norm": 1.5735267400741577,
-      "learning_rate": 3.317988064791134e-06,
-      "loss": 0.1907,
-      "step": 3919
-    },
-    {
-      "epoch": 0.6683538908277167,
-      "grad_norm": 1.0324760675430298,
-      "learning_rate": 3.3162830349531124e-06,
-      "loss": 0.1219,
-      "step": 3920
-    },
-    {
-      "epoch": 0.6685243892692543,
-      "grad_norm": 1.28639817237854,
-      "learning_rate": 3.31457800511509e-06,
-      "loss": 0.1547,
-      "step": 3921
-    },
-    {
-      "epoch": 0.668694887710792,
-      "grad_norm": 0.7513296604156494,
-      "learning_rate": 3.312872975277068e-06,
-      "loss": 0.0756,
-      "step": 3922
-    },
-    {
-      "epoch": 0.6688653861523297,
-      "grad_norm": 0.8335238695144653,
-      "learning_rate": 3.3111679454390454e-06,
-      "loss": 0.1004,
-      "step": 3923
-    },
-    {
-      "epoch": 0.6690358845938674,
-      "grad_norm": 1.2294464111328125,
-      "learning_rate": 3.3094629156010233e-06,
-      "loss": 0.1355,
-      "step": 3924
-    },
-    {
-      "epoch": 0.669206383035405,
-      "grad_norm": 1.066673755645752,
-      "learning_rate": 3.307757885763001e-06,
-      "loss": 0.0938,
-      "step": 3925
-    },
-    {
-      "epoch": 0.6693768814769427,
-      "grad_norm": 1.3101329803466797,
-      "learning_rate": 3.306052855924979e-06,
-      "loss": 0.1462,
-      "step": 3926
-    },
-    {
-      "epoch": 0.6695473799184805,
-      "grad_norm": 1.177607536315918,
-      "learning_rate": 3.3043478260869567e-06,
-      "loss": 0.1007,
-      "step": 3927
-    },
-    {
-      "epoch": 0.6697178783600182,
-      "grad_norm": 1.1281089782714844,
-      "learning_rate": 3.3026427962489343e-06,
-      "loss": 0.1019,
-      "step": 3928
-    },
-    {
-      "epoch": 0.6698883768015558,
-      "grad_norm": 1.3807610273361206,
-      "learning_rate": 3.3009377664109126e-06,
-      "loss": 0.1984,
-      "step": 3929
-    },
-    {
-      "epoch": 0.6700588752430935,
-      "grad_norm": 2.5573902130126953,
-      "learning_rate": 3.2992327365728906e-06,
-      "loss": 0.2041,
-      "step": 3930
-    },
-    {
-      "epoch": 0.6702293736846312,
-      "grad_norm": 1.2900224924087524,
-      "learning_rate": 3.297527706734868e-06,
-      "loss": 0.1484,
-      "step": 3931
-    },
-    {
-      "epoch": 0.6703998721261688,
-      "grad_norm": 1.0510343313217163,
-      "learning_rate": 3.295822676896846e-06,
-      "loss": 0.1248,
-      "step": 3932
-    },
-    {
-      "epoch": 0.6705703705677065,
-      "grad_norm": 1.1407992839813232,
-      "learning_rate": 3.2941176470588236e-06,
-      "loss": 0.1056,
-      "step": 3933
-    },
-    {
-      "epoch": 0.6707408690092442,
-      "grad_norm": 1.8074103593826294,
-      "learning_rate": 3.2924126172208015e-06,
-      "loss": 0.1311,
-      "step": 3934
-    },
-    {
-      "epoch": 0.670911367450782,
-      "grad_norm": 1.2502901554107666,
-      "learning_rate": 3.2907075873827795e-06,
-      "loss": 0.1774,
-      "step": 3935
-    },
-    {
-      "epoch": 0.6710818658923196,
-      "grad_norm": 1.6198917627334595,
-      "learning_rate": 3.289002557544757e-06,
-      "loss": 0.1086,
-      "step": 3936
-    },
-    {
-      "epoch": 0.6712523643338573,
-      "grad_norm": 1.1426345109939575,
-      "learning_rate": 3.2872975277067354e-06,
-      "loss": 0.1018,
-      "step": 3937
-    },
-    {
-      "epoch": 0.671422862775395,
-      "grad_norm": 1.3190062046051025,
-      "learning_rate": 3.2855924978687133e-06,
-      "loss": 0.1381,
-      "step": 3938
-    },
-    {
-      "epoch": 0.6715933612169326,
-      "grad_norm": 1.0100187063217163,
-      "learning_rate": 3.283887468030691e-06,
-      "loss": 0.0888,
-      "step": 3939
-    },
-    {
-      "epoch": 0.6717638596584703,
-      "grad_norm": 1.5038955211639404,
-      "learning_rate": 3.282182438192669e-06,
-      "loss": 0.1789,
-      "step": 3940
-    },
-    {
-      "epoch": 0.671934358100008,
-      "grad_norm": 1.029555320739746,
-      "learning_rate": 3.2804774083546463e-06,
-      "loss": 0.0934,
-      "step": 3941
-    },
-    {
-      "epoch": 0.6721048565415457,
-      "grad_norm": 2.1041479110717773,
-      "learning_rate": 3.2787723785166243e-06,
-      "loss": 0.1591,
-      "step": 3942
-    },
-    {
-      "epoch": 0.6722753549830833,
-      "grad_norm": 0.9763666987419128,
-      "learning_rate": 3.2770673486786018e-06,
-      "loss": 0.059,
-      "step": 3943
-    },
-    {
-      "epoch": 0.672445853424621,
-      "grad_norm": 1.5250053405761719,
-      "learning_rate": 3.2753623188405797e-06,
-      "loss": 0.1455,
-      "step": 3944
-    },
-    {
-      "epoch": 0.6726163518661588,
-      "grad_norm": 1.7633816003799438,
-      "learning_rate": 3.273657289002558e-06,
-      "loss": 0.2283,
-      "step": 3945
-    },
-    {
-      "epoch": 0.6727868503076964,
-      "grad_norm": 1.6598435640335083,
-      "learning_rate": 3.271952259164536e-06,
-      "loss": 0.1581,
-      "step": 3946
-    },
-    {
-      "epoch": 0.6729573487492341,
-      "grad_norm": 1.2925745248794556,
-      "learning_rate": 3.2702472293265136e-06,
-      "loss": 0.1303,
-      "step": 3947
-    },
-    {
-      "epoch": 0.6731278471907718,
-      "grad_norm": 1.2952626943588257,
-      "learning_rate": 3.2685421994884915e-06,
-      "loss": 0.1938,
-      "step": 3948
-    },
-    {
-      "epoch": 0.6732983456323095,
-      "grad_norm": 1.3310290575027466,
-      "learning_rate": 3.266837169650469e-06,
-      "loss": 0.0832,
-      "step": 3949
-    },
-    {
-      "epoch": 0.6734688440738471,
-      "grad_norm": 1.3135647773742676,
-      "learning_rate": 3.265132139812447e-06,
-      "loss": 0.1581,
-      "step": 3950
-    },
-    {
-      "epoch": 0.6736393425153848,
-      "grad_norm": 1.2514424324035645,
-      "learning_rate": 3.2634271099744245e-06,
-      "loss": 0.1717,
-      "step": 3951
-    },
-    {
-      "epoch": 0.6738098409569225,
-      "grad_norm": 2.424001693725586,
-      "learning_rate": 3.2617220801364025e-06,
-      "loss": 0.2024,
-      "step": 3952
-    },
-    {
-      "epoch": 0.6739803393984601,
-      "grad_norm": 1.5078508853912354,
-      "learning_rate": 3.2600170502983804e-06,
-      "loss": 0.2054,
-      "step": 3953
-    },
-    {
-      "epoch": 0.6741508378399979,
-      "grad_norm": 1.4088560342788696,
-      "learning_rate": 3.2583120204603584e-06,
-      "loss": 0.1318,
-      "step": 3954
-    },
-    {
-      "epoch": 0.6743213362815356,
-      "grad_norm": 1.03024423122406,
-      "learning_rate": 3.2566069906223363e-06,
-      "loss": 0.123,
-      "step": 3955
-    },
-    {
-      "epoch": 0.6744918347230733,
-      "grad_norm": 1.146548867225647,
-      "learning_rate": 3.2549019607843143e-06,
-      "loss": 0.1411,
-      "step": 3956
-    },
-    {
-      "epoch": 0.6746623331646109,
-      "grad_norm": 1.3883435726165771,
-      "learning_rate": 3.2531969309462918e-06,
-      "loss": 0.1419,
-      "step": 3957
-    },
-    {
-      "epoch": 0.6748328316061486,
-      "grad_norm": 1.400600552558899,
-      "learning_rate": 3.2514919011082697e-06,
-      "loss": 0.1479,
-      "step": 3958
-    },
-    {
-      "epoch": 0.6750033300476863,
-      "grad_norm": 0.9958585500717163,
-      "learning_rate": 3.2497868712702473e-06,
-      "loss": 0.0901,
-      "step": 3959
-    },
-    {
-      "epoch": 0.6751738284892239,
-      "grad_norm": 1.0494383573532104,
-      "learning_rate": 3.248081841432225e-06,
-      "loss": 0.0813,
-      "step": 3960
-    },
-    {
-      "epoch": 0.6753443269307616,
-      "grad_norm": 2.5229740142822266,
-      "learning_rate": 3.2463768115942027e-06,
-      "loss": 0.1091,
-      "step": 3961
-    },
-    {
-      "epoch": 0.6755148253722993,
-      "grad_norm": 1.1590737104415894,
-      "learning_rate": 3.244671781756181e-06,
-      "loss": 0.074,
-      "step": 3962
-    },
-    {
-      "epoch": 0.6756853238138371,
-      "grad_norm": 1.0404776334762573,
-      "learning_rate": 3.242966751918159e-06,
-      "loss": 0.0466,
-      "step": 3963
-    },
-    {
-      "epoch": 0.6758558222553747,
-      "grad_norm": 0.9535297751426697,
-      "learning_rate": 3.241261722080137e-06,
-      "loss": 0.1187,
-      "step": 3964
-    },
-    {
-      "epoch": 0.6760263206969124,
-      "grad_norm": 1.555312991142273,
-      "learning_rate": 3.2395566922421145e-06,
-      "loss": 0.0863,
-      "step": 3965
-    },
-    {
-      "epoch": 0.6761968191384501,
-      "grad_norm": 1.0932700634002686,
-      "learning_rate": 3.2378516624040925e-06,
-      "loss": 0.1172,
-      "step": 3966
-    },
-    {
-      "epoch": 0.6763673175799877,
-      "grad_norm": 1.41453218460083,
-      "learning_rate": 3.23614663256607e-06,
-      "loss": 0.1164,
-      "step": 3967
-    },
-    {
-      "epoch": 0.6765378160215254,
-      "grad_norm": 1.6799843311309814,
-      "learning_rate": 3.234441602728048e-06,
-      "loss": 0.1375,
-      "step": 3968
-    },
-    {
-      "epoch": 0.6767083144630631,
-      "grad_norm": 1.497155785560608,
-      "learning_rate": 3.2327365728900255e-06,
-      "loss": 0.1625,
-      "step": 3969
-    },
-    {
-      "epoch": 0.6768788129046008,
-      "grad_norm": 1.5166429281234741,
-      "learning_rate": 3.231031543052004e-06,
-      "loss": 0.1459,
-      "step": 3970
-    },
-    {
-      "epoch": 0.6770493113461384,
-      "grad_norm": 1.0572724342346191,
-      "learning_rate": 3.2293265132139818e-06,
-      "loss": 0.0913,
-      "step": 3971
-    },
-    {
-      "epoch": 0.6772198097876762,
-      "grad_norm": 1.0608195066452026,
-      "learning_rate": 3.2276214833759593e-06,
-      "loss": 0.102,
-      "step": 3972
-    },
-    {
-      "epoch": 0.6773903082292139,
-      "grad_norm": 1.6367521286010742,
-      "learning_rate": 3.2259164535379373e-06,
-      "loss": 0.151,
-      "step": 3973
-    },
-    {
-      "epoch": 0.6775608066707515,
-      "grad_norm": 1.0436595678329468,
-      "learning_rate": 3.224211423699915e-06,
-      "loss": 0.0479,
-      "step": 3974
-    },
-    {
-      "epoch": 0.6777313051122892,
-      "grad_norm": 0.6527305245399475,
-      "learning_rate": 3.2225063938618927e-06,
-      "loss": 0.0419,
-      "step": 3975
-    },
-    {
-      "epoch": 0.6779018035538269,
-      "grad_norm": 1.0341905355453491,
-      "learning_rate": 3.2208013640238707e-06,
-      "loss": 0.1052,
-      "step": 3976
-    },
-    {
-      "epoch": 0.6780723019953646,
-      "grad_norm": 1.846094012260437,
-      "learning_rate": 3.219096334185848e-06,
-      "loss": 0.1429,
-      "step": 3977
-    },
-    {
-      "epoch": 0.6782428004369022,
-      "grad_norm": 1.3365362882614136,
-      "learning_rate": 3.217391304347826e-06,
-      "loss": 0.176,
-      "step": 3978
-    },
-    {
-      "epoch": 0.6784132988784399,
-      "grad_norm": 1.9188679456710815,
-      "learning_rate": 3.2156862745098045e-06,
-      "loss": 0.1989,
-      "step": 3979
-    },
-    {
-      "epoch": 0.6785837973199776,
-      "grad_norm": 0.7098460793495178,
-      "learning_rate": 3.213981244671782e-06,
-      "loss": 0.0572,
-      "step": 3980
-    },
-    {
-      "epoch": 0.6787542957615154,
-      "grad_norm": 1.2970669269561768,
-      "learning_rate": 3.21227621483376e-06,
-      "loss": 0.0805,
-      "step": 3981
-    },
-    {
-      "epoch": 0.678924794203053,
-      "grad_norm": 0.8300220966339111,
-      "learning_rate": 3.2105711849957375e-06,
-      "loss": 0.0349,
-      "step": 3982
-    },
-    {
-      "epoch": 0.6790952926445907,
-      "grad_norm": 1.6387343406677246,
-      "learning_rate": 3.2088661551577155e-06,
-      "loss": 0.1912,
-      "step": 3983
-    },
-    {
-      "epoch": 0.6792657910861284,
-      "grad_norm": 1.5236952304840088,
-      "learning_rate": 3.2071611253196934e-06,
-      "loss": 0.1046,
-      "step": 3984
-    },
-    {
-      "epoch": 0.679436289527666,
-      "grad_norm": 1.3057942390441895,
-      "learning_rate": 3.205456095481671e-06,
-      "loss": 0.1108,
-      "step": 3985
-    },
-    {
-      "epoch": 0.6796067879692037,
-      "grad_norm": 1.1730042695999146,
-      "learning_rate": 3.203751065643649e-06,
-      "loss": 0.0957,
-      "step": 3986
-    },
-    {
-      "epoch": 0.6797772864107414,
-      "grad_norm": 1.2910196781158447,
-      "learning_rate": 3.2020460358056272e-06,
-      "loss": 0.0905,
-      "step": 3987
-    },
-    {
-      "epoch": 0.6799477848522791,
-      "grad_norm": 1.2115230560302734,
-      "learning_rate": 3.2003410059676048e-06,
-      "loss": 0.1406,
-      "step": 3988
-    },
-    {
-      "epoch": 0.6801182832938167,
-      "grad_norm": 1.7413958311080933,
-      "learning_rate": 3.1986359761295827e-06,
-      "loss": 0.1538,
-      "step": 3989
-    },
-    {
-      "epoch": 0.6802887817353545,
-      "grad_norm": 1.3703621625900269,
-      "learning_rate": 3.1969309462915602e-06,
-      "loss": 0.1362,
-      "step": 3990
-    },
-    {
-      "epoch": 0.6804592801768922,
-      "grad_norm": 1.4095655679702759,
-      "learning_rate": 3.195225916453538e-06,
-      "loss": 0.1855,
-      "step": 3991
-    },
-    {
-      "epoch": 0.6806297786184298,
-      "grad_norm": 1.5323487520217896,
-      "learning_rate": 3.193520886615516e-06,
-      "loss": 0.1647,
-      "step": 3992
-    },
-    {
-      "epoch": 0.6808002770599675,
-      "grad_norm": 1.5383172035217285,
-      "learning_rate": 3.1918158567774937e-06,
-      "loss": 0.1779,
-      "step": 3993
-    },
-    {
-      "epoch": 0.6809707755015052,
-      "grad_norm": 0.8922589421272278,
-      "learning_rate": 3.1901108269394716e-06,
-      "loss": 0.0839,
-      "step": 3994
-    },
-    {
-      "epoch": 0.6811412739430429,
-      "grad_norm": 1.0920547246932983,
-      "learning_rate": 3.188405797101449e-06,
-      "loss": 0.0963,
-      "step": 3995
-    },
-    {
-      "epoch": 0.6813117723845805,
-      "grad_norm": 1.4073317050933838,
-      "learning_rate": 3.1867007672634275e-06,
-      "loss": 0.1311,
-      "step": 3996
-    },
-    {
-      "epoch": 0.6814822708261182,
-      "grad_norm": 1.1026973724365234,
-      "learning_rate": 3.1849957374254055e-06,
-      "loss": 0.0933,
-      "step": 3997
-    },
-    {
-      "epoch": 0.681652769267656,
-      "grad_norm": 0.6690061688423157,
-      "learning_rate": 3.183290707587383e-06,
-      "loss": 0.0364,
-      "step": 3998
-    },
-    {
-      "epoch": 0.6818232677091935,
-      "grad_norm": 0.898353099822998,
-      "learning_rate": 3.181585677749361e-06,
-      "loss": 0.0633,
-      "step": 3999
-    },
-    {
-      "epoch": 0.6819937661507313,
-      "grad_norm": 1.5099992752075195,
-      "learning_rate": 3.1798806479113385e-06,
-      "loss": 0.1217,
-      "step": 4000
-    },
-    {
-      "epoch": 0.6819937661507313,
-      "eval_f1_score": 0.40487804878048783,
-      "eval_loss": 0.13649117946624756,
-      "eval_runtime": 182.8572,
-      "eval_samples_per_second": 54.687,
-      "eval_steps_per_second": 3.418,
-      "step": 4000
-    },
-    {
-      "epoch": 0.682164264592269,
-      "grad_norm": 1.6470274925231934,
-      "learning_rate": 3.1781756180733164e-06,
-      "loss": 0.1265,
-      "step": 4001
-    },
-    {
-      "epoch": 0.6823347630338067,
-      "grad_norm": 1.2008510828018188,
-      "learning_rate": 3.1764705882352943e-06,
-      "loss": 0.1013,
-      "step": 4002
-    },
-    {
-      "epoch": 0.6825052614753443,
-      "grad_norm": 1.8212544918060303,
-      "learning_rate": 3.174765558397272e-06,
-      "loss": 0.0905,
-      "step": 4003
-    },
-    {
-      "epoch": 0.682675759916882,
-      "grad_norm": 1.2132388353347778,
-      "learning_rate": 3.1730605285592502e-06,
-      "loss": 0.1148,
-      "step": 4004
-    },
-    {
-      "epoch": 0.6828462583584197,
-      "grad_norm": 1.000791311264038,
-      "learning_rate": 3.171355498721228e-06,
-      "loss": 0.0653,
-      "step": 4005
-    },
-    {
-      "epoch": 0.6830167567999573,
-      "grad_norm": 0.9185832738876343,
-      "learning_rate": 3.1696504688832057e-06,
-      "loss": 0.0586,
-      "step": 4006
-    },
-    {
-      "epoch": 0.683187255241495,
-      "grad_norm": 0.9736605882644653,
-      "learning_rate": 3.1679454390451837e-06,
-      "loss": 0.0481,
-      "step": 4007
-    },
-    {
-      "epoch": 0.6833577536830328,
-      "grad_norm": 1.5643205642700195,
-      "learning_rate": 3.166240409207161e-06,
-      "loss": 0.1509,
-      "step": 4008
-    },
-    {
-      "epoch": 0.6835282521245705,
-      "grad_norm": 1.8469020128250122,
-      "learning_rate": 3.164535379369139e-06,
-      "loss": 0.1601,
-      "step": 4009
-    },
-    {
-      "epoch": 0.6836987505661081,
-      "grad_norm": 2.011707067489624,
-      "learning_rate": 3.162830349531117e-06,
-      "loss": 0.1713,
-      "step": 4010
-    },
-    {
-      "epoch": 0.6838692490076458,
-      "grad_norm": 1.1956074237823486,
-      "learning_rate": 3.1611253196930946e-06,
-      "loss": 0.118,
-      "step": 4011
-    },
-    {
-      "epoch": 0.6840397474491835,
-      "grad_norm": 0.9073688387870789,
-      "learning_rate": 3.159420289855073e-06,
-      "loss": 0.0697,
-      "step": 4012
-    },
-    {
-      "epoch": 0.6842102458907211,
-      "grad_norm": 1.6583490371704102,
-      "learning_rate": 3.157715260017051e-06,
-      "loss": 0.1523,
-      "step": 4013
-    },
-    {
-      "epoch": 0.6843807443322588,
-      "grad_norm": 0.9067796468734741,
-      "learning_rate": 3.1560102301790284e-06,
-      "loss": 0.0845,
-      "step": 4014
-    },
-    {
-      "epoch": 0.6845512427737965,
-      "grad_norm": 1.350319743156433,
-      "learning_rate": 3.1543052003410064e-06,
-      "loss": 0.1404,
-      "step": 4015
-    },
-    {
-      "epoch": 0.6847217412153342,
-      "grad_norm": 0.9945334792137146,
-      "learning_rate": 3.152600170502984e-06,
-      "loss": 0.0774,
-      "step": 4016
-    },
-    {
-      "epoch": 0.6848922396568718,
-      "grad_norm": 1.5443482398986816,
-      "learning_rate": 3.150895140664962e-06,
-      "loss": 0.1621,
-      "step": 4017
-    },
-    {
-      "epoch": 0.6850627380984096,
-      "grad_norm": 1.2873508930206299,
-      "learning_rate": 3.1491901108269394e-06,
-      "loss": 0.093,
-      "step": 4018
-    },
-    {
-      "epoch": 0.6852332365399473,
-      "grad_norm": 1.3037817478179932,
-      "learning_rate": 3.1474850809889173e-06,
-      "loss": 0.1174,
-      "step": 4019
-    },
-    {
-      "epoch": 0.6854037349814849,
-      "grad_norm": 1.2276246547698975,
-      "learning_rate": 3.1457800511508953e-06,
-      "loss": 0.0997,
-      "step": 4020
-    },
-    {
-      "epoch": 0.6855742334230226,
-      "grad_norm": 0.9074063897132874,
-      "learning_rate": 3.1440750213128737e-06,
-      "loss": 0.0857,
-      "step": 4021
-    },
-    {
-      "epoch": 0.6857447318645603,
-      "grad_norm": 2.060265302658081,
-      "learning_rate": 3.142369991474851e-06,
-      "loss": 0.1786,
-      "step": 4022
-    },
-    {
-      "epoch": 0.685915230306098,
-      "grad_norm": 1.4568548202514648,
-      "learning_rate": 3.140664961636829e-06,
-      "loss": 0.1515,
-      "step": 4023
-    },
-    {
-      "epoch": 0.6860857287476356,
-      "grad_norm": 0.9228938817977905,
-      "learning_rate": 3.1389599317988067e-06,
-      "loss": 0.0982,
-      "step": 4024
-    },
-    {
-      "epoch": 0.6862562271891733,
-      "grad_norm": 1.6610695123672485,
-      "learning_rate": 3.1372549019607846e-06,
-      "loss": 0.1442,
-      "step": 4025
-    },
-    {
-      "epoch": 0.686426725630711,
-      "grad_norm": 2.3436145782470703,
-      "learning_rate": 3.135549872122762e-06,
-      "loss": 0.1543,
-      "step": 4026
-    },
-    {
-      "epoch": 0.6865972240722488,
-      "grad_norm": 1.4873566627502441,
-      "learning_rate": 3.13384484228474e-06,
-      "loss": 0.1753,
-      "step": 4027
-    },
-    {
-      "epoch": 0.6867677225137864,
-      "grad_norm": 1.3169639110565186,
-      "learning_rate": 3.132139812446718e-06,
-      "loss": 0.0885,
-      "step": 4028
-    },
-    {
-      "epoch": 0.6869382209553241,
-      "grad_norm": 1.3342516422271729,
-      "learning_rate": 3.130434782608696e-06,
-      "loss": 0.1357,
-      "step": 4029
-    },
-    {
-      "epoch": 0.6871087193968618,
-      "grad_norm": 1.2794122695922852,
-      "learning_rate": 3.128729752770674e-06,
-      "loss": 0.1444,
-      "step": 4030
-    },
-    {
-      "epoch": 0.6872792178383994,
-      "grad_norm": 1.5904749631881714,
-      "learning_rate": 3.127024722932652e-06,
-      "loss": 0.133,
-      "step": 4031
-    },
-    {
-      "epoch": 0.6874497162799371,
-      "grad_norm": 1.5989582538604736,
-      "learning_rate": 3.1253196930946294e-06,
-      "loss": 0.1855,
-      "step": 4032
-    },
-    {
-      "epoch": 0.6876202147214748,
-      "grad_norm": 1.0583912134170532,
-      "learning_rate": 3.1236146632566073e-06,
-      "loss": 0.093,
-      "step": 4033
-    },
-    {
-      "epoch": 0.6877907131630125,
-      "grad_norm": 0.9540891647338867,
-      "learning_rate": 3.121909633418585e-06,
-      "loss": 0.0612,
-      "step": 4034
-    },
-    {
-      "epoch": 0.6879612116045501,
-      "grad_norm": 1.2029814720153809,
-      "learning_rate": 3.120204603580563e-06,
-      "loss": 0.1068,
-      "step": 4035
-    },
-    {
-      "epoch": 0.6881317100460879,
-      "grad_norm": 1.2139424085617065,
-      "learning_rate": 3.1184995737425403e-06,
-      "loss": 0.1181,
-      "step": 4036
-    },
-    {
-      "epoch": 0.6883022084876256,
-      "grad_norm": 1.311047077178955,
-      "learning_rate": 3.1167945439045187e-06,
-      "loss": 0.1427,
-      "step": 4037
-    },
-    {
-      "epoch": 0.6884727069291632,
-      "grad_norm": 1.864897608757019,
-      "learning_rate": 3.1150895140664967e-06,
-      "loss": 0.1878,
-      "step": 4038
-    },
-    {
-      "epoch": 0.6886432053707009,
-      "grad_norm": 1.1222472190856934,
-      "learning_rate": 3.1133844842284746e-06,
-      "loss": 0.0889,
-      "step": 4039
-    },
-    {
-      "epoch": 0.6888137038122386,
-      "grad_norm": 1.16517174243927,
-      "learning_rate": 3.111679454390452e-06,
-      "loss": 0.0943,
-      "step": 4040
-    },
-    {
-      "epoch": 0.6889842022537763,
-      "grad_norm": 0.8637740015983582,
-      "learning_rate": 3.10997442455243e-06,
-      "loss": 0.083,
-      "step": 4041
-    },
-    {
-      "epoch": 0.6891547006953139,
-      "grad_norm": 1.0583702325820923,
-      "learning_rate": 3.1082693947144076e-06,
-      "loss": 0.0842,
-      "step": 4042
-    },
-    {
-      "epoch": 0.6893251991368516,
-      "grad_norm": 2.0392816066741943,
-      "learning_rate": 3.1065643648763855e-06,
-      "loss": 0.1701,
-      "step": 4043
-    },
-    {
-      "epoch": 0.6894956975783894,
-      "grad_norm": 0.7414406538009644,
-      "learning_rate": 3.104859335038363e-06,
-      "loss": 0.0429,
-      "step": 4044
-    },
-    {
-      "epoch": 0.689666196019927,
-      "grad_norm": 1.78953218460083,
-      "learning_rate": 3.103154305200341e-06,
-      "loss": 0.1752,
-      "step": 4045
-    },
-    {
-      "epoch": 0.6898366944614647,
-      "grad_norm": 1.2555387020111084,
-      "learning_rate": 3.1014492753623194e-06,
-      "loss": 0.1371,
-      "step": 4046
-    },
-    {
-      "epoch": 0.6900071929030024,
-      "grad_norm": 1.6776463985443115,
-      "learning_rate": 3.099744245524297e-06,
-      "loss": 0.1209,
-      "step": 4047
-    },
-    {
-      "epoch": 0.6901776913445401,
-      "grad_norm": 1.0086376667022705,
-      "learning_rate": 3.098039215686275e-06,
-      "loss": 0.0673,
-      "step": 4048
-    },
-    {
-      "epoch": 0.6903481897860777,
-      "grad_norm": 1.0416656732559204,
-      "learning_rate": 3.096334185848253e-06,
-      "loss": 0.0607,
-      "step": 4049
-    },
-    {
-      "epoch": 0.6905186882276154,
-      "grad_norm": 2.047964572906494,
-      "learning_rate": 3.0946291560102303e-06,
-      "loss": 0.2094,
-      "step": 4050
-    },
-    {
-      "epoch": 0.6906891866691531,
-      "grad_norm": 0.8563204407691956,
-      "learning_rate": 3.0929241261722083e-06,
-      "loss": 0.0883,
-      "step": 4051
-    },
-    {
-      "epoch": 0.6908596851106907,
-      "grad_norm": 2.1121885776519775,
-      "learning_rate": 3.091219096334186e-06,
-      "loss": 0.1619,
-      "step": 4052
-    },
-    {
-      "epoch": 0.6910301835522284,
-      "grad_norm": 1.5385748147964478,
-      "learning_rate": 3.0895140664961638e-06,
-      "loss": 0.1547,
-      "step": 4053
-    },
-    {
-      "epoch": 0.6912006819937662,
-      "grad_norm": 1.6678493022918701,
-      "learning_rate": 3.087809036658142e-06,
-      "loss": 0.1065,
-      "step": 4054
-    },
-    {
-      "epoch": 0.6913711804353039,
-      "grad_norm": 1.3417023420333862,
-      "learning_rate": 3.0861040068201196e-06,
-      "loss": 0.0756,
-      "step": 4055
-    },
-    {
-      "epoch": 0.6915416788768415,
-      "grad_norm": 0.9588251709938049,
-      "learning_rate": 3.0843989769820976e-06,
-      "loss": 0.0774,
-      "step": 4056
-    },
-    {
-      "epoch": 0.6917121773183792,
-      "grad_norm": 1.2807377576828003,
-      "learning_rate": 3.082693947144075e-06,
-      "loss": 0.1068,
-      "step": 4057
-    },
-    {
-      "epoch": 0.6918826757599169,
-      "grad_norm": 1.0533798933029175,
-      "learning_rate": 3.080988917306053e-06,
-      "loss": 0.1061,
-      "step": 4058
-    },
-    {
-      "epoch": 0.6920531742014545,
-      "grad_norm": 2.105905532836914,
-      "learning_rate": 3.079283887468031e-06,
-      "loss": 0.1713,
-      "step": 4059
-    },
-    {
-      "epoch": 0.6922236726429922,
-      "grad_norm": 1.4359697103500366,
-      "learning_rate": 3.0775788576300085e-06,
-      "loss": 0.1761,
-      "step": 4060
-    },
-    {
-      "epoch": 0.6923941710845299,
-      "grad_norm": 1.4418187141418457,
-      "learning_rate": 3.0758738277919865e-06,
-      "loss": 0.1127,
-      "step": 4061
-    },
-    {
-      "epoch": 0.6925646695260677,
-      "grad_norm": 1.7965011596679688,
-      "learning_rate": 3.074168797953965e-06,
-      "loss": 0.1559,
-      "step": 4062
-    },
-    {
-      "epoch": 0.6927351679676053,
-      "grad_norm": 0.8797532916069031,
-      "learning_rate": 3.0724637681159424e-06,
-      "loss": 0.0962,
-      "step": 4063
-    },
-    {
-      "epoch": 0.692905666409143,
-      "grad_norm": 1.4908859729766846,
-      "learning_rate": 3.0707587382779203e-06,
-      "loss": 0.1333,
-      "step": 4064
-    },
-    {
-      "epoch": 0.6930761648506807,
-      "grad_norm": 1.1060150861740112,
-      "learning_rate": 3.069053708439898e-06,
-      "loss": 0.1229,
-      "step": 4065
-    },
-    {
-      "epoch": 0.6932466632922183,
-      "grad_norm": 0.9352675080299377,
-      "learning_rate": 3.067348678601876e-06,
-      "loss": 0.0877,
-      "step": 4066
-    },
-    {
-      "epoch": 0.693417161733756,
-      "grad_norm": 1.0809539556503296,
-      "learning_rate": 3.0656436487638537e-06,
-      "loss": 0.1242,
-      "step": 4067
-    },
-    {
-      "epoch": 0.6935876601752937,
-      "grad_norm": 1.624000072479248,
-      "learning_rate": 3.0639386189258313e-06,
-      "loss": 0.1616,
-      "step": 4068
-    },
-    {
-      "epoch": 0.6937581586168314,
-      "grad_norm": 1.7362465858459473,
-      "learning_rate": 3.0622335890878092e-06,
-      "loss": 0.1814,
-      "step": 4069
-    },
-    {
-      "epoch": 0.693928657058369,
-      "grad_norm": 1.0399138927459717,
-      "learning_rate": 3.0605285592497867e-06,
-      "loss": 0.1063,
-      "step": 4070
-    },
-    {
-      "epoch": 0.6940991554999068,
-      "grad_norm": 1.5530822277069092,
-      "learning_rate": 3.058823529411765e-06,
-      "loss": 0.0814,
-      "step": 4071
-    },
-    {
-      "epoch": 0.6942696539414445,
-      "grad_norm": 1.186566710472107,
-      "learning_rate": 3.057118499573743e-06,
-      "loss": 0.098,
-      "step": 4072
-    },
-    {
-      "epoch": 0.6944401523829822,
-      "grad_norm": 1.128265142440796,
-      "learning_rate": 3.0554134697357206e-06,
-      "loss": 0.0888,
-      "step": 4073
-    },
-    {
-      "epoch": 0.6946106508245198,
-      "grad_norm": 1.3690623044967651,
-      "learning_rate": 3.0537084398976985e-06,
-      "loss": 0.1201,
-      "step": 4074
-    },
-    {
-      "epoch": 0.6947811492660575,
-      "grad_norm": 1.3120359182357788,
-      "learning_rate": 3.052003410059676e-06,
-      "loss": 0.0738,
-      "step": 4075
-    },
-    {
-      "epoch": 0.6949516477075952,
-      "grad_norm": 1.1368895769119263,
-      "learning_rate": 3.050298380221654e-06,
-      "loss": 0.1423,
-      "step": 4076
-    },
-    {
-      "epoch": 0.6951221461491328,
-      "grad_norm": 1.0708452463150024,
-      "learning_rate": 3.048593350383632e-06,
-      "loss": 0.0803,
-      "step": 4077
-    },
-    {
-      "epoch": 0.6952926445906705,
-      "grad_norm": 1.069138526916504,
-      "learning_rate": 3.0468883205456095e-06,
-      "loss": 0.1072,
-      "step": 4078
-    },
-    {
-      "epoch": 0.6954631430322082,
-      "grad_norm": 1.5960021018981934,
-      "learning_rate": 3.045183290707588e-06,
-      "loss": 0.2298,
-      "step": 4079
-    },
-    {
-      "epoch": 0.695633641473746,
-      "grad_norm": 1.1349047422409058,
-      "learning_rate": 3.043478260869566e-06,
-      "loss": 0.0945,
-      "step": 4080
-    },
-    {
-      "epoch": 0.6958041399152836,
-      "grad_norm": 0.9507064819335938,
-      "learning_rate": 3.0417732310315433e-06,
-      "loss": 0.0711,
-      "step": 4081
-    },
-    {
-      "epoch": 0.6959746383568213,
-      "grad_norm": 2.050335168838501,
-      "learning_rate": 3.0400682011935213e-06,
-      "loss": 0.183,
-      "step": 4082
-    },
-    {
-      "epoch": 0.696145136798359,
-      "grad_norm": 1.2211310863494873,
-      "learning_rate": 3.038363171355499e-06,
-      "loss": 0.1362,
-      "step": 4083
-    },
-    {
-      "epoch": 0.6963156352398966,
-      "grad_norm": 1.4060834646224976,
-      "learning_rate": 3.0366581415174767e-06,
-      "loss": 0.1536,
-      "step": 4084
-    },
-    {
-      "epoch": 0.6964861336814343,
-      "grad_norm": 0.946904182434082,
-      "learning_rate": 3.0349531116794547e-06,
-      "loss": 0.084,
-      "step": 4085
-    },
-    {
-      "epoch": 0.696656632122972,
-      "grad_norm": 1.6174473762512207,
-      "learning_rate": 3.0332480818414322e-06,
-      "loss": 0.1601,
-      "step": 4086
-    },
-    {
-      "epoch": 0.6968271305645097,
-      "grad_norm": 0.7835435271263123,
-      "learning_rate": 3.03154305200341e-06,
-      "loss": 0.0676,
-      "step": 4087
-    },
-    {
-      "epoch": 0.6969976290060473,
-      "grad_norm": 0.8758642673492432,
-      "learning_rate": 3.0298380221653885e-06,
-      "loss": 0.0672,
-      "step": 4088
-    },
-    {
-      "epoch": 0.697168127447585,
-      "grad_norm": 1.4428333044052124,
-      "learning_rate": 3.028132992327366e-06,
-      "loss": 0.1454,
-      "step": 4089
-    },
-    {
-      "epoch": 0.6973386258891228,
-      "grad_norm": 1.5521656274795532,
-      "learning_rate": 3.026427962489344e-06,
-      "loss": 0.1404,
-      "step": 4090
-    },
-    {
-      "epoch": 0.6975091243306604,
-      "grad_norm": 1.1999118328094482,
-      "learning_rate": 3.0247229326513215e-06,
-      "loss": 0.0745,
-      "step": 4091
-    },
-    {
-      "epoch": 0.6976796227721981,
-      "grad_norm": 1.1342631578445435,
-      "learning_rate": 3.0230179028132995e-06,
-      "loss": 0.1336,
-      "step": 4092
-    },
-    {
-      "epoch": 0.6978501212137358,
-      "grad_norm": 2.592848300933838,
-      "learning_rate": 3.021312872975277e-06,
-      "loss": 0.217,
-      "step": 4093
-    },
-    {
-      "epoch": 0.6980206196552735,
-      "grad_norm": 1.6239129304885864,
-      "learning_rate": 3.019607843137255e-06,
-      "loss": 0.1574,
-      "step": 4094
-    },
-    {
-      "epoch": 0.6981911180968111,
-      "grad_norm": 0.7634371519088745,
-      "learning_rate": 3.017902813299233e-06,
-      "loss": 0.0563,
-      "step": 4095
-    },
-    {
-      "epoch": 0.6983616165383488,
-      "grad_norm": 1.2029179334640503,
-      "learning_rate": 3.0161977834612113e-06,
-      "loss": 0.0893,
-      "step": 4096
-    },
-    {
-      "epoch": 0.6985321149798865,
-      "grad_norm": 1.4013758897781372,
-      "learning_rate": 3.014492753623189e-06,
-      "loss": 0.1706,
-      "step": 4097
-    },
-    {
-      "epoch": 0.6987026134214241,
-      "grad_norm": 1.3356106281280518,
-      "learning_rate": 3.0127877237851667e-06,
-      "loss": 0.1122,
-      "step": 4098
-    },
-    {
-      "epoch": 0.6988731118629619,
-      "grad_norm": 0.9753081202507019,
-      "learning_rate": 3.0110826939471443e-06,
-      "loss": 0.0938,
-      "step": 4099
-    },
-    {
-      "epoch": 0.6990436103044996,
-      "grad_norm": 1.8733267784118652,
-      "learning_rate": 3.009377664109122e-06,
-      "loss": 0.1907,
-      "step": 4100
-    },
-    {
-      "epoch": 0.6990436103044996,
-      "eval_f1_score": 0.40389294403892945,
-      "eval_loss": 0.13582302629947662,
-      "eval_runtime": 183.3886,
-      "eval_samples_per_second": 54.529,
-      "eval_steps_per_second": 3.408,
-      "step": 4100
-    },
-    {
-      "epoch": 0.6992141087460373,
-      "grad_norm": 1.568135142326355,
-      "learning_rate": 3.0076726342710997e-06,
-      "loss": 0.164,
-      "step": 4101
-    },
-    {
-      "epoch": 0.6993846071875749,
-      "grad_norm": 1.4794212579727173,
-      "learning_rate": 3.0059676044330777e-06,
-      "loss": 0.1211,
-      "step": 4102
-    },
-    {
-      "epoch": 0.6995551056291126,
-      "grad_norm": 1.19968843460083,
-      "learning_rate": 3.0042625745950556e-06,
-      "loss": 0.0848,
-      "step": 4103
-    },
-    {
-      "epoch": 0.6997256040706503,
-      "grad_norm": 0.8885430693626404,
-      "learning_rate": 3.0025575447570336e-06,
-      "loss": 0.0861,
-      "step": 4104
-    },
-    {
-      "epoch": 0.6998961025121879,
-      "grad_norm": 1.1285784244537354,
-      "learning_rate": 3.0008525149190115e-06,
-      "loss": 0.0899,
-      "step": 4105
-    },
-    {
-      "epoch": 0.7000666009537256,
-      "grad_norm": 1.2111326456069946,
-      "learning_rate": 2.9991474850809895e-06,
-      "loss": 0.1186,
-      "step": 4106
-    },
-    {
-      "epoch": 0.7002370993952634,
-      "grad_norm": 1.571638584136963,
-      "learning_rate": 2.997442455242967e-06,
-      "loss": 0.119,
-      "step": 4107
-    },
-    {
-      "epoch": 0.7004075978368011,
-      "grad_norm": 1.3643468618392944,
-      "learning_rate": 2.995737425404945e-06,
-      "loss": 0.1753,
-      "step": 4108
-    },
-    {
-      "epoch": 0.7005780962783387,
-      "grad_norm": 1.8419893980026245,
-      "learning_rate": 2.9940323955669225e-06,
-      "loss": 0.1758,
-      "step": 4109
-    },
-    {
-      "epoch": 0.7007485947198764,
-      "grad_norm": 1.5465821027755737,
-      "learning_rate": 2.9923273657289004e-06,
-      "loss": 0.1774,
-      "step": 4110
-    },
-    {
-      "epoch": 0.7009190931614141,
-      "grad_norm": 0.8390033841133118,
-      "learning_rate": 2.990622335890878e-06,
-      "loss": 0.0841,
-      "step": 4111
-    },
-    {
-      "epoch": 0.7010895916029517,
-      "grad_norm": 1.2229130268096924,
-      "learning_rate": 2.988917306052856e-06,
-      "loss": 0.1386,
-      "step": 4112
-    },
-    {
-      "epoch": 0.7012600900444894,
-      "grad_norm": 1.1811176538467407,
-      "learning_rate": 2.9872122762148343e-06,
-      "loss": 0.1048,
-      "step": 4113
-    },
-    {
-      "epoch": 0.7014305884860271,
-      "grad_norm": 1.3642230033874512,
-      "learning_rate": 2.9855072463768118e-06,
-      "loss": 0.0605,
-      "step": 4114
-    },
-    {
-      "epoch": 0.7016010869275648,
-      "grad_norm": 2.046272039413452,
-      "learning_rate": 2.9838022165387897e-06,
-      "loss": 0.1666,
-      "step": 4115
-    },
-    {
-      "epoch": 0.7017715853691024,
-      "grad_norm": 0.9750205874443054,
-      "learning_rate": 2.9820971867007677e-06,
-      "loss": 0.1009,
-      "step": 4116
-    },
-    {
-      "epoch": 0.7019420838106402,
-      "grad_norm": 1.9477434158325195,
-      "learning_rate": 2.980392156862745e-06,
-      "loss": 0.1598,
-      "step": 4117
-    },
-    {
-      "epoch": 0.7021125822521779,
-      "grad_norm": 1.599894642829895,
-      "learning_rate": 2.978687127024723e-06,
-      "loss": 0.1349,
-      "step": 4118
-    },
-    {
-      "epoch": 0.7022830806937156,
-      "grad_norm": 0.8582209944725037,
-      "learning_rate": 2.9769820971867007e-06,
-      "loss": 0.0778,
-      "step": 4119
-    },
-    {
-      "epoch": 0.7024535791352532,
-      "grad_norm": 1.3194434642791748,
-      "learning_rate": 2.9752770673486786e-06,
-      "loss": 0.1299,
-      "step": 4120
-    },
-    {
-      "epoch": 0.7026240775767909,
-      "grad_norm": 1.538150668144226,
-      "learning_rate": 2.973572037510657e-06,
-      "loss": 0.1519,
-      "step": 4121
-    },
-    {
-      "epoch": 0.7027945760183286,
-      "grad_norm": 1.2453093528747559,
-      "learning_rate": 2.9718670076726345e-06,
-      "loss": 0.1228,
-      "step": 4122
-    },
-    {
-      "epoch": 0.7029650744598662,
-      "grad_norm": 0.9578561186790466,
-      "learning_rate": 2.9701619778346125e-06,
-      "loss": 0.0931,
-      "step": 4123
-    },
-    {
-      "epoch": 0.7031355729014039,
-      "grad_norm": 2.0141830444335938,
-      "learning_rate": 2.9684569479965904e-06,
-      "loss": 0.096,
-      "step": 4124
-    },
-    {
-      "epoch": 0.7033060713429417,
-      "grad_norm": 1.8215370178222656,
-      "learning_rate": 2.966751918158568e-06,
-      "loss": 0.1514,
-      "step": 4125
-    },
-    {
-      "epoch": 0.7034765697844794,
-      "grad_norm": 1.2377793788909912,
-      "learning_rate": 2.965046888320546e-06,
-      "loss": 0.1562,
-      "step": 4126
-    },
-    {
-      "epoch": 0.703647068226017,
-      "grad_norm": 1.0801706314086914,
-      "learning_rate": 2.9633418584825234e-06,
-      "loss": 0.1053,
-      "step": 4127
-    },
-    {
-      "epoch": 0.7038175666675547,
-      "grad_norm": 1.0231587886810303,
-      "learning_rate": 2.9616368286445014e-06,
-      "loss": 0.0982,
-      "step": 4128
-    },
-    {
-      "epoch": 0.7039880651090924,
-      "grad_norm": 1.714358925819397,
-      "learning_rate": 2.9599317988064797e-06,
-      "loss": 0.1542,
-      "step": 4129
-    },
-    {
-      "epoch": 0.70415856355063,
-      "grad_norm": 1.3221189975738525,
-      "learning_rate": 2.9582267689684573e-06,
-      "loss": 0.1671,
-      "step": 4130
-    },
-    {
-      "epoch": 0.7043290619921677,
-      "grad_norm": 1.0923004150390625,
-      "learning_rate": 2.956521739130435e-06,
-      "loss": 0.0877,
-      "step": 4131
-    },
-    {
-      "epoch": 0.7044995604337054,
-      "grad_norm": 1.122705340385437,
-      "learning_rate": 2.9548167092924127e-06,
-      "loss": 0.0929,
-      "step": 4132
-    },
-    {
-      "epoch": 0.7046700588752431,
-      "grad_norm": 1.1311506032943726,
-      "learning_rate": 2.9531116794543907e-06,
-      "loss": 0.0871,
-      "step": 4133
-    },
-    {
-      "epoch": 0.7048405573167807,
-      "grad_norm": 1.2216159105300903,
-      "learning_rate": 2.9514066496163686e-06,
-      "loss": 0.1291,
-      "step": 4134
-    },
-    {
-      "epoch": 0.7050110557583185,
-      "grad_norm": 1.0954818725585938,
-      "learning_rate": 2.949701619778346e-06,
-      "loss": 0.0986,
-      "step": 4135
-    },
-    {
-      "epoch": 0.7051815541998562,
-      "grad_norm": 1.4039642810821533,
-      "learning_rate": 2.947996589940324e-06,
-      "loss": 0.142,
-      "step": 4136
-    },
-    {
-      "epoch": 0.7053520526413938,
-      "grad_norm": 1.124967098236084,
-      "learning_rate": 2.9462915601023016e-06,
-      "loss": 0.1346,
-      "step": 4137
-    },
-    {
-      "epoch": 0.7055225510829315,
-      "grad_norm": 1.3469061851501465,
-      "learning_rate": 2.94458653026428e-06,
-      "loss": 0.0981,
-      "step": 4138
-    },
-    {
-      "epoch": 0.7056930495244692,
-      "grad_norm": 1.1395944356918335,
-      "learning_rate": 2.942881500426258e-06,
-      "loss": 0.07,
-      "step": 4139
-    },
-    {
-      "epoch": 0.7058635479660069,
-      "grad_norm": 2.9476287364959717,
-      "learning_rate": 2.9411764705882355e-06,
-      "loss": 0.1005,
-      "step": 4140
-    },
-    {
-      "epoch": 0.7060340464075445,
-      "grad_norm": 0.6996238231658936,
-      "learning_rate": 2.9394714407502134e-06,
-      "loss": 0.0563,
-      "step": 4141
-    },
-    {
-      "epoch": 0.7062045448490822,
-      "grad_norm": 1.2846816778182983,
-      "learning_rate": 2.9377664109121914e-06,
-      "loss": 0.0969,
-      "step": 4142
-    },
-    {
-      "epoch": 0.70637504329062,
-      "grad_norm": 1.0030068159103394,
-      "learning_rate": 2.936061381074169e-06,
-      "loss": 0.0635,
-      "step": 4143
-    },
-    {
-      "epoch": 0.7065455417321576,
-      "grad_norm": 1.61794114112854,
-      "learning_rate": 2.934356351236147e-06,
-      "loss": 0.1276,
-      "step": 4144
-    },
-    {
-      "epoch": 0.7067160401736953,
-      "grad_norm": 1.2051515579223633,
-      "learning_rate": 2.9326513213981244e-06,
-      "loss": 0.1229,
-      "step": 4145
-    },
-    {
-      "epoch": 0.706886538615233,
-      "grad_norm": 1.1893885135650635,
-      "learning_rate": 2.9309462915601027e-06,
-      "loss": 0.1083,
-      "step": 4146
-    },
-    {
-      "epoch": 0.7070570370567707,
-      "grad_norm": 1.0019936561584473,
-      "learning_rate": 2.9292412617220807e-06,
-      "loss": 0.0553,
-      "step": 4147
-    },
-    {
-      "epoch": 0.7072275354983083,
-      "grad_norm": 1.0891516208648682,
-      "learning_rate": 2.927536231884058e-06,
-      "loss": 0.1181,
-      "step": 4148
-    },
-    {
-      "epoch": 0.707398033939846,
-      "grad_norm": 1.939622163772583,
-      "learning_rate": 2.925831202046036e-06,
-      "loss": 0.1261,
-      "step": 4149
-    },
-    {
-      "epoch": 0.7075685323813837,
-      "grad_norm": 1.939447283744812,
-      "learning_rate": 2.9241261722080137e-06,
-      "loss": 0.2224,
-      "step": 4150
-    },
-    {
-      "epoch": 0.7077390308229213,
-      "grad_norm": 1.927830457687378,
-      "learning_rate": 2.9224211423699916e-06,
-      "loss": 0.1948,
-      "step": 4151
-    },
-    {
-      "epoch": 0.707909529264459,
-      "grad_norm": 1.7494922876358032,
-      "learning_rate": 2.9207161125319696e-06,
-      "loss": 0.1486,
-      "step": 4152
-    },
-    {
-      "epoch": 0.7080800277059968,
-      "grad_norm": 1.2902439832687378,
-      "learning_rate": 2.919011082693947e-06,
-      "loss": 0.0676,
-      "step": 4153
-    },
-    {
-      "epoch": 0.7082505261475345,
-      "grad_norm": 1.1832592487335205,
-      "learning_rate": 2.917306052855925e-06,
-      "loss": 0.1257,
-      "step": 4154
-    },
-    {
-      "epoch": 0.7084210245890721,
-      "grad_norm": 1.3747416734695435,
-      "learning_rate": 2.9156010230179034e-06,
-      "loss": 0.0945,
-      "step": 4155
-    },
-    {
-      "epoch": 0.7085915230306098,
-      "grad_norm": 1.8421868085861206,
-      "learning_rate": 2.913895993179881e-06,
-      "loss": 0.1391,
-      "step": 4156
-    },
-    {
-      "epoch": 0.7087620214721475,
-      "grad_norm": 1.0732749700546265,
-      "learning_rate": 2.912190963341859e-06,
-      "loss": 0.0927,
-      "step": 4157
-    },
-    {
-      "epoch": 0.7089325199136851,
-      "grad_norm": 0.9617205858230591,
-      "learning_rate": 2.9104859335038364e-06,
-      "loss": 0.0774,
-      "step": 4158
-    },
-    {
-      "epoch": 0.7091030183552228,
-      "grad_norm": 1.061026692390442,
-      "learning_rate": 2.9087809036658143e-06,
-      "loss": 0.0845,
-      "step": 4159
-    },
-    {
-      "epoch": 0.7092735167967605,
-      "grad_norm": 1.1157773733139038,
-      "learning_rate": 2.9070758738277923e-06,
-      "loss": 0.1153,
-      "step": 4160
-    },
-    {
-      "epoch": 0.7094440152382983,
-      "grad_norm": 1.2340047359466553,
-      "learning_rate": 2.90537084398977e-06,
-      "loss": 0.0949,
-      "step": 4161
-    },
-    {
-      "epoch": 0.7096145136798359,
-      "grad_norm": 0.9534417986869812,
-      "learning_rate": 2.9036658141517478e-06,
-      "loss": 0.1063,
-      "step": 4162
-    },
-    {
-      "epoch": 0.7097850121213736,
-      "grad_norm": 1.6519148349761963,
-      "learning_rate": 2.901960784313726e-06,
-      "loss": 0.1103,
-      "step": 4163
-    },
-    {
-      "epoch": 0.7099555105629113,
-      "grad_norm": 1.2995315790176392,
-      "learning_rate": 2.9002557544757037e-06,
-      "loss": 0.1218,
-      "step": 4164
-    },
-    {
-      "epoch": 0.710126009004449,
-      "grad_norm": 1.3705964088439941,
-      "learning_rate": 2.8985507246376816e-06,
-      "loss": 0.1569,
-      "step": 4165
-    },
-    {
-      "epoch": 0.7102965074459866,
-      "grad_norm": 1.0150760412216187,
-      "learning_rate": 2.896845694799659e-06,
-      "loss": 0.0828,
-      "step": 4166
-    },
-    {
-      "epoch": 0.7104670058875243,
-      "grad_norm": 1.2685070037841797,
-      "learning_rate": 2.895140664961637e-06,
-      "loss": 0.1415,
-      "step": 4167
-    },
-    {
-      "epoch": 0.710637504329062,
-      "grad_norm": 2.023293972015381,
-      "learning_rate": 2.8934356351236146e-06,
-      "loss": 0.1005,
-      "step": 4168
-    },
-    {
-      "epoch": 0.7108080027705996,
-      "grad_norm": 1.7603518962860107,
-      "learning_rate": 2.8917306052855926e-06,
-      "loss": 0.1835,
-      "step": 4169
-    },
-    {
-      "epoch": 0.7109785012121373,
-      "grad_norm": 1.636339545249939,
-      "learning_rate": 2.8900255754475705e-06,
-      "loss": 0.1074,
-      "step": 4170
-    },
-    {
-      "epoch": 0.7111489996536751,
-      "grad_norm": 0.8647012114524841,
-      "learning_rate": 2.888320545609549e-06,
-      "loss": 0.0743,
-      "step": 4171
-    },
-    {
-      "epoch": 0.7113194980952128,
-      "grad_norm": 1.7029750347137451,
-      "learning_rate": 2.8866155157715264e-06,
-      "loss": 0.1231,
-      "step": 4172
-    },
-    {
-      "epoch": 0.7114899965367504,
-      "grad_norm": 0.8566877245903015,
-      "learning_rate": 2.8849104859335043e-06,
-      "loss": 0.0483,
-      "step": 4173
-    },
-    {
-      "epoch": 0.7116604949782881,
-      "grad_norm": 1.5091886520385742,
-      "learning_rate": 2.883205456095482e-06,
-      "loss": 0.1455,
-      "step": 4174
-    },
-    {
-      "epoch": 0.7118309934198258,
-      "grad_norm": 1.0542126893997192,
-      "learning_rate": 2.88150042625746e-06,
-      "loss": 0.0929,
-      "step": 4175
-    },
-    {
-      "epoch": 0.7120014918613634,
-      "grad_norm": 1.0845410823822021,
-      "learning_rate": 2.8797953964194373e-06,
-      "loss": 0.0825,
-      "step": 4176
-    },
-    {
-      "epoch": 0.7121719903029011,
-      "grad_norm": 1.2993086576461792,
-      "learning_rate": 2.8780903665814153e-06,
-      "loss": 0.1161,
-      "step": 4177
-    },
-    {
-      "epoch": 0.7123424887444388,
-      "grad_norm": 1.0071661472320557,
-      "learning_rate": 2.8763853367433932e-06,
-      "loss": 0.1235,
-      "step": 4178
-    },
-    {
-      "epoch": 0.7125129871859766,
-      "grad_norm": 1.3799209594726562,
-      "learning_rate": 2.8746803069053708e-06,
-      "loss": 0.1558,
-      "step": 4179
-    },
-    {
-      "epoch": 0.7126834856275142,
-      "grad_norm": 1.4128234386444092,
-      "learning_rate": 2.872975277067349e-06,
-      "loss": 0.0733,
-      "step": 4180
-    },
-    {
-      "epoch": 0.7128539840690519,
-      "grad_norm": 2.611358880996704,
-      "learning_rate": 2.871270247229327e-06,
-      "loss": 0.162,
-      "step": 4181
-    },
-    {
-      "epoch": 0.7130244825105896,
-      "grad_norm": 1.768095850944519,
-      "learning_rate": 2.8695652173913046e-06,
-      "loss": 0.1795,
-      "step": 4182
-    },
-    {
-      "epoch": 0.7131949809521272,
-      "grad_norm": 1.983628511428833,
-      "learning_rate": 2.8678601875532826e-06,
-      "loss": 0.2206,
-      "step": 4183
-    },
-    {
-      "epoch": 0.7133654793936649,
-      "grad_norm": 0.8268606066703796,
-      "learning_rate": 2.86615515771526e-06,
-      "loss": 0.0654,
-      "step": 4184
-    },
-    {
-      "epoch": 0.7135359778352026,
-      "grad_norm": 2.2783286571502686,
-      "learning_rate": 2.864450127877238e-06,
-      "loss": 0.1699,
-      "step": 4185
-    },
-    {
-      "epoch": 0.7137064762767403,
-      "grad_norm": 1.525234341621399,
-      "learning_rate": 2.8627450980392155e-06,
-      "loss": 0.115,
-      "step": 4186
-    },
-    {
-      "epoch": 0.7138769747182779,
-      "grad_norm": 1.6166064739227295,
-      "learning_rate": 2.8610400682011935e-06,
-      "loss": 0.1621,
-      "step": 4187
-    },
-    {
-      "epoch": 0.7140474731598156,
-      "grad_norm": 1.0671355724334717,
-      "learning_rate": 2.859335038363172e-06,
-      "loss": 0.0987,
-      "step": 4188
-    },
-    {
-      "epoch": 0.7142179716013534,
-      "grad_norm": 1.1636643409729004,
-      "learning_rate": 2.8576300085251494e-06,
-      "loss": 0.0428,
-      "step": 4189
-    },
-    {
-      "epoch": 0.714388470042891,
-      "grad_norm": 1.1787340641021729,
-      "learning_rate": 2.8559249786871273e-06,
-      "loss": 0.1225,
-      "step": 4190
-    },
-    {
-      "epoch": 0.7145589684844287,
-      "grad_norm": 1.5593682527542114,
-      "learning_rate": 2.8542199488491053e-06,
-      "loss": 0.0722,
-      "step": 4191
-    },
-    {
-      "epoch": 0.7147294669259664,
-      "grad_norm": 1.4937877655029297,
-      "learning_rate": 2.852514919011083e-06,
-      "loss": 0.1429,
-      "step": 4192
-    },
-    {
-      "epoch": 0.7148999653675041,
-      "grad_norm": 1.341630220413208,
-      "learning_rate": 2.8508098891730608e-06,
-      "loss": 0.1719,
-      "step": 4193
-    },
-    {
-      "epoch": 0.7150704638090417,
-      "grad_norm": 1.9753972291946411,
-      "learning_rate": 2.8491048593350383e-06,
-      "loss": 0.1641,
-      "step": 4194
-    },
-    {
-      "epoch": 0.7152409622505794,
-      "grad_norm": 1.4761886596679688,
-      "learning_rate": 2.8473998294970162e-06,
-      "loss": 0.1814,
-      "step": 4195
-    },
-    {
-      "epoch": 0.7154114606921171,
-      "grad_norm": 1.4910000562667847,
-      "learning_rate": 2.8456947996589946e-06,
-      "loss": 0.1179,
-      "step": 4196
-    },
-    {
-      "epoch": 0.7155819591336547,
-      "grad_norm": 1.1857808828353882,
-      "learning_rate": 2.843989769820972e-06,
-      "loss": 0.1288,
-      "step": 4197
-    },
-    {
-      "epoch": 0.7157524575751925,
-      "grad_norm": 1.2303504943847656,
-      "learning_rate": 2.84228473998295e-06,
-      "loss": 0.0762,
-      "step": 4198
-    },
-    {
-      "epoch": 0.7159229560167302,
-      "grad_norm": 1.384795904159546,
-      "learning_rate": 2.840579710144928e-06,
-      "loss": 0.1227,
-      "step": 4199
-    },
-    {
-      "epoch": 0.7160934544582679,
-      "grad_norm": 1.126552939414978,
-      "learning_rate": 2.8388746803069055e-06,
-      "loss": 0.106,
-      "step": 4200
-    },
-    {
-      "epoch": 0.7160934544582679,
-      "eval_f1_score": 0.41904761904761906,
-      "eval_loss": 0.1373097002506256,
-      "eval_runtime": 183.3693,
-      "eval_samples_per_second": 54.535,
-      "eval_steps_per_second": 3.408,
-      "step": 4200
-    },
-    {
-      "epoch": 0.7162639528998055,
-      "grad_norm": 1.542959213256836,
-      "learning_rate": 2.8371696504688835e-06,
-      "loss": 0.1462,
-      "step": 4201
-    },
-    {
-      "epoch": 0.7164344513413432,
-      "grad_norm": 1.6853175163269043,
-      "learning_rate": 2.835464620630861e-06,
-      "loss": 0.155,
-      "step": 4202
-    },
-    {
-      "epoch": 0.7166049497828809,
-      "grad_norm": 1.2853667736053467,
-      "learning_rate": 2.833759590792839e-06,
-      "loss": 0.1062,
-      "step": 4203
-    },
-    {
-      "epoch": 0.7167754482244185,
-      "grad_norm": 1.1857186555862427,
-      "learning_rate": 2.8320545609548165e-06,
-      "loss": 0.1316,
-      "step": 4204
-    },
-    {
-      "epoch": 0.7169459466659562,
-      "grad_norm": 1.0130743980407715,
-      "learning_rate": 2.830349531116795e-06,
-      "loss": 0.0867,
-      "step": 4205
-    },
-    {
-      "epoch": 0.717116445107494,
-      "grad_norm": 2.006258249282837,
-      "learning_rate": 2.828644501278773e-06,
-      "loss": 0.103,
-      "step": 4206
-    },
-    {
-      "epoch": 0.7172869435490317,
-      "grad_norm": 0.9961538910865784,
-      "learning_rate": 2.8269394714407503e-06,
-      "loss": 0.1068,
-      "step": 4207
-    },
-    {
-      "epoch": 0.7174574419905693,
-      "grad_norm": 1.3043984174728394,
-      "learning_rate": 2.8252344416027283e-06,
-      "loss": 0.1233,
-      "step": 4208
-    },
-    {
-      "epoch": 0.717627940432107,
-      "grad_norm": 1.4612125158309937,
-      "learning_rate": 2.8235294117647062e-06,
-      "loss": 0.1093,
-      "step": 4209
-    },
-    {
-      "epoch": 0.7177984388736447,
-      "grad_norm": 1.5382938385009766,
-      "learning_rate": 2.8218243819266838e-06,
-      "loss": 0.1405,
-      "step": 4210
-    },
-    {
-      "epoch": 0.7179689373151823,
-      "grad_norm": 0.8454248309135437,
-      "learning_rate": 2.8201193520886617e-06,
-      "loss": 0.0629,
-      "step": 4211
-    },
-    {
-      "epoch": 0.71813943575672,
-      "grad_norm": 1.388529658317566,
-      "learning_rate": 2.8184143222506392e-06,
-      "loss": 0.1287,
-      "step": 4212
-    },
-    {
-      "epoch": 0.7183099341982577,
-      "grad_norm": 1.1707777976989746,
-      "learning_rate": 2.8167092924126176e-06,
-      "loss": 0.0902,
-      "step": 4213
-    },
-    {
-      "epoch": 0.7184804326397954,
-      "grad_norm": 1.8198975324630737,
-      "learning_rate": 2.8150042625745955e-06,
-      "loss": 0.1535,
-      "step": 4214
-    },
-    {
-      "epoch": 0.718650931081333,
-      "grad_norm": 0.971316397190094,
-      "learning_rate": 2.813299232736573e-06,
-      "loss": 0.0817,
-      "step": 4215
-    },
-    {
-      "epoch": 0.7188214295228708,
-      "grad_norm": 2.003599166870117,
-      "learning_rate": 2.811594202898551e-06,
-      "loss": 0.1548,
-      "step": 4216
-    },
-    {
-      "epoch": 0.7189919279644085,
-      "grad_norm": 1.0757346153259277,
-      "learning_rate": 2.809889173060529e-06,
-      "loss": 0.0863,
-      "step": 4217
-    },
-    {
-      "epoch": 0.7191624264059462,
-      "grad_norm": 1.4649049043655396,
-      "learning_rate": 2.8081841432225065e-06,
-      "loss": 0.0921,
-      "step": 4218
-    },
-    {
-      "epoch": 0.7193329248474838,
-      "grad_norm": 1.4503200054168701,
-      "learning_rate": 2.8064791133844844e-06,
-      "loss": 0.1385,
-      "step": 4219
-    },
-    {
-      "epoch": 0.7195034232890215,
-      "grad_norm": 1.5700362920761108,
-      "learning_rate": 2.804774083546462e-06,
-      "loss": 0.1012,
-      "step": 4220
-    },
-    {
-      "epoch": 0.7196739217305592,
-      "grad_norm": 1.4330483675003052,
-      "learning_rate": 2.8030690537084403e-06,
-      "loss": 0.1198,
-      "step": 4221
-    },
-    {
-      "epoch": 0.7198444201720968,
-      "grad_norm": 1.444413661956787,
-      "learning_rate": 2.8013640238704183e-06,
-      "loss": 0.1491,
-      "step": 4222
-    },
-    {
-      "epoch": 0.7200149186136345,
-      "grad_norm": 1.5214699506759644,
-      "learning_rate": 2.799658994032396e-06,
-      "loss": 0.1354,
-      "step": 4223
-    },
-    {
-      "epoch": 0.7201854170551723,
-      "grad_norm": 1.3857860565185547,
-      "learning_rate": 2.7979539641943737e-06,
-      "loss": 0.1061,
-      "step": 4224
-    },
-    {
-      "epoch": 0.72035591549671,
-      "grad_norm": 1.6461029052734375,
-      "learning_rate": 2.7962489343563513e-06,
-      "loss": 0.1131,
-      "step": 4225
-    },
-    {
-      "epoch": 0.7205264139382476,
-      "grad_norm": 1.16617751121521,
-      "learning_rate": 2.7945439045183292e-06,
-      "loss": 0.104,
-      "step": 4226
-    },
-    {
-      "epoch": 0.7206969123797853,
-      "grad_norm": 1.5141948461532593,
-      "learning_rate": 2.792838874680307e-06,
-      "loss": 0.1551,
-      "step": 4227
-    },
-    {
-      "epoch": 0.720867410821323,
-      "grad_norm": 1.6117165088653564,
-      "learning_rate": 2.7911338448422847e-06,
-      "loss": 0.1252,
-      "step": 4228
-    },
-    {
-      "epoch": 0.7210379092628606,
-      "grad_norm": 1.0355428457260132,
-      "learning_rate": 2.7894288150042626e-06,
-      "loss": 0.113,
-      "step": 4229
-    },
-    {
-      "epoch": 0.7212084077043983,
-      "grad_norm": 1.0461231470108032,
-      "learning_rate": 2.787723785166241e-06,
-      "loss": 0.076,
-      "step": 4230
-    },
-    {
-      "epoch": 0.721378906145936,
-      "grad_norm": 0.8947053551673889,
-      "learning_rate": 2.7860187553282185e-06,
-      "loss": 0.0648,
-      "step": 4231
-    },
-    {
-      "epoch": 0.7215494045874737,
-      "grad_norm": 0.8495338559150696,
-      "learning_rate": 2.7843137254901965e-06,
-      "loss": 0.0694,
-      "step": 4232
-    },
-    {
-      "epoch": 0.7217199030290113,
-      "grad_norm": 1.1855164766311646,
-      "learning_rate": 2.782608695652174e-06,
-      "loss": 0.119,
-      "step": 4233
-    },
-    {
-      "epoch": 0.7218904014705491,
-      "grad_norm": 1.35049569606781,
-      "learning_rate": 2.780903665814152e-06,
-      "loss": 0.0734,
-      "step": 4234
-    },
-    {
-      "epoch": 0.7220608999120868,
-      "grad_norm": 0.6907497048377991,
-      "learning_rate": 2.77919863597613e-06,
-      "loss": 0.0636,
-      "step": 4235
-    },
-    {
-      "epoch": 0.7222313983536244,
-      "grad_norm": 1.5610991716384888,
-      "learning_rate": 2.7774936061381074e-06,
-      "loss": 0.128,
-      "step": 4236
-    },
-    {
-      "epoch": 0.7224018967951621,
-      "grad_norm": 1.1115251779556274,
-      "learning_rate": 2.7757885763000854e-06,
-      "loss": 0.1234,
-      "step": 4237
-    },
-    {
-      "epoch": 0.7225723952366998,
-      "grad_norm": 1.9755184650421143,
-      "learning_rate": 2.7740835464620637e-06,
-      "loss": 0.1831,
-      "step": 4238
-    },
-    {
-      "epoch": 0.7227428936782375,
-      "grad_norm": 1.4163349866867065,
-      "learning_rate": 2.7723785166240413e-06,
-      "loss": 0.1199,
-      "step": 4239
-    },
-    {
-      "epoch": 0.7229133921197751,
-      "grad_norm": 1.1437273025512695,
-      "learning_rate": 2.7706734867860192e-06,
-      "loss": 0.1224,
-      "step": 4240
-    },
-    {
-      "epoch": 0.7230838905613128,
-      "grad_norm": 0.7897486090660095,
-      "learning_rate": 2.7689684569479967e-06,
-      "loss": 0.0404,
-      "step": 4241
-    },
-    {
-      "epoch": 0.7232543890028506,
-      "grad_norm": 0.8485286235809326,
-      "learning_rate": 2.7672634271099747e-06,
-      "loss": 0.106,
-      "step": 4242
-    },
-    {
-      "epoch": 0.7234248874443882,
-      "grad_norm": 1.5928601026535034,
-      "learning_rate": 2.7655583972719522e-06,
-      "loss": 0.1414,
-      "step": 4243
-    },
-    {
-      "epoch": 0.7235953858859259,
-      "grad_norm": 1.260733962059021,
-      "learning_rate": 2.76385336743393e-06,
-      "loss": 0.127,
-      "step": 4244
-    },
-    {
-      "epoch": 0.7237658843274636,
-      "grad_norm": 1.6917012929916382,
-      "learning_rate": 2.762148337595908e-06,
-      "loss": 0.1737,
-      "step": 4245
-    },
-    {
-      "epoch": 0.7239363827690013,
-      "grad_norm": 1.3605605363845825,
-      "learning_rate": 2.7604433077578856e-06,
-      "loss": 0.1303,
-      "step": 4246
-    },
-    {
-      "epoch": 0.7241068812105389,
-      "grad_norm": 0.9476409554481506,
-      "learning_rate": 2.758738277919864e-06,
-      "loss": 0.0679,
-      "step": 4247
-    },
-    {
-      "epoch": 0.7242773796520766,
-      "grad_norm": 1.033415675163269,
-      "learning_rate": 2.757033248081842e-06,
-      "loss": 0.083,
-      "step": 4248
-    },
-    {
-      "epoch": 0.7244478780936143,
-      "grad_norm": 1.710024118423462,
-      "learning_rate": 2.7553282182438195e-06,
-      "loss": 0.1698,
-      "step": 4249
-    },
-    {
-      "epoch": 0.7246183765351519,
-      "grad_norm": 1.2438137531280518,
-      "learning_rate": 2.7536231884057974e-06,
-      "loss": 0.0689,
-      "step": 4250
-    },
-    {
-      "epoch": 0.7247888749766896,
-      "grad_norm": 0.8587600588798523,
-      "learning_rate": 2.751918158567775e-06,
-      "loss": 0.0427,
-      "step": 4251
-    },
-    {
-      "epoch": 0.7249593734182274,
-      "grad_norm": 0.9046088457107544,
-      "learning_rate": 2.750213128729753e-06,
-      "loss": 0.0771,
-      "step": 4252
-    },
-    {
-      "epoch": 0.7251298718597651,
-      "grad_norm": 1.3793909549713135,
-      "learning_rate": 2.748508098891731e-06,
-      "loss": 0.1231,
-      "step": 4253
-    },
-    {
-      "epoch": 0.7253003703013027,
-      "grad_norm": 0.9312161207199097,
-      "learning_rate": 2.7468030690537084e-06,
-      "loss": 0.07,
-      "step": 4254
-    },
-    {
-      "epoch": 0.7254708687428404,
-      "grad_norm": 0.8483788967132568,
-      "learning_rate": 2.7450980392156867e-06,
-      "loss": 0.0578,
-      "step": 4255
-    },
-    {
-      "epoch": 0.7256413671843781,
-      "grad_norm": 0.8836938142776489,
-      "learning_rate": 2.7433930093776647e-06,
-      "loss": 0.0601,
-      "step": 4256
-    },
-    {
-      "epoch": 0.7258118656259157,
-      "grad_norm": 2.299175977706909,
-      "learning_rate": 2.7416879795396422e-06,
-      "loss": 0.263,
-      "step": 4257
-    },
-    {
-      "epoch": 0.7259823640674534,
-      "grad_norm": 1.600693941116333,
-      "learning_rate": 2.73998294970162e-06,
-      "loss": 0.1501,
-      "step": 4258
-    },
-    {
-      "epoch": 0.7261528625089911,
-      "grad_norm": 1.5393942594528198,
-      "learning_rate": 2.7382779198635977e-06,
-      "loss": 0.106,
-      "step": 4259
-    },
-    {
-      "epoch": 0.7263233609505289,
-      "grad_norm": 1.9113017320632935,
-      "learning_rate": 2.7365728900255756e-06,
-      "loss": 0.1802,
-      "step": 4260
-    },
-    {
-      "epoch": 0.7264938593920665,
-      "grad_norm": 1.0482639074325562,
-      "learning_rate": 2.734867860187553e-06,
-      "loss": 0.1159,
-      "step": 4261
-    },
-    {
-      "epoch": 0.7266643578336042,
-      "grad_norm": 1.4692020416259766,
-      "learning_rate": 2.733162830349531e-06,
-      "loss": 0.1104,
-      "step": 4262
-    },
-    {
-      "epoch": 0.7268348562751419,
-      "grad_norm": 1.8054784536361694,
-      "learning_rate": 2.7314578005115095e-06,
-      "loss": 0.1175,
-      "step": 4263
-    },
-    {
-      "epoch": 0.7270053547166796,
-      "grad_norm": 1.0958493947982788,
-      "learning_rate": 2.729752770673487e-06,
-      "loss": 0.1065,
-      "step": 4264
-    },
-    {
-      "epoch": 0.7271758531582172,
-      "grad_norm": 1.799086570739746,
-      "learning_rate": 2.728047740835465e-06,
-      "loss": 0.1377,
-      "step": 4265
-    },
-    {
-      "epoch": 0.7273463515997549,
-      "grad_norm": 1.0919666290283203,
-      "learning_rate": 2.726342710997443e-06,
-      "loss": 0.0761,
-      "step": 4266
-    },
-    {
-      "epoch": 0.7275168500412926,
-      "grad_norm": 0.9615311622619629,
-      "learning_rate": 2.7246376811594204e-06,
-      "loss": 0.1008,
-      "step": 4267
-    },
-    {
-      "epoch": 0.7276873484828302,
-      "grad_norm": 1.4227288961410522,
-      "learning_rate": 2.7229326513213984e-06,
-      "loss": 0.1297,
-      "step": 4268
-    },
-    {
-      "epoch": 0.727857846924368,
-      "grad_norm": 2.2517104148864746,
-      "learning_rate": 2.721227621483376e-06,
-      "loss": 0.1958,
-      "step": 4269
-    },
-    {
-      "epoch": 0.7280283453659057,
-      "grad_norm": 0.826910674571991,
-      "learning_rate": 2.719522591645354e-06,
-      "loss": 0.0842,
-      "step": 4270
-    },
-    {
-      "epoch": 0.7281988438074434,
-      "grad_norm": 1.608984112739563,
-      "learning_rate": 2.7178175618073318e-06,
-      "loss": 0.1673,
-      "step": 4271
-    },
-    {
-      "epoch": 0.728369342248981,
-      "grad_norm": 2.3391051292419434,
-      "learning_rate": 2.7161125319693097e-06,
-      "loss": 0.2653,
-      "step": 4272
-    },
-    {
-      "epoch": 0.7285398406905187,
-      "grad_norm": 1.6860833168029785,
-      "learning_rate": 2.7144075021312877e-06,
-      "loss": 0.142,
-      "step": 4273
-    },
-    {
-      "epoch": 0.7287103391320564,
-      "grad_norm": 1.008806586265564,
-      "learning_rate": 2.7127024722932656e-06,
-      "loss": 0.0992,
-      "step": 4274
-    },
-    {
-      "epoch": 0.728880837573594,
-      "grad_norm": 0.8580942153930664,
-      "learning_rate": 2.710997442455243e-06,
-      "loss": 0.0887,
-      "step": 4275
-    },
-    {
-      "epoch": 0.7290513360151317,
-      "grad_norm": 1.5033763647079468,
-      "learning_rate": 2.709292412617221e-06,
-      "loss": 0.1098,
-      "step": 4276
-    },
-    {
-      "epoch": 0.7292218344566694,
-      "grad_norm": 1.1777950525283813,
-      "learning_rate": 2.7075873827791986e-06,
-      "loss": 0.1614,
-      "step": 4277
-    },
-    {
-      "epoch": 0.7293923328982072,
-      "grad_norm": 1.6841565370559692,
-      "learning_rate": 2.7058823529411766e-06,
-      "loss": 0.1036,
-      "step": 4278
-    },
-    {
-      "epoch": 0.7295628313397448,
-      "grad_norm": 1.382503628730774,
-      "learning_rate": 2.704177323103154e-06,
-      "loss": 0.1626,
-      "step": 4279
-    },
-    {
-      "epoch": 0.7297333297812825,
-      "grad_norm": 0.8704547882080078,
-      "learning_rate": 2.7024722932651325e-06,
-      "loss": 0.0475,
-      "step": 4280
-    },
-    {
-      "epoch": 0.7299038282228202,
-      "grad_norm": 0.9458300471305847,
-      "learning_rate": 2.7007672634271104e-06,
-      "loss": 0.0854,
-      "step": 4281
-    },
-    {
-      "epoch": 0.7300743266643578,
-      "grad_norm": 1.3750553131103516,
-      "learning_rate": 2.699062233589088e-06,
-      "loss": 0.1525,
-      "step": 4282
-    },
-    {
-      "epoch": 0.7302448251058955,
-      "grad_norm": 1.59765625,
-      "learning_rate": 2.697357203751066e-06,
-      "loss": 0.1628,
-      "step": 4283
-    },
-    {
-      "epoch": 0.7304153235474332,
-      "grad_norm": 0.8783493638038635,
-      "learning_rate": 2.695652173913044e-06,
-      "loss": 0.0422,
-      "step": 4284
-    },
-    {
-      "epoch": 0.7305858219889709,
-      "grad_norm": 1.324453592300415,
-      "learning_rate": 2.6939471440750214e-06,
-      "loss": 0.1313,
-      "step": 4285
-    },
-    {
-      "epoch": 0.7307563204305085,
-      "grad_norm": 1.4083690643310547,
-      "learning_rate": 2.6922421142369993e-06,
-      "loss": 0.1731,
-      "step": 4286
-    },
-    {
-      "epoch": 0.7309268188720462,
-      "grad_norm": 1.37704336643219,
-      "learning_rate": 2.690537084398977e-06,
-      "loss": 0.1982,
-      "step": 4287
-    },
-    {
-      "epoch": 0.731097317313584,
-      "grad_norm": 1.0839030742645264,
-      "learning_rate": 2.688832054560955e-06,
-      "loss": 0.1559,
-      "step": 4288
-    },
-    {
-      "epoch": 0.7312678157551216,
-      "grad_norm": 0.9779359698295593,
-      "learning_rate": 2.687127024722933e-06,
-      "loss": 0.1027,
-      "step": 4289
-    },
-    {
-      "epoch": 0.7314383141966593,
-      "grad_norm": 0.768750011920929,
-      "learning_rate": 2.6854219948849107e-06,
-      "loss": 0.0665,
-      "step": 4290
-    },
-    {
-      "epoch": 0.731608812638197,
-      "grad_norm": 1.0980817079544067,
-      "learning_rate": 2.6837169650468886e-06,
-      "loss": 0.1393,
-      "step": 4291
-    },
-    {
-      "epoch": 0.7317793110797347,
-      "grad_norm": 1.476439356803894,
-      "learning_rate": 2.6820119352088666e-06,
-      "loss": 0.1317,
-      "step": 4292
-    },
-    {
-      "epoch": 0.7319498095212723,
-      "grad_norm": 2.1569950580596924,
-      "learning_rate": 2.680306905370844e-06,
-      "loss": 0.2718,
-      "step": 4293
-    },
-    {
-      "epoch": 0.73212030796281,
-      "grad_norm": 1.522475242614746,
-      "learning_rate": 2.678601875532822e-06,
-      "loss": 0.1145,
-      "step": 4294
-    },
-    {
-      "epoch": 0.7322908064043477,
-      "grad_norm": 0.9734368324279785,
-      "learning_rate": 2.6768968456947996e-06,
-      "loss": 0.0975,
-      "step": 4295
-    },
-    {
-      "epoch": 0.7324613048458853,
-      "grad_norm": 1.308054804801941,
-      "learning_rate": 2.6751918158567775e-06,
-      "loss": 0.1428,
-      "step": 4296
-    },
-    {
-      "epoch": 0.7326318032874231,
-      "grad_norm": 1.8643138408660889,
-      "learning_rate": 2.673486786018756e-06,
-      "loss": 0.1347,
-      "step": 4297
-    },
-    {
-      "epoch": 0.7328023017289608,
-      "grad_norm": 1.876632809638977,
-      "learning_rate": 2.6717817561807334e-06,
-      "loss": 0.1321,
-      "step": 4298
-    },
-    {
-      "epoch": 0.7329728001704985,
-      "grad_norm": 1.1990265846252441,
-      "learning_rate": 2.6700767263427114e-06,
-      "loss": 0.1154,
-      "step": 4299
-    },
-    {
-      "epoch": 0.7331432986120361,
-      "grad_norm": 1.6998820304870605,
-      "learning_rate": 2.668371696504689e-06,
-      "loss": 0.2055,
-      "step": 4300
-    },
-    {
-      "epoch": 0.7331432986120361,
-      "eval_f1_score": 0.38242894056847543,
-      "eval_loss": 0.13687299191951752,
-      "eval_runtime": 183.3795,
-      "eval_samples_per_second": 54.532,
-      "eval_steps_per_second": 3.408,
-      "step": 4300
-    },
-    {
-      "epoch": 0.7333137970535738,
-      "grad_norm": 1.0364716053009033,
-      "learning_rate": 2.666666666666667e-06,
-      "loss": 0.1074,
-      "step": 4301
-    },
-    {
-      "epoch": 0.7334842954951115,
-      "grad_norm": 1.6394861936569214,
-      "learning_rate": 2.6649616368286448e-06,
-      "loss": 0.14,
-      "step": 4302
-    },
-    {
-      "epoch": 0.7336547939366491,
-      "grad_norm": 1.0912730693817139,
-      "learning_rate": 2.6632566069906223e-06,
-      "loss": 0.1327,
-      "step": 4303
-    },
-    {
-      "epoch": 0.7338252923781868,
-      "grad_norm": 1.2479007244110107,
-      "learning_rate": 2.6615515771526002e-06,
-      "loss": 0.1071,
-      "step": 4304
-    },
-    {
-      "epoch": 0.7339957908197245,
-      "grad_norm": 1.0514239072799683,
-      "learning_rate": 2.6598465473145786e-06,
-      "loss": 0.0963,
-      "step": 4305
-    },
-    {
-      "epoch": 0.7341662892612623,
-      "grad_norm": 1.7217882871627808,
-      "learning_rate": 2.658141517476556e-06,
-      "loss": 0.1427,
-      "step": 4306
-    },
-    {
-      "epoch": 0.7343367877027999,
-      "grad_norm": 1.6627919673919678,
-      "learning_rate": 2.656436487638534e-06,
-      "loss": 0.1591,
-      "step": 4307
-    },
-    {
-      "epoch": 0.7345072861443376,
-      "grad_norm": 1.7781330347061157,
-      "learning_rate": 2.6547314578005116e-06,
-      "loss": 0.2108,
-      "step": 4308
-    },
-    {
-      "epoch": 0.7346777845858753,
-      "grad_norm": 0.9791902303695679,
-      "learning_rate": 2.6530264279624896e-06,
-      "loss": 0.0956,
-      "step": 4309
-    },
-    {
-      "epoch": 0.734848283027413,
-      "grad_norm": 1.3857060670852661,
-      "learning_rate": 2.6513213981244675e-06,
-      "loss": 0.1057,
-      "step": 4310
-    },
-    {
-      "epoch": 0.7350187814689506,
-      "grad_norm": 1.7408908605575562,
-      "learning_rate": 2.649616368286445e-06,
-      "loss": 0.1027,
-      "step": 4311
-    },
-    {
-      "epoch": 0.7351892799104883,
-      "grad_norm": 0.9111679792404175,
-      "learning_rate": 2.647911338448423e-06,
-      "loss": 0.0736,
-      "step": 4312
-    },
-    {
-      "epoch": 0.735359778352026,
-      "grad_norm": 1.2416996955871582,
-      "learning_rate": 2.6462063086104014e-06,
-      "loss": 0.1333,
-      "step": 4313
-    },
-    {
-      "epoch": 0.7355302767935636,
-      "grad_norm": 1.0298309326171875,
-      "learning_rate": 2.644501278772379e-06,
-      "loss": 0.11,
-      "step": 4314
-    },
-    {
-      "epoch": 0.7357007752351014,
-      "grad_norm": 0.9649758338928223,
-      "learning_rate": 2.642796248934357e-06,
-      "loss": 0.0822,
-      "step": 4315
-    },
-    {
-      "epoch": 0.7358712736766391,
-      "grad_norm": 2.0734007358551025,
-      "learning_rate": 2.6410912190963343e-06,
-      "loss": 0.2293,
-      "step": 4316
-    },
-    {
-      "epoch": 0.7360417721181768,
-      "grad_norm": 1.662545919418335,
-      "learning_rate": 2.6393861892583123e-06,
-      "loss": 0.1653,
-      "step": 4317
-    },
-    {
-      "epoch": 0.7362122705597144,
-      "grad_norm": 0.939878523349762,
-      "learning_rate": 2.63768115942029e-06,
-      "loss": 0.1111,
-      "step": 4318
-    },
-    {
-      "epoch": 0.7363827690012521,
-      "grad_norm": 1.9839119911193848,
-      "learning_rate": 2.6359761295822678e-06,
-      "loss": 0.2171,
-      "step": 4319
-    },
-    {
-      "epoch": 0.7365532674427898,
-      "grad_norm": 1.956137776374817,
-      "learning_rate": 2.6342710997442457e-06,
-      "loss": 0.1606,
-      "step": 4320
-    },
-    {
-      "epoch": 0.7367237658843274,
-      "grad_norm": 1.4634205102920532,
-      "learning_rate": 2.6325660699062232e-06,
-      "loss": 0.1347,
-      "step": 4321
-    },
-    {
-      "epoch": 0.7368942643258651,
-      "grad_norm": 1.2985095977783203,
-      "learning_rate": 2.6308610400682016e-06,
-      "loss": 0.1213,
-      "step": 4322
-    },
-    {
-      "epoch": 0.7370647627674028,
-      "grad_norm": 1.6353211402893066,
-      "learning_rate": 2.6291560102301796e-06,
-      "loss": 0.1546,
-      "step": 4323
-    },
-    {
-      "epoch": 0.7372352612089406,
-      "grad_norm": 1.1860822439193726,
-      "learning_rate": 2.627450980392157e-06,
-      "loss": 0.1052,
-      "step": 4324
-    },
-    {
-      "epoch": 0.7374057596504782,
-      "grad_norm": 0.985072910785675,
-      "learning_rate": 2.625745950554135e-06,
-      "loss": 0.0964,
-      "step": 4325
-    },
-    {
-      "epoch": 0.7375762580920159,
-      "grad_norm": 2.0165584087371826,
-      "learning_rate": 2.6240409207161126e-06,
-      "loss": 0.1631,
-      "step": 4326
-    },
-    {
-      "epoch": 0.7377467565335536,
-      "grad_norm": 1.6807523965835571,
-      "learning_rate": 2.6223358908780905e-06,
-      "loss": 0.1516,
-      "step": 4327
-    },
-    {
-      "epoch": 0.7379172549750912,
-      "grad_norm": 1.2592802047729492,
-      "learning_rate": 2.6206308610400685e-06,
-      "loss": 0.0987,
-      "step": 4328
-    },
-    {
-      "epoch": 0.7380877534166289,
-      "grad_norm": 1.1863470077514648,
-      "learning_rate": 2.618925831202046e-06,
-      "loss": 0.1345,
-      "step": 4329
-    },
-    {
-      "epoch": 0.7382582518581666,
-      "grad_norm": 2.7960240840911865,
-      "learning_rate": 2.6172208013640243e-06,
-      "loss": 0.1873,
-      "step": 4330
-    },
-    {
-      "epoch": 0.7384287502997043,
-      "grad_norm": 1.290603756904602,
-      "learning_rate": 2.6155157715260023e-06,
-      "loss": 0.1136,
-      "step": 4331
-    },
-    {
-      "epoch": 0.7385992487412419,
-      "grad_norm": 0.8807956576347351,
-      "learning_rate": 2.61381074168798e-06,
-      "loss": 0.0936,
-      "step": 4332
-    },
-    {
-      "epoch": 0.7387697471827797,
-      "grad_norm": 1.3434580564498901,
-      "learning_rate": 2.6121057118499578e-06,
-      "loss": 0.1259,
-      "step": 4333
-    },
-    {
-      "epoch": 0.7389402456243174,
-      "grad_norm": 1.2540873289108276,
-      "learning_rate": 2.6104006820119353e-06,
-      "loss": 0.0962,
-      "step": 4334
-    },
-    {
-      "epoch": 0.739110744065855,
-      "grad_norm": 2.339768648147583,
-      "learning_rate": 2.6086956521739132e-06,
-      "loss": 0.1698,
-      "step": 4335
-    },
-    {
-      "epoch": 0.7392812425073927,
-      "grad_norm": 1.055652379989624,
-      "learning_rate": 2.6069906223358908e-06,
-      "loss": 0.1213,
-      "step": 4336
-    },
-    {
-      "epoch": 0.7394517409489304,
-      "grad_norm": 1.548759937286377,
-      "learning_rate": 2.6052855924978687e-06,
-      "loss": 0.182,
-      "step": 4337
-    },
-    {
-      "epoch": 0.7396222393904681,
-      "grad_norm": 2.2357752323150635,
-      "learning_rate": 2.6035805626598467e-06,
-      "loss": 0.0535,
-      "step": 4338
-    },
-    {
-      "epoch": 0.7397927378320057,
-      "grad_norm": 1.687495470046997,
-      "learning_rate": 2.6018755328218246e-06,
-      "loss": 0.1639,
-      "step": 4339
-    },
-    {
-      "epoch": 0.7399632362735434,
-      "grad_norm": 1.113901972770691,
-      "learning_rate": 2.6001705029838026e-06,
-      "loss": 0.115,
-      "step": 4340
-    },
-    {
-      "epoch": 0.7401337347150811,
-      "grad_norm": 1.1302788257598877,
-      "learning_rate": 2.5984654731457805e-06,
-      "loss": 0.0996,
-      "step": 4341
-    },
-    {
-      "epoch": 0.7403042331566188,
-      "grad_norm": 1.3049522638320923,
-      "learning_rate": 2.596760443307758e-06,
-      "loss": 0.1667,
-      "step": 4342
-    },
-    {
-      "epoch": 0.7404747315981565,
-      "grad_norm": 1.607433557510376,
-      "learning_rate": 2.595055413469736e-06,
-      "loss": 0.1755,
-      "step": 4343
-    },
-    {
-      "epoch": 0.7406452300396942,
-      "grad_norm": 0.6860228776931763,
-      "learning_rate": 2.5933503836317135e-06,
-      "loss": 0.0924,
-      "step": 4344
-    },
-    {
-      "epoch": 0.7408157284812319,
-      "grad_norm": 1.0273911952972412,
-      "learning_rate": 2.5916453537936914e-06,
-      "loss": 0.1122,
-      "step": 4345
-    },
-    {
-      "epoch": 0.7409862269227695,
-      "grad_norm": 1.5581039190292358,
-      "learning_rate": 2.5899403239556694e-06,
-      "loss": 0.1814,
-      "step": 4346
-    },
-    {
-      "epoch": 0.7411567253643072,
-      "grad_norm": 1.4215240478515625,
-      "learning_rate": 2.5882352941176473e-06,
-      "loss": 0.089,
-      "step": 4347
-    },
-    {
-      "epoch": 0.7413272238058449,
-      "grad_norm": 1.4152482748031616,
-      "learning_rate": 2.5865302642796253e-06,
-      "loss": 0.1896,
-      "step": 4348
-    },
-    {
-      "epoch": 0.7414977222473825,
-      "grad_norm": 1.5308514833450317,
-      "learning_rate": 2.5848252344416032e-06,
-      "loss": 0.1202,
-      "step": 4349
-    },
-    {
-      "epoch": 0.7416682206889202,
-      "grad_norm": 0.9181138277053833,
-      "learning_rate": 2.5831202046035808e-06,
-      "loss": 0.1222,
-      "step": 4350
-    },
-    {
-      "epoch": 0.741838719130458,
-      "grad_norm": 1.5006439685821533,
-      "learning_rate": 2.5814151747655587e-06,
-      "loss": 0.1436,
-      "step": 4351
-    },
-    {
-      "epoch": 0.7420092175719957,
-      "grad_norm": 1.5951142311096191,
-      "learning_rate": 2.5797101449275362e-06,
-      "loss": 0.1597,
-      "step": 4352
-    },
-    {
-      "epoch": 0.7421797160135333,
-      "grad_norm": 1.2084048986434937,
-      "learning_rate": 2.578005115089514e-06,
-      "loss": 0.1351,
-      "step": 4353
-    },
-    {
-      "epoch": 0.742350214455071,
-      "grad_norm": 1.3123643398284912,
-      "learning_rate": 2.5763000852514917e-06,
-      "loss": 0.1916,
-      "step": 4354
-    },
-    {
-      "epoch": 0.7425207128966087,
-      "grad_norm": 1.2805266380310059,
-      "learning_rate": 2.57459505541347e-06,
-      "loss": 0.1025,
-      "step": 4355
-    },
-    {
-      "epoch": 0.7426912113381464,
-      "grad_norm": 0.7374964356422424,
-      "learning_rate": 2.572890025575448e-06,
-      "loss": 0.0495,
-      "step": 4356
-    },
-    {
-      "epoch": 0.742861709779684,
-      "grad_norm": 1.0054583549499512,
-      "learning_rate": 2.5711849957374255e-06,
-      "loss": 0.0919,
-      "step": 4357
-    },
-    {
-      "epoch": 0.7430322082212217,
-      "grad_norm": 1.0947014093399048,
-      "learning_rate": 2.5694799658994035e-06,
-      "loss": 0.1137,
-      "step": 4358
-    },
-    {
-      "epoch": 0.7432027066627594,
-      "grad_norm": 1.2278332710266113,
-      "learning_rate": 2.5677749360613814e-06,
-      "loss": 0.1139,
-      "step": 4359
-    },
-    {
-      "epoch": 0.743373205104297,
-      "grad_norm": 1.3202364444732666,
-      "learning_rate": 2.566069906223359e-06,
-      "loss": 0.1668,
-      "step": 4360
-    },
-    {
-      "epoch": 0.7435437035458348,
-      "grad_norm": 1.5769107341766357,
-      "learning_rate": 2.564364876385337e-06,
-      "loss": 0.1187,
-      "step": 4361
-    },
-    {
-      "epoch": 0.7437142019873725,
-      "grad_norm": 1.1885755062103271,
-      "learning_rate": 2.5626598465473144e-06,
-      "loss": 0.129,
-      "step": 4362
-    },
-    {
-      "epoch": 0.7438847004289102,
-      "grad_norm": 1.2387917041778564,
-      "learning_rate": 2.5609548167092924e-06,
-      "loss": 0.1576,
-      "step": 4363
-    },
-    {
-      "epoch": 0.7440551988704478,
-      "grad_norm": 1.4065356254577637,
-      "learning_rate": 2.5592497868712708e-06,
-      "loss": 0.1399,
-      "step": 4364
-    },
-    {
-      "epoch": 0.7442256973119855,
-      "grad_norm": 1.1759954690933228,
-      "learning_rate": 2.5575447570332483e-06,
-      "loss": 0.1346,
-      "step": 4365
-    },
-    {
-      "epoch": 0.7443961957535232,
-      "grad_norm": 1.0152462720870972,
-      "learning_rate": 2.5558397271952262e-06,
-      "loss": 0.1211,
-      "step": 4366
-    },
-    {
-      "epoch": 0.7445666941950608,
-      "grad_norm": 1.4754053354263306,
-      "learning_rate": 2.554134697357204e-06,
-      "loss": 0.0932,
-      "step": 4367
-    },
-    {
-      "epoch": 0.7447371926365985,
-      "grad_norm": 1.8604658842086792,
-      "learning_rate": 2.5524296675191817e-06,
-      "loss": 0.1788,
-      "step": 4368
-    },
-    {
-      "epoch": 0.7449076910781363,
-      "grad_norm": 1.0769208669662476,
-      "learning_rate": 2.5507246376811596e-06,
-      "loss": 0.1241,
-      "step": 4369
-    },
-    {
-      "epoch": 0.745078189519674,
-      "grad_norm": 2.2280867099761963,
-      "learning_rate": 2.549019607843137e-06,
-      "loss": 0.2057,
-      "step": 4370
-    },
-    {
-      "epoch": 0.7452486879612116,
-      "grad_norm": 1.528975248336792,
-      "learning_rate": 2.547314578005115e-06,
-      "loss": 0.1394,
-      "step": 4371
-    },
-    {
-      "epoch": 0.7454191864027493,
-      "grad_norm": 1.2800006866455078,
-      "learning_rate": 2.5456095481670935e-06,
-      "loss": 0.1064,
-      "step": 4372
-    },
-    {
-      "epoch": 0.745589684844287,
-      "grad_norm": 0.9087845683097839,
-      "learning_rate": 2.543904518329071e-06,
-      "loss": 0.0957,
-      "step": 4373
-    },
-    {
-      "epoch": 0.7457601832858246,
-      "grad_norm": 1.1006022691726685,
-      "learning_rate": 2.542199488491049e-06,
-      "loss": 0.1061,
-      "step": 4374
-    },
-    {
-      "epoch": 0.7459306817273623,
-      "grad_norm": 1.150396466255188,
-      "learning_rate": 2.5404944586530265e-06,
-      "loss": 0.1074,
-      "step": 4375
-    },
-    {
-      "epoch": 0.7461011801689,
-      "grad_norm": 0.9069077968597412,
-      "learning_rate": 2.5387894288150044e-06,
-      "loss": 0.0905,
-      "step": 4376
-    },
-    {
-      "epoch": 0.7462716786104377,
-      "grad_norm": 1.106367588043213,
-      "learning_rate": 2.5370843989769824e-06,
-      "loss": 0.1537,
-      "step": 4377
-    },
-    {
-      "epoch": 0.7464421770519754,
-      "grad_norm": 1.9207699298858643,
-      "learning_rate": 2.53537936913896e-06,
-      "loss": 0.2223,
-      "step": 4378
-    },
-    {
-      "epoch": 0.7466126754935131,
-      "grad_norm": 0.9847105145454407,
-      "learning_rate": 2.533674339300938e-06,
-      "loss": 0.0824,
-      "step": 4379
-    },
-    {
-      "epoch": 0.7467831739350508,
-      "grad_norm": 1.7170462608337402,
-      "learning_rate": 2.5319693094629162e-06,
-      "loss": 0.1208,
-      "step": 4380
-    },
-    {
-      "epoch": 0.7469536723765884,
-      "grad_norm": 1.0074905157089233,
-      "learning_rate": 2.5302642796248938e-06,
-      "loss": 0.1053,
-      "step": 4381
-    },
-    {
-      "epoch": 0.7471241708181261,
-      "grad_norm": 2.5140938758850098,
-      "learning_rate": 2.5285592497868717e-06,
-      "loss": 0.2067,
-      "step": 4382
-    },
-    {
-      "epoch": 0.7472946692596638,
-      "grad_norm": 1.0962971448898315,
-      "learning_rate": 2.5268542199488492e-06,
-      "loss": 0.1203,
-      "step": 4383
-    },
-    {
-      "epoch": 0.7474651677012015,
-      "grad_norm": 1.105176568031311,
-      "learning_rate": 2.525149190110827e-06,
-      "loss": 0.0925,
-      "step": 4384
-    },
-    {
-      "epoch": 0.7476356661427391,
-      "grad_norm": 1.3398061990737915,
-      "learning_rate": 2.523444160272805e-06,
-      "loss": 0.1793,
-      "step": 4385
-    },
-    {
-      "epoch": 0.7478061645842768,
-      "grad_norm": 1.1867016553878784,
-      "learning_rate": 2.5217391304347826e-06,
-      "loss": 0.1557,
-      "step": 4386
-    },
-    {
-      "epoch": 0.7479766630258146,
-      "grad_norm": 1.3503540754318237,
-      "learning_rate": 2.5200341005967606e-06,
-      "loss": 0.1372,
-      "step": 4387
-    },
-    {
-      "epoch": 0.7481471614673522,
-      "grad_norm": 1.147642970085144,
-      "learning_rate": 2.518329070758738e-06,
-      "loss": 0.1113,
-      "step": 4388
-    },
-    {
-      "epoch": 0.7483176599088899,
-      "grad_norm": 1.246585488319397,
-      "learning_rate": 2.5166240409207165e-06,
-      "loss": 0.1425,
-      "step": 4389
-    },
-    {
-      "epoch": 0.7484881583504276,
-      "grad_norm": 1.5160149335861206,
-      "learning_rate": 2.5149190110826944e-06,
-      "loss": 0.07,
-      "step": 4390
-    },
-    {
-      "epoch": 0.7486586567919653,
-      "grad_norm": 1.313860535621643,
-      "learning_rate": 2.513213981244672e-06,
-      "loss": 0.1085,
-      "step": 4391
-    },
-    {
-      "epoch": 0.7488291552335029,
-      "grad_norm": 0.953596830368042,
-      "learning_rate": 2.51150895140665e-06,
-      "loss": 0.0741,
-      "step": 4392
-    },
-    {
-      "epoch": 0.7489996536750406,
-      "grad_norm": 0.7907807230949402,
-      "learning_rate": 2.5098039215686274e-06,
-      "loss": 0.0392,
-      "step": 4393
-    },
-    {
-      "epoch": 0.7491701521165783,
-      "grad_norm": 1.9617118835449219,
-      "learning_rate": 2.5080988917306054e-06,
-      "loss": 0.2056,
-      "step": 4394
-    },
-    {
-      "epoch": 0.7493406505581159,
-      "grad_norm": 0.8299599885940552,
-      "learning_rate": 2.5063938618925833e-06,
-      "loss": 0.0701,
-      "step": 4395
-    },
-    {
-      "epoch": 0.7495111489996537,
-      "grad_norm": 1.3734976053237915,
-      "learning_rate": 2.504688832054561e-06,
-      "loss": 0.146,
-      "step": 4396
-    },
-    {
-      "epoch": 0.7496816474411914,
-      "grad_norm": 1.1142525672912598,
-      "learning_rate": 2.5029838022165392e-06,
-      "loss": 0.1469,
-      "step": 4397
-    },
-    {
-      "epoch": 0.7498521458827291,
-      "grad_norm": 1.0340555906295776,
-      "learning_rate": 2.501278772378517e-06,
-      "loss": 0.109,
-      "step": 4398
-    },
-    {
-      "epoch": 0.7500226443242667,
-      "grad_norm": 1.8724783658981323,
-      "learning_rate": 2.4995737425404947e-06,
-      "loss": 0.2038,
-      "step": 4399
-    },
-    {
-      "epoch": 0.7501931427658044,
-      "grad_norm": 1.4347814321517944,
-      "learning_rate": 2.4978687127024726e-06,
-      "loss": 0.1388,
-      "step": 4400
-    },
-    {
-      "epoch": 0.7501931427658044,
-      "eval_f1_score": 0.4258373205741627,
-      "eval_loss": 0.13514938950538635,
-      "eval_runtime": 183.3576,
-      "eval_samples_per_second": 54.538,
-      "eval_steps_per_second": 3.409,
-      "step": 4400
-    },
-    {
-      "epoch": 0.7503636412073421,
-      "grad_norm": 2.3867857456207275,
-      "learning_rate": 2.49616368286445e-06,
-      "loss": 0.1136,
-      "step": 4401
-    },
-    {
-      "epoch": 0.7505341396488797,
-      "grad_norm": 1.4532997608184814,
-      "learning_rate": 2.494458653026428e-06,
-      "loss": 0.132,
-      "step": 4402
-    },
-    {
-      "epoch": 0.7507046380904174,
-      "grad_norm": 1.3240541219711304,
-      "learning_rate": 2.492753623188406e-06,
-      "loss": 0.1223,
-      "step": 4403
-    },
-    {
-      "epoch": 0.7508751365319551,
-      "grad_norm": 1.0867290496826172,
-      "learning_rate": 2.491048593350384e-06,
-      "loss": 0.1179,
-      "step": 4404
-    },
-    {
-      "epoch": 0.7510456349734929,
-      "grad_norm": 1.9877395629882812,
-      "learning_rate": 2.4893435635123615e-06,
-      "loss": 0.1358,
-      "step": 4405
-    },
-    {
-      "epoch": 0.7512161334150305,
-      "grad_norm": 0.991084098815918,
-      "learning_rate": 2.4876385336743395e-06,
-      "loss": 0.0833,
-      "step": 4406
-    },
-    {
-      "epoch": 0.7513866318565682,
-      "grad_norm": 1.5853980779647827,
-      "learning_rate": 2.485933503836317e-06,
-      "loss": 0.1493,
-      "step": 4407
-    },
-    {
-      "epoch": 0.7515571302981059,
-      "grad_norm": 1.4868155717849731,
-      "learning_rate": 2.4842284739982954e-06,
-      "loss": 0.167,
-      "step": 4408
-    },
-    {
-      "epoch": 0.7517276287396436,
-      "grad_norm": 1.1008561849594116,
-      "learning_rate": 2.482523444160273e-06,
-      "loss": 0.1315,
-      "step": 4409
-    },
-    {
-      "epoch": 0.7518981271811812,
-      "grad_norm": 1.977602243423462,
-      "learning_rate": 2.480818414322251e-06,
-      "loss": 0.2289,
-      "step": 4410
-    },
-    {
-      "epoch": 0.7520686256227189,
-      "grad_norm": 1.166752576828003,
-      "learning_rate": 2.4791133844842284e-06,
-      "loss": 0.0632,
-      "step": 4411
-    },
-    {
-      "epoch": 0.7522391240642566,
-      "grad_norm": 0.9453310370445251,
-      "learning_rate": 2.4774083546462067e-06,
-      "loss": 0.0958,
-      "step": 4412
-    },
-    {
-      "epoch": 0.7524096225057942,
-      "grad_norm": 1.2444056272506714,
-      "learning_rate": 2.4757033248081843e-06,
-      "loss": 0.1525,
-      "step": 4413
-    },
-    {
-      "epoch": 0.752580120947332,
-      "grad_norm": 1.092913031578064,
-      "learning_rate": 2.4739982949701622e-06,
-      "loss": 0.0853,
-      "step": 4414
-    },
-    {
-      "epoch": 0.7527506193888697,
-      "grad_norm": 0.9639697074890137,
-      "learning_rate": 2.4722932651321397e-06,
-      "loss": 0.1036,
-      "step": 4415
-    },
-    {
-      "epoch": 0.7529211178304074,
-      "grad_norm": 1.2381495237350464,
-      "learning_rate": 2.470588235294118e-06,
-      "loss": 0.0988,
-      "step": 4416
-    },
-    {
-      "epoch": 0.753091616271945,
-      "grad_norm": 1.0427268743515015,
-      "learning_rate": 2.4688832054560956e-06,
-      "loss": 0.0812,
-      "step": 4417
-    },
-    {
-      "epoch": 0.7532621147134827,
-      "grad_norm": 0.9707403182983398,
-      "learning_rate": 2.4671781756180736e-06,
-      "loss": 0.0893,
-      "step": 4418
-    },
-    {
-      "epoch": 0.7534326131550204,
-      "grad_norm": 0.9385908842086792,
-      "learning_rate": 2.465473145780051e-06,
-      "loss": 0.0769,
-      "step": 4419
-    },
-    {
-      "epoch": 0.753603111596558,
-      "grad_norm": 1.4962774515151978,
-      "learning_rate": 2.4637681159420295e-06,
-      "loss": 0.1397,
-      "step": 4420
-    },
-    {
-      "epoch": 0.7537736100380957,
-      "grad_norm": 0.847218930721283,
-      "learning_rate": 2.462063086104007e-06,
-      "loss": 0.0686,
-      "step": 4421
-    },
-    {
-      "epoch": 0.7539441084796334,
-      "grad_norm": 0.836495041847229,
-      "learning_rate": 2.460358056265985e-06,
-      "loss": 0.0678,
-      "step": 4422
-    },
-    {
-      "epoch": 0.7541146069211712,
-      "grad_norm": 0.7686975002288818,
-      "learning_rate": 2.4586530264279625e-06,
-      "loss": 0.0581,
-      "step": 4423
-    },
-    {
-      "epoch": 0.7542851053627088,
-      "grad_norm": 1.2687360048294067,
-      "learning_rate": 2.456947996589941e-06,
-      "loss": 0.1058,
-      "step": 4424
-    },
-    {
-      "epoch": 0.7544556038042465,
-      "grad_norm": 0.7973870635032654,
-      "learning_rate": 2.4552429667519184e-06,
-      "loss": 0.058,
-      "step": 4425
-    },
-    {
-      "epoch": 0.7546261022457842,
-      "grad_norm": 1.092147707939148,
-      "learning_rate": 2.4535379369138963e-06,
-      "loss": 0.0922,
-      "step": 4426
-    },
-    {
-      "epoch": 0.7547966006873218,
-      "grad_norm": 1.2011126279830933,
-      "learning_rate": 2.451832907075874e-06,
-      "loss": 0.0785,
-      "step": 4427
-    },
-    {
-      "epoch": 0.7549670991288595,
-      "grad_norm": 1.0751177072525024,
-      "learning_rate": 2.4501278772378518e-06,
-      "loss": 0.1342,
-      "step": 4428
-    },
-    {
-      "epoch": 0.7551375975703972,
-      "grad_norm": 1.8159641027450562,
-      "learning_rate": 2.4484228473998297e-06,
-      "loss": 0.129,
-      "step": 4429
-    },
-    {
-      "epoch": 0.7553080960119349,
-      "grad_norm": 0.9125354290008545,
-      "learning_rate": 2.4467178175618077e-06,
-      "loss": 0.1658,
-      "step": 4430
-    },
-    {
-      "epoch": 0.7554785944534725,
-      "grad_norm": 1.7383756637573242,
-      "learning_rate": 2.445012787723785e-06,
-      "loss": 0.1719,
-      "step": 4431
-    },
-    {
-      "epoch": 0.7556490928950103,
-      "grad_norm": 1.2764480113983154,
-      "learning_rate": 2.443307757885763e-06,
-      "loss": 0.0987,
-      "step": 4432
-    },
-    {
-      "epoch": 0.755819591336548,
-      "grad_norm": 1.415343165397644,
-      "learning_rate": 2.441602728047741e-06,
-      "loss": 0.0729,
-      "step": 4433
-    },
-    {
-      "epoch": 0.7559900897780856,
-      "grad_norm": 1.5825785398483276,
-      "learning_rate": 2.439897698209719e-06,
-      "loss": 0.1534,
-      "step": 4434
-    },
-    {
-      "epoch": 0.7561605882196233,
-      "grad_norm": 1.9668054580688477,
-      "learning_rate": 2.4381926683716966e-06,
-      "loss": 0.2567,
-      "step": 4435
-    },
-    {
-      "epoch": 0.756331086661161,
-      "grad_norm": 1.0087342262268066,
-      "learning_rate": 2.4364876385336745e-06,
-      "loss": 0.0754,
-      "step": 4436
-    },
-    {
-      "epoch": 0.7565015851026987,
-      "grad_norm": 1.6823344230651855,
-      "learning_rate": 2.4347826086956525e-06,
-      "loss": 0.0898,
-      "step": 4437
-    },
-    {
-      "epoch": 0.7566720835442363,
-      "grad_norm": 1.1069707870483398,
-      "learning_rate": 2.4330775788576304e-06,
-      "loss": 0.1075,
-      "step": 4438
-    },
-    {
-      "epoch": 0.756842581985774,
-      "grad_norm": 1.4310376644134521,
-      "learning_rate": 2.431372549019608e-06,
-      "loss": 0.1312,
-      "step": 4439
-    },
-    {
-      "epoch": 0.7570130804273117,
-      "grad_norm": 1.1864593029022217,
-      "learning_rate": 2.429667519181586e-06,
-      "loss": 0.1176,
-      "step": 4440
-    },
-    {
-      "epoch": 0.7571835788688493,
-      "grad_norm": 0.9127157330513,
-      "learning_rate": 2.427962489343564e-06,
-      "loss": 0.049,
-      "step": 4441
-    },
-    {
-      "epoch": 0.7573540773103871,
-      "grad_norm": 1.7356191873550415,
-      "learning_rate": 2.4262574595055418e-06,
-      "loss": 0.1689,
-      "step": 4442
-    },
-    {
-      "epoch": 0.7575245757519248,
-      "grad_norm": 1.5915813446044922,
-      "learning_rate": 2.4245524296675193e-06,
-      "loss": 0.1468,
-      "step": 4443
-    },
-    {
-      "epoch": 0.7576950741934625,
-      "grad_norm": 1.4367433786392212,
-      "learning_rate": 2.4228473998294973e-06,
-      "loss": 0.1447,
-      "step": 4444
-    },
-    {
-      "epoch": 0.7578655726350001,
-      "grad_norm": 1.4117807149887085,
-      "learning_rate": 2.421142369991475e-06,
-      "loss": 0.1131,
-      "step": 4445
-    },
-    {
-      "epoch": 0.7580360710765378,
-      "grad_norm": 1.4179257154464722,
-      "learning_rate": 2.419437340153453e-06,
-      "loss": 0.1428,
-      "step": 4446
-    },
-    {
-      "epoch": 0.7582065695180755,
-      "grad_norm": 1.7990797758102417,
-      "learning_rate": 2.4177323103154307e-06,
-      "loss": 0.1489,
-      "step": 4447
-    },
-    {
-      "epoch": 0.7583770679596131,
-      "grad_norm": 0.9080357551574707,
-      "learning_rate": 2.4160272804774086e-06,
-      "loss": 0.0824,
-      "step": 4448
-    },
-    {
-      "epoch": 0.7585475664011508,
-      "grad_norm": 1.3778406381607056,
-      "learning_rate": 2.4143222506393866e-06,
-      "loss": 0.1252,
-      "step": 4449
-    },
-    {
-      "epoch": 0.7587180648426886,
-      "grad_norm": 1.2085192203521729,
-      "learning_rate": 2.412617220801364e-06,
-      "loss": 0.1522,
-      "step": 4450
-    },
-    {
-      "epoch": 0.7588885632842263,
-      "grad_norm": 0.9220867156982422,
-      "learning_rate": 2.410912190963342e-06,
-      "loss": 0.0691,
-      "step": 4451
-    },
-    {
-      "epoch": 0.7590590617257639,
-      "grad_norm": 1.1078466176986694,
-      "learning_rate": 2.40920716112532e-06,
-      "loss": 0.0868,
-      "step": 4452
-    },
-    {
-      "epoch": 0.7592295601673016,
-      "grad_norm": 1.5541517734527588,
-      "learning_rate": 2.4075021312872975e-06,
-      "loss": 0.153,
-      "step": 4453
-    },
-    {
-      "epoch": 0.7594000586088393,
-      "grad_norm": 1.4000375270843506,
-      "learning_rate": 2.4057971014492755e-06,
-      "loss": 0.1326,
-      "step": 4454
-    },
-    {
-      "epoch": 0.759570557050377,
-      "grad_norm": 1.2640565633773804,
-      "learning_rate": 2.4040920716112534e-06,
-      "loss": 0.0853,
-      "step": 4455
-    },
-    {
-      "epoch": 0.7597410554919146,
-      "grad_norm": 0.8130654692649841,
-      "learning_rate": 2.4023870417732314e-06,
-      "loss": 0.0956,
-      "step": 4456
-    },
-    {
-      "epoch": 0.7599115539334523,
-      "grad_norm": 1.007473111152649,
-      "learning_rate": 2.400682011935209e-06,
-      "loss": 0.0567,
-      "step": 4457
-    },
-    {
-      "epoch": 0.76008205237499,
-      "grad_norm": 0.8968228697776794,
-      "learning_rate": 2.398976982097187e-06,
-      "loss": 0.0818,
-      "step": 4458
-    },
-    {
-      "epoch": 0.7602525508165277,
-      "grad_norm": 1.0275651216506958,
-      "learning_rate": 2.3972719522591648e-06,
-      "loss": 0.1403,
-      "step": 4459
-    },
-    {
-      "epoch": 0.7604230492580654,
-      "grad_norm": 1.5832469463348389,
-      "learning_rate": 2.3955669224211427e-06,
-      "loss": 0.1343,
-      "step": 4460
-    },
-    {
-      "epoch": 0.7605935476996031,
-      "grad_norm": 0.951404869556427,
-      "learning_rate": 2.3938618925831202e-06,
-      "loss": 0.0802,
-      "step": 4461
-    },
-    {
-      "epoch": 0.7607640461411408,
-      "grad_norm": 0.8085765242576599,
-      "learning_rate": 2.392156862745098e-06,
-      "loss": 0.0801,
-      "step": 4462
-    },
-    {
-      "epoch": 0.7609345445826784,
-      "grad_norm": 1.6486432552337646,
-      "learning_rate": 2.390451832907076e-06,
-      "loss": 0.1259,
-      "step": 4463
-    },
-    {
-      "epoch": 0.7611050430242161,
-      "grad_norm": 1.229943871498108,
-      "learning_rate": 2.3887468030690537e-06,
-      "loss": 0.0928,
-      "step": 4464
-    },
-    {
-      "epoch": 0.7612755414657538,
-      "grad_norm": 1.0928643941879272,
-      "learning_rate": 2.3870417732310316e-06,
-      "loss": 0.1039,
-      "step": 4465
-    },
-    {
-      "epoch": 0.7614460399072914,
-      "grad_norm": 1.9494421482086182,
-      "learning_rate": 2.3853367433930096e-06,
-      "loss": 0.1601,
-      "step": 4466
-    },
-    {
-      "epoch": 0.7616165383488291,
-      "grad_norm": 1.933903694152832,
-      "learning_rate": 2.3836317135549875e-06,
-      "loss": 0.1642,
-      "step": 4467
-    },
-    {
-      "epoch": 0.7617870367903669,
-      "grad_norm": 1.6810734272003174,
-      "learning_rate": 2.381926683716965e-06,
-      "loss": 0.186,
-      "step": 4468
-    },
-    {
-      "epoch": 0.7619575352319046,
-      "grad_norm": 0.9030094742774963,
-      "learning_rate": 2.380221653878943e-06,
-      "loss": 0.0896,
-      "step": 4469
-    },
-    {
-      "epoch": 0.7621280336734422,
-      "grad_norm": 1.5973889827728271,
-      "learning_rate": 2.378516624040921e-06,
-      "loss": 0.1563,
-      "step": 4470
-    },
-    {
-      "epoch": 0.7622985321149799,
-      "grad_norm": 1.6514700651168823,
-      "learning_rate": 2.376811594202899e-06,
-      "loss": 0.1492,
-      "step": 4471
-    },
-    {
-      "epoch": 0.7624690305565176,
-      "grad_norm": 1.5564332008361816,
-      "learning_rate": 2.3751065643648764e-06,
-      "loss": 0.1419,
-      "step": 4472
-    },
-    {
-      "epoch": 0.7626395289980552,
-      "grad_norm": 1.4853767156600952,
-      "learning_rate": 2.3734015345268544e-06,
-      "loss": 0.1656,
-      "step": 4473
-    },
-    {
-      "epoch": 0.7628100274395929,
-      "grad_norm": 1.4599570035934448,
-      "learning_rate": 2.3716965046888323e-06,
-      "loss": 0.1308,
-      "step": 4474
-    },
-    {
-      "epoch": 0.7629805258811306,
-      "grad_norm": 1.1921581029891968,
-      "learning_rate": 2.3699914748508102e-06,
-      "loss": 0.1128,
-      "step": 4475
-    },
-    {
-      "epoch": 0.7631510243226683,
-      "grad_norm": 1.045434832572937,
-      "learning_rate": 2.3682864450127878e-06,
-      "loss": 0.0534,
-      "step": 4476
-    },
-    {
-      "epoch": 0.763321522764206,
-      "grad_norm": 1.5900884866714478,
-      "learning_rate": 2.3665814151747657e-06,
-      "loss": 0.1463,
-      "step": 4477
-    },
-    {
-      "epoch": 0.7634920212057437,
-      "grad_norm": 1.286139965057373,
-      "learning_rate": 2.3648763853367437e-06,
-      "loss": 0.1475,
-      "step": 4478
-    },
-    {
-      "epoch": 0.7636625196472814,
-      "grad_norm": 1.091168761253357,
-      "learning_rate": 2.3631713554987216e-06,
-      "loss": 0.1028,
-      "step": 4479
-    },
-    {
-      "epoch": 0.763833018088819,
-      "grad_norm": 1.0828571319580078,
-      "learning_rate": 2.361466325660699e-06,
-      "loss": 0.1278,
-      "step": 4480
-    },
-    {
-      "epoch": 0.7640035165303567,
-      "grad_norm": 2.2095322608947754,
-      "learning_rate": 2.359761295822677e-06,
-      "loss": 0.2602,
-      "step": 4481
-    },
-    {
-      "epoch": 0.7641740149718944,
-      "grad_norm": 1.4665558338165283,
-      "learning_rate": 2.3580562659846546e-06,
-      "loss": 0.1733,
-      "step": 4482
-    },
-    {
-      "epoch": 0.7643445134134321,
-      "grad_norm": 1.0846915245056152,
-      "learning_rate": 2.356351236146633e-06,
-      "loss": 0.1111,
-      "step": 4483
-    },
-    {
-      "epoch": 0.7645150118549697,
-      "grad_norm": 1.8451837301254272,
-      "learning_rate": 2.3546462063086105e-06,
-      "loss": 0.1506,
-      "step": 4484
-    },
-    {
-      "epoch": 0.7646855102965074,
-      "grad_norm": 1.1041662693023682,
-      "learning_rate": 2.3529411764705885e-06,
-      "loss": 0.1269,
-      "step": 4485
-    },
-    {
-      "epoch": 0.7648560087380452,
-      "grad_norm": 1.6395684480667114,
-      "learning_rate": 2.351236146632566e-06,
-      "loss": 0.1488,
-      "step": 4486
-    },
-    {
-      "epoch": 0.7650265071795828,
-      "grad_norm": 1.3365883827209473,
-      "learning_rate": 2.3495311167945443e-06,
-      "loss": 0.1094,
-      "step": 4487
-    },
-    {
-      "epoch": 0.7651970056211205,
-      "grad_norm": 1.0546587705612183,
-      "learning_rate": 2.347826086956522e-06,
-      "loss": 0.1041,
-      "step": 4488
-    },
-    {
-      "epoch": 0.7653675040626582,
-      "grad_norm": 2.1003003120422363,
-      "learning_rate": 2.3461210571185e-06,
-      "loss": 0.2355,
-      "step": 4489
-    },
-    {
-      "epoch": 0.7655380025041959,
-      "grad_norm": 1.71974778175354,
-      "learning_rate": 2.3444160272804773e-06,
-      "loss": 0.1549,
-      "step": 4490
-    },
-    {
-      "epoch": 0.7657085009457335,
-      "grad_norm": 1.3704986572265625,
-      "learning_rate": 2.3427109974424557e-06,
-      "loss": 0.0997,
-      "step": 4491
-    },
-    {
-      "epoch": 0.7658789993872712,
-      "grad_norm": 0.7516347765922546,
-      "learning_rate": 2.3410059676044332e-06,
-      "loss": 0.0518,
-      "step": 4492
-    },
-    {
-      "epoch": 0.7660494978288089,
-      "grad_norm": 1.6383329629898071,
-      "learning_rate": 2.339300937766411e-06,
-      "loss": 0.096,
-      "step": 4493
-    },
-    {
-      "epoch": 0.7662199962703465,
-      "grad_norm": 1.824157953262329,
-      "learning_rate": 2.3375959079283887e-06,
-      "loss": 0.1887,
-      "step": 4494
-    },
-    {
-      "epoch": 0.7663904947118843,
-      "grad_norm": 1.9850565195083618,
-      "learning_rate": 2.335890878090367e-06,
-      "loss": 0.1113,
-      "step": 4495
-    },
-    {
-      "epoch": 0.766560993153422,
-      "grad_norm": 1.7400462627410889,
-      "learning_rate": 2.3341858482523446e-06,
-      "loss": 0.1139,
-      "step": 4496
-    },
-    {
-      "epoch": 0.7667314915949597,
-      "grad_norm": 0.9568600654602051,
-      "learning_rate": 2.3324808184143226e-06,
-      "loss": 0.0496,
-      "step": 4497
-    },
-    {
-      "epoch": 0.7669019900364973,
-      "grad_norm": 1.088996171951294,
-      "learning_rate": 2.3307757885763e-06,
-      "loss": 0.0943,
-      "step": 4498
-    },
-    {
-      "epoch": 0.767072488478035,
-      "grad_norm": 1.688697338104248,
-      "learning_rate": 2.329070758738278e-06,
-      "loss": 0.1739,
-      "step": 4499
-    },
-    {
-      "epoch": 0.7672429869195727,
-      "grad_norm": 0.9527409076690674,
-      "learning_rate": 2.327365728900256e-06,
-      "loss": 0.0959,
-      "step": 4500
-    },
-    {
-      "epoch": 0.7672429869195727,
-      "eval_f1_score": 0.4222737819025522,
-      "eval_loss": 0.1333077847957611,
-      "eval_runtime": 183.4295,
-      "eval_samples_per_second": 54.517,
-      "eval_steps_per_second": 3.407,
-      "step": 4500
-    },
-    {
-      "epoch": 0.7674134853611104,
-      "grad_norm": 2.166240692138672,
-      "learning_rate": 2.325660699062234e-06,
-      "loss": 0.1514,
-      "step": 4501
-    },
-    {
-      "epoch": 0.767583983802648,
-      "grad_norm": 2.316793441772461,
-      "learning_rate": 2.3239556692242114e-06,
-      "loss": 0.1376,
-      "step": 4502
-    },
-    {
-      "epoch": 0.7677544822441857,
-      "grad_norm": 1.5623490810394287,
-      "learning_rate": 2.3222506393861894e-06,
-      "loss": 0.1548,
-      "step": 4503
-    },
-    {
-      "epoch": 0.7679249806857235,
-      "grad_norm": 1.0329327583312988,
-      "learning_rate": 2.3205456095481673e-06,
-      "loss": 0.1004,
-      "step": 4504
-    },
-    {
-      "epoch": 0.7680954791272611,
-      "grad_norm": 1.3712180852890015,
-      "learning_rate": 2.3188405797101453e-06,
-      "loss": 0.1254,
-      "step": 4505
-    },
-    {
-      "epoch": 0.7682659775687988,
-      "grad_norm": 1.363422155380249,
-      "learning_rate": 2.317135549872123e-06,
-      "loss": 0.1472,
-      "step": 4506
-    },
-    {
-      "epoch": 0.7684364760103365,
-      "grad_norm": 1.4349713325500488,
-      "learning_rate": 2.3154305200341008e-06,
-      "loss": 0.1377,
-      "step": 4507
-    },
-    {
-      "epoch": 0.7686069744518742,
-      "grad_norm": 1.409959316253662,
-      "learning_rate": 2.3137254901960787e-06,
-      "loss": 0.096,
-      "step": 4508
-    },
-    {
-      "epoch": 0.7687774728934118,
-      "grad_norm": 1.6371657848358154,
-      "learning_rate": 2.3120204603580567e-06,
-      "loss": 0.1444,
-      "step": 4509
-    },
-    {
-      "epoch": 0.7689479713349495,
-      "grad_norm": 1.1964426040649414,
-      "learning_rate": 2.310315430520034e-06,
-      "loss": 0.0879,
-      "step": 4510
-    },
-    {
-      "epoch": 0.7691184697764872,
-      "grad_norm": 2.026524782180786,
-      "learning_rate": 2.308610400682012e-06,
-      "loss": 0.1026,
-      "step": 4511
-    },
-    {
-      "epoch": 0.7692889682180248,
-      "grad_norm": 1.329511284828186,
-      "learning_rate": 2.30690537084399e-06,
-      "loss": 0.1055,
-      "step": 4512
-    },
-    {
-      "epoch": 0.7694594666595626,
-      "grad_norm": 1.415218710899353,
-      "learning_rate": 2.305200341005968e-06,
-      "loss": 0.14,
-      "step": 4513
-    },
-    {
-      "epoch": 0.7696299651011003,
-      "grad_norm": 1.076926827430725,
-      "learning_rate": 2.3034953111679455e-06,
-      "loss": 0.1197,
-      "step": 4514
-    },
-    {
-      "epoch": 0.769800463542638,
-      "grad_norm": 0.8736515045166016,
-      "learning_rate": 2.3017902813299235e-06,
-      "loss": 0.0643,
-      "step": 4515
-    },
-    {
-      "epoch": 0.7699709619841756,
-      "grad_norm": 0.9680917859077454,
-      "learning_rate": 2.3000852514919014e-06,
-      "loss": 0.0981,
-      "step": 4516
-    },
-    {
-      "epoch": 0.7701414604257133,
-      "grad_norm": 1.3549295663833618,
-      "learning_rate": 2.2983802216538794e-06,
-      "loss": 0.1523,
-      "step": 4517
-    },
-    {
-      "epoch": 0.770311958867251,
-      "grad_norm": 1.550164818763733,
-      "learning_rate": 2.296675191815857e-06,
-      "loss": 0.1469,
-      "step": 4518
-    },
-    {
-      "epoch": 0.7704824573087886,
-      "grad_norm": 0.9949040412902832,
-      "learning_rate": 2.294970161977835e-06,
-      "loss": 0.0529,
-      "step": 4519
-    },
-    {
-      "epoch": 0.7706529557503263,
-      "grad_norm": 1.1915141344070435,
-      "learning_rate": 2.2932651321398124e-06,
-      "loss": 0.1141,
-      "step": 4520
-    },
-    {
-      "epoch": 0.770823454191864,
-      "grad_norm": 1.5403510332107544,
-      "learning_rate": 2.2915601023017903e-06,
-      "loss": 0.1201,
-      "step": 4521
-    },
-    {
-      "epoch": 0.7709939526334018,
-      "grad_norm": 0.8535776138305664,
-      "learning_rate": 2.2898550724637683e-06,
-      "loss": 0.0777,
-      "step": 4522
-    },
-    {
-      "epoch": 0.7711644510749394,
-      "grad_norm": 1.0403300523757935,
-      "learning_rate": 2.2881500426257462e-06,
-      "loss": 0.0568,
-      "step": 4523
-    },
-    {
-      "epoch": 0.7713349495164771,
-      "grad_norm": 1.436591625213623,
-      "learning_rate": 2.2864450127877238e-06,
-      "loss": 0.1344,
-      "step": 4524
-    },
-    {
-      "epoch": 0.7715054479580148,
-      "grad_norm": 1.678331732749939,
-      "learning_rate": 2.2847399829497017e-06,
-      "loss": 0.1449,
-      "step": 4525
-    },
-    {
-      "epoch": 0.7716759463995524,
-      "grad_norm": 1.3845634460449219,
-      "learning_rate": 2.2830349531116796e-06,
-      "loss": 0.1028,
-      "step": 4526
-    },
-    {
-      "epoch": 0.7718464448410901,
-      "grad_norm": 1.875254511833191,
-      "learning_rate": 2.2813299232736576e-06,
-      "loss": 0.2095,
-      "step": 4527
-    },
-    {
-      "epoch": 0.7720169432826278,
-      "grad_norm": 1.1919282674789429,
-      "learning_rate": 2.279624893435635e-06,
-      "loss": 0.12,
-      "step": 4528
-    },
-    {
-      "epoch": 0.7721874417241655,
-      "grad_norm": 1.207365870475769,
-      "learning_rate": 2.277919863597613e-06,
-      "loss": 0.0699,
-      "step": 4529
-    },
-    {
-      "epoch": 0.7723579401657031,
-      "grad_norm": 1.3329031467437744,
-      "learning_rate": 2.276214833759591e-06,
-      "loss": 0.1017,
-      "step": 4530
-    },
-    {
-      "epoch": 0.7725284386072409,
-      "grad_norm": 1.2029889822006226,
-      "learning_rate": 2.274509803921569e-06,
-      "loss": 0.0918,
-      "step": 4531
-    },
-    {
-      "epoch": 0.7726989370487786,
-      "grad_norm": 1.4061421155929565,
-      "learning_rate": 2.2728047740835465e-06,
-      "loss": 0.1064,
-      "step": 4532
-    },
-    {
-      "epoch": 0.7728694354903162,
-      "grad_norm": 1.9878129959106445,
-      "learning_rate": 2.2710997442455244e-06,
-      "loss": 0.1742,
-      "step": 4533
-    },
-    {
-      "epoch": 0.7730399339318539,
-      "grad_norm": 1.2761764526367188,
-      "learning_rate": 2.2693947144075024e-06,
-      "loss": 0.0986,
-      "step": 4534
-    },
-    {
-      "epoch": 0.7732104323733916,
-      "grad_norm": 0.9076769948005676,
-      "learning_rate": 2.2676896845694803e-06,
-      "loss": 0.0792,
-      "step": 4535
-    },
-    {
-      "epoch": 0.7733809308149293,
-      "grad_norm": 0.8845469355583191,
-      "learning_rate": 2.265984654731458e-06,
-      "loss": 0.0581,
-      "step": 4536
-    },
-    {
-      "epoch": 0.7735514292564669,
-      "grad_norm": 1.3369200229644775,
-      "learning_rate": 2.264279624893436e-06,
-      "loss": 0.082,
-      "step": 4537
-    },
-    {
-      "epoch": 0.7737219276980046,
-      "grad_norm": 1.3708652257919312,
-      "learning_rate": 2.2625745950554138e-06,
-      "loss": 0.1127,
-      "step": 4538
-    },
-    {
-      "epoch": 0.7738924261395423,
-      "grad_norm": 1.0232014656066895,
-      "learning_rate": 2.2608695652173913e-06,
-      "loss": 0.0548,
-      "step": 4539
-    },
-    {
-      "epoch": 0.77406292458108,
-      "grad_norm": 1.4051992893218994,
-      "learning_rate": 2.2591645353793692e-06,
-      "loss": 0.157,
-      "step": 4540
-    },
-    {
-      "epoch": 0.7742334230226177,
-      "grad_norm": 0.9153316617012024,
-      "learning_rate": 2.257459505541347e-06,
-      "loss": 0.0903,
-      "step": 4541
-    },
-    {
-      "epoch": 0.7744039214641554,
-      "grad_norm": 0.7737805843353271,
-      "learning_rate": 2.255754475703325e-06,
-      "loss": 0.0629,
-      "step": 4542
-    },
-    {
-      "epoch": 0.7745744199056931,
-      "grad_norm": 1.317366123199463,
-      "learning_rate": 2.2540494458653026e-06,
-      "loss": 0.1337,
-      "step": 4543
-    },
-    {
-      "epoch": 0.7747449183472307,
-      "grad_norm": 1.833511471748352,
-      "learning_rate": 2.2523444160272806e-06,
-      "loss": 0.2257,
-      "step": 4544
-    },
-    {
-      "epoch": 0.7749154167887684,
-      "grad_norm": 1.0858491659164429,
-      "learning_rate": 2.2506393861892585e-06,
-      "loss": 0.0734,
-      "step": 4545
-    },
-    {
-      "epoch": 0.7750859152303061,
-      "grad_norm": 1.2199957370758057,
-      "learning_rate": 2.2489343563512365e-06,
-      "loss": 0.1057,
-      "step": 4546
-    },
-    {
-      "epoch": 0.7752564136718438,
-      "grad_norm": 1.3122528791427612,
-      "learning_rate": 2.247229326513214e-06,
-      "loss": 0.0903,
-      "step": 4547
-    },
-    {
-      "epoch": 0.7754269121133814,
-      "grad_norm": 1.3773401975631714,
-      "learning_rate": 2.245524296675192e-06,
-      "loss": 0.1184,
-      "step": 4548
-    },
-    {
-      "epoch": 0.7755974105549192,
-      "grad_norm": 0.9603626132011414,
-      "learning_rate": 2.24381926683717e-06,
-      "loss": 0.1162,
-      "step": 4549
-    },
-    {
-      "epoch": 0.7757679089964569,
-      "grad_norm": 1.2490376234054565,
-      "learning_rate": 2.242114236999148e-06,
-      "loss": 0.1059,
-      "step": 4550
-    },
-    {
-      "epoch": 0.7759384074379945,
-      "grad_norm": 1.633994221687317,
-      "learning_rate": 2.2404092071611254e-06,
-      "loss": 0.0966,
-      "step": 4551
-    },
-    {
-      "epoch": 0.7761089058795322,
-      "grad_norm": 1.8625892400741577,
-      "learning_rate": 2.2387041773231033e-06,
-      "loss": 0.1668,
-      "step": 4552
-    },
-    {
-      "epoch": 0.7762794043210699,
-      "grad_norm": 1.9468538761138916,
-      "learning_rate": 2.2369991474850813e-06,
-      "loss": 0.1483,
-      "step": 4553
-    },
-    {
-      "epoch": 0.7764499027626076,
-      "grad_norm": 1.5350470542907715,
-      "learning_rate": 2.2352941176470592e-06,
-      "loss": 0.1652,
-      "step": 4554
-    },
-    {
-      "epoch": 0.7766204012041452,
-      "grad_norm": 1.5547540187835693,
-      "learning_rate": 2.2335890878090367e-06,
-      "loss": 0.1503,
-      "step": 4555
-    },
-    {
-      "epoch": 0.7767908996456829,
-      "grad_norm": 1.884204387664795,
-      "learning_rate": 2.2318840579710147e-06,
-      "loss": 0.1608,
-      "step": 4556
-    },
-    {
-      "epoch": 0.7769613980872206,
-      "grad_norm": 1.4525254964828491,
-      "learning_rate": 2.2301790281329922e-06,
-      "loss": 0.1111,
-      "step": 4557
-    },
-    {
-      "epoch": 0.7771318965287582,
-      "grad_norm": 1.7327322959899902,
-      "learning_rate": 2.2284739982949706e-06,
-      "loss": 0.178,
-      "step": 4558
-    },
-    {
-      "epoch": 0.777302394970296,
-      "grad_norm": 1.8476815223693848,
-      "learning_rate": 2.226768968456948e-06,
-      "loss": 0.1236,
-      "step": 4559
-    },
-    {
-      "epoch": 0.7774728934118337,
-      "grad_norm": 1.6292099952697754,
-      "learning_rate": 2.225063938618926e-06,
-      "loss": 0.1675,
-      "step": 4560
-    },
-    {
-      "epoch": 0.7776433918533714,
-      "grad_norm": 1.247784972190857,
-      "learning_rate": 2.2233589087809036e-06,
-      "loss": 0.1312,
-      "step": 4561
-    },
-    {
-      "epoch": 0.777813890294909,
-      "grad_norm": 1.5058189630508423,
-      "learning_rate": 2.221653878942882e-06,
-      "loss": 0.1229,
-      "step": 4562
-    },
-    {
-      "epoch": 0.7779843887364467,
-      "grad_norm": 1.342382788658142,
-      "learning_rate": 2.2199488491048595e-06,
-      "loss": 0.0972,
-      "step": 4563
-    },
-    {
-      "epoch": 0.7781548871779844,
-      "grad_norm": 1.530906319618225,
-      "learning_rate": 2.2182438192668374e-06,
-      "loss": 0.095,
-      "step": 4564
-    },
-    {
-      "epoch": 0.778325385619522,
-      "grad_norm": 1.5444718599319458,
-      "learning_rate": 2.216538789428815e-06,
-      "loss": 0.1519,
-      "step": 4565
-    },
-    {
-      "epoch": 0.7784958840610597,
-      "grad_norm": 0.7900330424308777,
-      "learning_rate": 2.214833759590793e-06,
-      "loss": 0.0426,
-      "step": 4566
-    },
-    {
-      "epoch": 0.7786663825025975,
-      "grad_norm": 1.4255989789962769,
-      "learning_rate": 2.213128729752771e-06,
-      "loss": 0.1045,
-      "step": 4567
-    },
-    {
-      "epoch": 0.7788368809441352,
-      "grad_norm": 1.0543164014816284,
-      "learning_rate": 2.211423699914749e-06,
-      "loss": 0.11,
-      "step": 4568
-    },
-    {
-      "epoch": 0.7790073793856728,
-      "grad_norm": 1.1432523727416992,
-      "learning_rate": 2.2097186700767263e-06,
-      "loss": 0.1237,
-      "step": 4569
-    },
-    {
-      "epoch": 0.7791778778272105,
-      "grad_norm": 1.9459341764450073,
-      "learning_rate": 2.2080136402387043e-06,
-      "loss": 0.1522,
-      "step": 4570
-    },
-    {
-      "epoch": 0.7793483762687482,
-      "grad_norm": 2.064140558242798,
-      "learning_rate": 2.2063086104006822e-06,
-      "loss": 0.1779,
-      "step": 4571
-    },
-    {
-      "epoch": 0.7795188747102858,
-      "grad_norm": 1.0942349433898926,
-      "learning_rate": 2.20460358056266e-06,
-      "loss": 0.112,
-      "step": 4572
-    },
-    {
-      "epoch": 0.7796893731518235,
-      "grad_norm": 1.4263490438461304,
-      "learning_rate": 2.2028985507246377e-06,
-      "loss": 0.1094,
-      "step": 4573
-    },
-    {
-      "epoch": 0.7798598715933612,
-      "grad_norm": 1.002995252609253,
-      "learning_rate": 2.2011935208866156e-06,
-      "loss": 0.1146,
-      "step": 4574
-    },
-    {
-      "epoch": 0.780030370034899,
-      "grad_norm": 1.8714383840560913,
-      "learning_rate": 2.1994884910485936e-06,
-      "loss": 0.1763,
-      "step": 4575
-    },
-    {
-      "epoch": 0.7802008684764365,
-      "grad_norm": 1.303443193435669,
-      "learning_rate": 2.1977834612105715e-06,
-      "loss": 0.1416,
-      "step": 4576
-    },
-    {
-      "epoch": 0.7803713669179743,
-      "grad_norm": 1.5888121128082275,
-      "learning_rate": 2.196078431372549e-06,
-      "loss": 0.2209,
-      "step": 4577
-    },
-    {
-      "epoch": 0.780541865359512,
-      "grad_norm": 1.908546805381775,
-      "learning_rate": 2.194373401534527e-06,
-      "loss": 0.1524,
-      "step": 4578
-    },
-    {
-      "epoch": 0.7807123638010496,
-      "grad_norm": 2.010317087173462,
-      "learning_rate": 2.192668371696505e-06,
-      "loss": 0.1973,
-      "step": 4579
-    },
-    {
-      "epoch": 0.7808828622425873,
-      "grad_norm": 1.123651385307312,
-      "learning_rate": 2.190963341858483e-06,
-      "loss": 0.1256,
-      "step": 4580
-    },
-    {
-      "epoch": 0.781053360684125,
-      "grad_norm": 1.4087166786193848,
-      "learning_rate": 2.1892583120204604e-06,
-      "loss": 0.1015,
-      "step": 4581
-    },
-    {
-      "epoch": 0.7812238591256627,
-      "grad_norm": 1.4993146657943726,
-      "learning_rate": 2.1875532821824384e-06,
-      "loss": 0.1835,
-      "step": 4582
-    },
-    {
-      "epoch": 0.7813943575672003,
-      "grad_norm": 1.4932217597961426,
-      "learning_rate": 2.1858482523444163e-06,
-      "loss": 0.1404,
-      "step": 4583
-    },
-    {
-      "epoch": 0.781564856008738,
-      "grad_norm": 2.5937252044677734,
-      "learning_rate": 2.1841432225063943e-06,
-      "loss": 0.1626,
-      "step": 4584
-    },
-    {
-      "epoch": 0.7817353544502758,
-      "grad_norm": 1.0918924808502197,
-      "learning_rate": 2.1824381926683718e-06,
-      "loss": 0.0876,
-      "step": 4585
-    },
-    {
-      "epoch": 0.7819058528918134,
-      "grad_norm": 0.8180877566337585,
-      "learning_rate": 2.1807331628303497e-06,
-      "loss": 0.0312,
-      "step": 4586
-    },
-    {
-      "epoch": 0.7820763513333511,
-      "grad_norm": 3.8481969833374023,
-      "learning_rate": 2.1790281329923277e-06,
-      "loss": 0.2022,
-      "step": 4587
-    },
-    {
-      "epoch": 0.7822468497748888,
-      "grad_norm": 0.9197229146957397,
-      "learning_rate": 2.1773231031543056e-06,
-      "loss": 0.0463,
-      "step": 4588
-    },
-    {
-      "epoch": 0.7824173482164265,
-      "grad_norm": 0.9426055550575256,
-      "learning_rate": 2.175618073316283e-06,
-      "loss": 0.066,
-      "step": 4589
-    },
-    {
-      "epoch": 0.7825878466579641,
-      "grad_norm": 1.4380828142166138,
-      "learning_rate": 2.173913043478261e-06,
-      "loss": 0.1376,
-      "step": 4590
-    },
-    {
-      "epoch": 0.7827583450995018,
-      "grad_norm": 1.3979727029800415,
-      "learning_rate": 2.1722080136402386e-06,
-      "loss": 0.1191,
-      "step": 4591
-    },
-    {
-      "epoch": 0.7829288435410395,
-      "grad_norm": 1.105899691581726,
-      "learning_rate": 2.170502983802217e-06,
-      "loss": 0.0955,
-      "step": 4592
-    },
-    {
-      "epoch": 0.7830993419825771,
-      "grad_norm": 1.049070954322815,
-      "learning_rate": 2.1687979539641945e-06,
-      "loss": 0.1086,
-      "step": 4593
-    },
-    {
-      "epoch": 0.7832698404241148,
-      "grad_norm": 1.0583131313323975,
-      "learning_rate": 2.1670929241261725e-06,
-      "loss": 0.1121,
-      "step": 4594
-    },
-    {
-      "epoch": 0.7834403388656526,
-      "grad_norm": 2.0508458614349365,
-      "learning_rate": 2.16538789428815e-06,
-      "loss": 0.2152,
-      "step": 4595
-    },
-    {
-      "epoch": 0.7836108373071903,
-      "grad_norm": 1.0575108528137207,
-      "learning_rate": 2.163682864450128e-06,
-      "loss": 0.0627,
-      "step": 4596
-    },
-    {
-      "epoch": 0.7837813357487279,
-      "grad_norm": 1.286759614944458,
-      "learning_rate": 2.161977834612106e-06,
-      "loss": 0.1289,
-      "step": 4597
-    },
-    {
-      "epoch": 0.7839518341902656,
-      "grad_norm": 2.3939502239227295,
-      "learning_rate": 2.160272804774084e-06,
-      "loss": 0.1199,
-      "step": 4598
-    },
-    {
-      "epoch": 0.7841223326318033,
-      "grad_norm": 1.2347538471221924,
-      "learning_rate": 2.1585677749360614e-06,
-      "loss": 0.145,
-      "step": 4599
-    },
-    {
-      "epoch": 0.784292831073341,
-      "grad_norm": 0.7571415901184082,
-      "learning_rate": 2.1568627450980393e-06,
-      "loss": 0.0456,
-      "step": 4600
-    },
-    {
-      "epoch": 0.784292831073341,
-      "eval_f1_score": 0.3776595744680851,
-      "eval_loss": 0.13533847033977509,
-      "eval_runtime": 183.3624,
-      "eval_samples_per_second": 54.537,
-      "eval_steps_per_second": 3.409,
-      "step": 4600
-    },
-    {
-      "epoch": 0.7844633295148786,
-      "grad_norm": 0.9207397103309631,
-      "learning_rate": 2.1551577152600173e-06,
-      "loss": 0.0878,
-      "step": 4601
-    },
-    {
-      "epoch": 0.7846338279564163,
-      "grad_norm": 1.8424160480499268,
-      "learning_rate": 2.153452685421995e-06,
-      "loss": 0.106,
-      "step": 4602
-    },
-    {
-      "epoch": 0.784804326397954,
-      "grad_norm": 1.3503949642181396,
-      "learning_rate": 2.1517476555839727e-06,
-      "loss": 0.1342,
-      "step": 4603
-    },
-    {
-      "epoch": 0.7849748248394917,
-      "grad_norm": 1.642559289932251,
-      "learning_rate": 2.1500426257459507e-06,
-      "loss": 0.1322,
-      "step": 4604
-    },
-    {
-      "epoch": 0.7851453232810294,
-      "grad_norm": 1.4871376752853394,
-      "learning_rate": 2.1483375959079286e-06,
-      "loss": 0.14,
-      "step": 4605
-    },
-    {
-      "epoch": 0.7853158217225671,
-      "grad_norm": 1.3946406841278076,
-      "learning_rate": 2.1466325660699066e-06,
-      "loss": 0.1535,
-      "step": 4606
-    },
-    {
-      "epoch": 0.7854863201641048,
-      "grad_norm": 1.222882866859436,
-      "learning_rate": 2.144927536231884e-06,
-      "loss": 0.0676,
-      "step": 4607
-    },
-    {
-      "epoch": 0.7856568186056424,
-      "grad_norm": 1.5095902681350708,
-      "learning_rate": 2.143222506393862e-06,
-      "loss": 0.1177,
-      "step": 4608
-    },
-    {
-      "epoch": 0.7858273170471801,
-      "grad_norm": 1.2253878116607666,
-      "learning_rate": 2.14151747655584e-06,
-      "loss": 0.1358,
-      "step": 4609
-    },
-    {
-      "epoch": 0.7859978154887178,
-      "grad_norm": 1.1032277345657349,
-      "learning_rate": 2.139812446717818e-06,
-      "loss": 0.0618,
-      "step": 4610
-    },
-    {
-      "epoch": 0.7861683139302554,
-      "grad_norm": 1.2632951736450195,
-      "learning_rate": 2.1381074168797955e-06,
-      "loss": 0.101,
-      "step": 4611
-    },
-    {
-      "epoch": 0.7863388123717931,
-      "grad_norm": 0.7547646164894104,
-      "learning_rate": 2.1364023870417734e-06,
-      "loss": 0.0467,
-      "step": 4612
-    },
-    {
-      "epoch": 0.7865093108133309,
-      "grad_norm": 1.7445420026779175,
-      "learning_rate": 2.1346973572037514e-06,
-      "loss": 0.1337,
-      "step": 4613
-    },
-    {
-      "epoch": 0.7866798092548686,
-      "grad_norm": 3.457361936569214,
-      "learning_rate": 2.132992327365729e-06,
-      "loss": 0.224,
-      "step": 4614
-    },
-    {
-      "epoch": 0.7868503076964062,
-      "grad_norm": 1.6196421384811401,
-      "learning_rate": 2.131287297527707e-06,
-      "loss": 0.0935,
-      "step": 4615
-    },
-    {
-      "epoch": 0.7870208061379439,
-      "grad_norm": 1.171519160270691,
-      "learning_rate": 2.1295822676896848e-06,
-      "loss": 0.0683,
-      "step": 4616
-    },
-    {
-      "epoch": 0.7871913045794816,
-      "grad_norm": 1.2372194528579712,
-      "learning_rate": 2.1278772378516627e-06,
-      "loss": 0.1254,
-      "step": 4617
-    },
-    {
-      "epoch": 0.7873618030210192,
-      "grad_norm": 1.5402429103851318,
-      "learning_rate": 2.1261722080136403e-06,
-      "loss": 0.1376,
-      "step": 4618
-    },
-    {
-      "epoch": 0.7875323014625569,
-      "grad_norm": 1.385901689529419,
-      "learning_rate": 2.124467178175618e-06,
-      "loss": 0.1525,
-      "step": 4619
-    },
-    {
-      "epoch": 0.7877027999040946,
-      "grad_norm": 1.2449439764022827,
-      "learning_rate": 2.122762148337596e-06,
-      "loss": 0.1271,
-      "step": 4620
-    },
-    {
-      "epoch": 0.7878732983456324,
-      "grad_norm": 1.6943930387496948,
-      "learning_rate": 2.121057118499574e-06,
-      "loss": 0.1269,
-      "step": 4621
-    },
-    {
-      "epoch": 0.78804379678717,
-      "grad_norm": 1.9401750564575195,
-      "learning_rate": 2.1193520886615516e-06,
-      "loss": 0.2205,
-      "step": 4622
-    },
-    {
-      "epoch": 0.7882142952287077,
-      "grad_norm": 1.010846734046936,
-      "learning_rate": 2.1176470588235296e-06,
-      "loss": 0.0951,
-      "step": 4623
-    },
-    {
-      "epoch": 0.7883847936702454,
-      "grad_norm": 0.9300770163536072,
-      "learning_rate": 2.1159420289855075e-06,
-      "loss": 0.0448,
-      "step": 4624
-    },
-    {
-      "epoch": 0.788555292111783,
-      "grad_norm": 1.5240856409072876,
-      "learning_rate": 2.1142369991474855e-06,
-      "loss": 0.1576,
-      "step": 4625
-    },
-    {
-      "epoch": 0.7887257905533207,
-      "grad_norm": 1.2999317646026611,
-      "learning_rate": 2.112531969309463e-06,
-      "loss": 0.0857,
-      "step": 4626
-    },
-    {
-      "epoch": 0.7888962889948584,
-      "grad_norm": 1.4437685012817383,
-      "learning_rate": 2.110826939471441e-06,
-      "loss": 0.134,
-      "step": 4627
-    },
-    {
-      "epoch": 0.7890667874363961,
-      "grad_norm": 1.392773151397705,
-      "learning_rate": 2.109121909633419e-06,
-      "loss": 0.135,
-      "step": 4628
-    },
-    {
-      "epoch": 0.7892372858779337,
-      "grad_norm": 1.3370131254196167,
-      "learning_rate": 2.107416879795397e-06,
-      "loss": 0.1207,
-      "step": 4629
-    },
-    {
-      "epoch": 0.7894077843194715,
-      "grad_norm": 1.3597493171691895,
-      "learning_rate": 2.1057118499573744e-06,
-      "loss": 0.1099,
-      "step": 4630
-    },
-    {
-      "epoch": 0.7895782827610092,
-      "grad_norm": 2.7039904594421387,
-      "learning_rate": 2.1040068201193523e-06,
-      "loss": 0.2182,
-      "step": 4631
-    },
-    {
-      "epoch": 0.7897487812025468,
-      "grad_norm": 1.292164921760559,
-      "learning_rate": 2.10230179028133e-06,
-      "loss": 0.1013,
-      "step": 4632
-    },
-    {
-      "epoch": 0.7899192796440845,
-      "grad_norm": 1.3703480958938599,
-      "learning_rate": 2.1005967604433078e-06,
-      "loss": 0.0705,
-      "step": 4633
-    },
-    {
-      "epoch": 0.7900897780856222,
-      "grad_norm": 1.4119629859924316,
-      "learning_rate": 2.0988917306052857e-06,
-      "loss": 0.1079,
-      "step": 4634
-    },
-    {
-      "epoch": 0.7902602765271599,
-      "grad_norm": 2.1801319122314453,
-      "learning_rate": 2.0971867007672637e-06,
-      "loss": 0.1929,
-      "step": 4635
-    },
-    {
-      "epoch": 0.7904307749686975,
-      "grad_norm": 1.5141770839691162,
-      "learning_rate": 2.095481670929241e-06,
-      "loss": 0.1497,
-      "step": 4636
-    },
-    {
-      "epoch": 0.7906012734102352,
-      "grad_norm": 1.2041051387786865,
-      "learning_rate": 2.093776641091219e-06,
-      "loss": 0.0976,
-      "step": 4637
-    },
-    {
-      "epoch": 0.7907717718517729,
-      "grad_norm": 2.3297648429870605,
-      "learning_rate": 2.092071611253197e-06,
-      "loss": 0.2369,
-      "step": 4638
-    },
-    {
-      "epoch": 0.7909422702933105,
-      "grad_norm": 0.6728577613830566,
-      "learning_rate": 2.090366581415175e-06,
-      "loss": 0.0651,
-      "step": 4639
-    },
-    {
-      "epoch": 0.7911127687348483,
-      "grad_norm": 1.3690725564956665,
-      "learning_rate": 2.0886615515771526e-06,
-      "loss": 0.0873,
-      "step": 4640
-    },
-    {
-      "epoch": 0.791283267176386,
-      "grad_norm": 1.7001179456710815,
-      "learning_rate": 2.0869565217391305e-06,
-      "loss": 0.1586,
-      "step": 4641
-    },
-    {
-      "epoch": 0.7914537656179237,
-      "grad_norm": 1.7392905950546265,
-      "learning_rate": 2.0852514919011085e-06,
-      "loss": 0.1737,
-      "step": 4642
-    },
-    {
-      "epoch": 0.7916242640594613,
-      "grad_norm": 2.2579967975616455,
-      "learning_rate": 2.0835464620630864e-06,
-      "loss": 0.125,
-      "step": 4643
-    },
-    {
-      "epoch": 0.791794762500999,
-      "grad_norm": 1.8998128175735474,
-      "learning_rate": 2.081841432225064e-06,
-      "loss": 0.1891,
-      "step": 4644
-    },
-    {
-      "epoch": 0.7919652609425367,
-      "grad_norm": 1.4053045511245728,
-      "learning_rate": 2.080136402387042e-06,
-      "loss": 0.0873,
-      "step": 4645
-    },
-    {
-      "epoch": 0.7921357593840744,
-      "grad_norm": 2.0661635398864746,
-      "learning_rate": 2.07843137254902e-06,
-      "loss": 0.169,
-      "step": 4646
-    },
-    {
-      "epoch": 0.792306257825612,
-      "grad_norm": 1.3633352518081665,
-      "learning_rate": 2.0767263427109978e-06,
-      "loss": 0.0958,
-      "step": 4647
-    },
-    {
-      "epoch": 0.7924767562671498,
-      "grad_norm": 0.9656442999839783,
-      "learning_rate": 2.0750213128729753e-06,
-      "loss": 0.0431,
-      "step": 4648
-    },
-    {
-      "epoch": 0.7926472547086875,
-      "grad_norm": 1.8652012348175049,
-      "learning_rate": 2.0733162830349532e-06,
-      "loss": 0.2104,
-      "step": 4649
-    },
-    {
-      "epoch": 0.7928177531502251,
-      "grad_norm": 1.1847926378250122,
-      "learning_rate": 2.071611253196931e-06,
-      "loss": 0.103,
-      "step": 4650
-    },
-    {
-      "epoch": 0.7929882515917628,
-      "grad_norm": 1.349178671836853,
-      "learning_rate": 2.069906223358909e-06,
-      "loss": 0.1452,
-      "step": 4651
-    },
-    {
-      "epoch": 0.7931587500333005,
-      "grad_norm": 2.4015870094299316,
-      "learning_rate": 2.0682011935208867e-06,
-      "loss": 0.1767,
-      "step": 4652
-    },
-    {
-      "epoch": 0.7933292484748382,
-      "grad_norm": 1.461780309677124,
-      "learning_rate": 2.0664961636828646e-06,
-      "loss": 0.0683,
-      "step": 4653
-    },
-    {
-      "epoch": 0.7934997469163758,
-      "grad_norm": 1.072913646697998,
-      "learning_rate": 2.0647911338448426e-06,
-      "loss": 0.1258,
-      "step": 4654
-    },
-    {
-      "epoch": 0.7936702453579135,
-      "grad_norm": 1.6158761978149414,
-      "learning_rate": 2.0630861040068205e-06,
-      "loss": 0.1743,
-      "step": 4655
-    },
-    {
-      "epoch": 0.7938407437994512,
-      "grad_norm": 1.0536906719207764,
-      "learning_rate": 2.061381074168798e-06,
-      "loss": 0.1707,
-      "step": 4656
-    },
-    {
-      "epoch": 0.7940112422409888,
-      "grad_norm": 0.9950072169303894,
-      "learning_rate": 2.059676044330776e-06,
-      "loss": 0.0788,
-      "step": 4657
-    },
-    {
-      "epoch": 0.7941817406825266,
-      "grad_norm": 1.13616943359375,
-      "learning_rate": 2.0579710144927535e-06,
-      "loss": 0.1079,
-      "step": 4658
-    },
-    {
-      "epoch": 0.7943522391240643,
-      "grad_norm": 1.375840663909912,
-      "learning_rate": 2.056265984654732e-06,
-      "loss": 0.123,
-      "step": 4659
-    },
-    {
-      "epoch": 0.794522737565602,
-      "grad_norm": 1.3938781023025513,
-      "learning_rate": 2.0545609548167094e-06,
-      "loss": 0.1818,
-      "step": 4660
-    },
-    {
-      "epoch": 0.7946932360071396,
-      "grad_norm": 1.6234029531478882,
-      "learning_rate": 2.0528559249786873e-06,
-      "loss": 0.1373,
-      "step": 4661
-    },
-    {
-      "epoch": 0.7948637344486773,
-      "grad_norm": 1.4031246900558472,
-      "learning_rate": 2.051150895140665e-06,
-      "loss": 0.1111,
-      "step": 4662
-    },
-    {
-      "epoch": 0.795034232890215,
-      "grad_norm": 0.9812381267547607,
-      "learning_rate": 2.0494458653026432e-06,
-      "loss": 0.1086,
-      "step": 4663
-    },
-    {
-      "epoch": 0.7952047313317526,
-      "grad_norm": 1.050758957862854,
-      "learning_rate": 2.0477408354646208e-06,
-      "loss": 0.1264,
-      "step": 4664
-    },
-    {
-      "epoch": 0.7953752297732903,
-      "grad_norm": 1.0812690258026123,
-      "learning_rate": 2.0460358056265987e-06,
-      "loss": 0.1103,
-      "step": 4665
-    },
-    {
-      "epoch": 0.795545728214828,
-      "grad_norm": 1.3163347244262695,
-      "learning_rate": 2.0443307757885762e-06,
-      "loss": 0.1515,
-      "step": 4666
-    },
-    {
-      "epoch": 0.7957162266563658,
-      "grad_norm": 1.0045158863067627,
-      "learning_rate": 2.0426257459505546e-06,
-      "loss": 0.0468,
-      "step": 4667
-    },
-    {
-      "epoch": 0.7958867250979034,
-      "grad_norm": 0.9080355763435364,
-      "learning_rate": 2.040920716112532e-06,
-      "loss": 0.0406,
-      "step": 4668
-    },
-    {
-      "epoch": 0.7960572235394411,
-      "grad_norm": 1.759256362915039,
-      "learning_rate": 2.03921568627451e-06,
-      "loss": 0.1086,
-      "step": 4669
-    },
-    {
-      "epoch": 0.7962277219809788,
-      "grad_norm": 1.5688270330429077,
-      "learning_rate": 2.0375106564364876e-06,
-      "loss": 0.1196,
-      "step": 4670
-    },
-    {
-      "epoch": 0.7963982204225164,
-      "grad_norm": 1.7334173917770386,
-      "learning_rate": 2.0358056265984655e-06,
-      "loss": 0.1583,
-      "step": 4671
-    },
-    {
-      "epoch": 0.7965687188640541,
-      "grad_norm": 1.627323865890503,
-      "learning_rate": 2.0341005967604435e-06,
-      "loss": 0.0697,
-      "step": 4672
-    },
-    {
-      "epoch": 0.7967392173055918,
-      "grad_norm": 1.538482666015625,
-      "learning_rate": 2.0323955669224214e-06,
-      "loss": 0.1348,
-      "step": 4673
-    },
-    {
-      "epoch": 0.7969097157471295,
-      "grad_norm": 1.303102731704712,
-      "learning_rate": 2.030690537084399e-06,
-      "loss": 0.1172,
-      "step": 4674
-    },
-    {
-      "epoch": 0.7970802141886671,
-      "grad_norm": 1.290687918663025,
-      "learning_rate": 2.028985507246377e-06,
-      "loss": 0.1082,
-      "step": 4675
-    },
-    {
-      "epoch": 0.7972507126302049,
-      "grad_norm": 2.2509918212890625,
-      "learning_rate": 2.027280477408355e-06,
-      "loss": 0.1668,
-      "step": 4676
-    },
-    {
-      "epoch": 0.7974212110717426,
-      "grad_norm": 1.1365306377410889,
-      "learning_rate": 2.025575447570333e-06,
-      "loss": 0.1073,
-      "step": 4677
-    },
-    {
-      "epoch": 0.7975917095132802,
-      "grad_norm": 1.3142699003219604,
-      "learning_rate": 2.0238704177323103e-06,
-      "loss": 0.1258,
-      "step": 4678
-    },
-    {
-      "epoch": 0.7977622079548179,
-      "grad_norm": 1.2691103219985962,
-      "learning_rate": 2.0221653878942883e-06,
-      "loss": 0.0808,
-      "step": 4679
-    },
-    {
-      "epoch": 0.7979327063963556,
-      "grad_norm": 1.9458601474761963,
-      "learning_rate": 2.0204603580562662e-06,
-      "loss": 0.1571,
-      "step": 4680
-    },
-    {
-      "epoch": 0.7981032048378933,
-      "grad_norm": 0.9146658182144165,
-      "learning_rate": 2.018755328218244e-06,
-      "loss": 0.0478,
-      "step": 4681
-    },
-    {
-      "epoch": 0.7982737032794309,
-      "grad_norm": 1.0852900743484497,
-      "learning_rate": 2.0170502983802217e-06,
-      "loss": 0.1092,
-      "step": 4682
-    },
-    {
-      "epoch": 0.7984442017209686,
-      "grad_norm": 1.535971999168396,
-      "learning_rate": 2.0153452685421997e-06,
-      "loss": 0.103,
-      "step": 4683
-    },
-    {
-      "epoch": 0.7986147001625064,
-      "grad_norm": 1.4463566541671753,
-      "learning_rate": 2.0136402387041776e-06,
-      "loss": 0.1268,
-      "step": 4684
-    },
-    {
-      "epoch": 0.798785198604044,
-      "grad_norm": 1.2712289094924927,
-      "learning_rate": 2.0119352088661555e-06,
-      "loss": 0.0915,
-      "step": 4685
-    },
-    {
-      "epoch": 0.7989556970455817,
-      "grad_norm": 0.8879340887069702,
-      "learning_rate": 2.010230179028133e-06,
-      "loss": 0.0396,
-      "step": 4686
-    },
-    {
-      "epoch": 0.7991261954871194,
-      "grad_norm": 1.5370845794677734,
-      "learning_rate": 2.008525149190111e-06,
-      "loss": 0.1355,
-      "step": 4687
-    },
-    {
-      "epoch": 0.7992966939286571,
-      "grad_norm": 1.276149868965149,
-      "learning_rate": 2.006820119352089e-06,
-      "loss": 0.0813,
-      "step": 4688
-    },
-    {
-      "epoch": 0.7994671923701947,
-      "grad_norm": 1.219150185585022,
-      "learning_rate": 2.0051150895140665e-06,
-      "loss": 0.1205,
-      "step": 4689
-    },
-    {
-      "epoch": 0.7996376908117324,
-      "grad_norm": 1.353977918624878,
-      "learning_rate": 2.0034100596760444e-06,
-      "loss": 0.0943,
-      "step": 4690
-    },
-    {
-      "epoch": 0.7998081892532701,
-      "grad_norm": 1.6376045942306519,
-      "learning_rate": 2.0017050298380224e-06,
-      "loss": 0.0888,
-      "step": 4691
-    },
-    {
-      "epoch": 0.7999786876948078,
-      "grad_norm": 0.8724763989448547,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.0747,
-      "step": 4692
-    },
-    {
-      "epoch": 0.8001491861363454,
-      "grad_norm": 1.2550649642944336,
-      "learning_rate": 1.998294970161978e-06,
-      "loss": 0.1348,
-      "step": 4693
-    },
-    {
-      "epoch": 0.8003196845778832,
-      "grad_norm": 1.7393126487731934,
-      "learning_rate": 1.996589940323956e-06,
-      "loss": 0.1726,
-      "step": 4694
-    },
-    {
-      "epoch": 0.8004901830194209,
-      "grad_norm": 0.9823253154754639,
-      "learning_rate": 1.9948849104859338e-06,
-      "loss": 0.1048,
-      "step": 4695
-    },
-    {
-      "epoch": 0.8006606814609585,
-      "grad_norm": 2.0837788581848145,
-      "learning_rate": 1.9931798806479117e-06,
-      "loss": 0.1357,
-      "step": 4696
-    },
-    {
-      "epoch": 0.8008311799024962,
-      "grad_norm": 1.061660647392273,
-      "learning_rate": 1.9914748508098892e-06,
-      "loss": 0.0809,
-      "step": 4697
-    },
-    {
-      "epoch": 0.8010016783440339,
-      "grad_norm": 1.0696051120758057,
-      "learning_rate": 1.989769820971867e-06,
-      "loss": 0.0773,
-      "step": 4698
-    },
-    {
-      "epoch": 0.8011721767855716,
-      "grad_norm": 0.8411040306091309,
-      "learning_rate": 1.988064791133845e-06,
-      "loss": 0.0861,
-      "step": 4699
-    },
-    {
-      "epoch": 0.8013426752271092,
-      "grad_norm": 2.4212546348571777,
-      "learning_rate": 1.986359761295823e-06,
-      "loss": 0.1266,
-      "step": 4700
-    },
-    {
-      "epoch": 0.8013426752271092,
-      "eval_f1_score": 0.4120603015075377,
-      "eval_loss": 0.13393385708332062,
-      "eval_runtime": 183.4415,
-      "eval_samples_per_second": 54.513,
-      "eval_steps_per_second": 3.407,
-      "step": 4700
-    },
-    {
-      "epoch": 0.8015131736686469,
-      "grad_norm": 1.3715274333953857,
-      "learning_rate": 1.9846547314578006e-06,
-      "loss": 0.1156,
-      "step": 4701
-    },
-    {
-      "epoch": 0.8016836721101847,
-      "grad_norm": 2.326615571975708,
-      "learning_rate": 1.9829497016197785e-06,
-      "loss": 0.1544,
-      "step": 4702
-    },
-    {
-      "epoch": 0.8018541705517223,
-      "grad_norm": 1.6274570226669312,
-      "learning_rate": 1.9812446717817565e-06,
-      "loss": 0.1187,
-      "step": 4703
-    },
-    {
-      "epoch": 0.80202466899326,
-      "grad_norm": 1.5323635339736938,
-      "learning_rate": 1.979539641943734e-06,
-      "loss": 0.1284,
-      "step": 4704
-    },
-    {
-      "epoch": 0.8021951674347977,
-      "grad_norm": 1.5731792449951172,
-      "learning_rate": 1.977834612105712e-06,
-      "loss": 0.1295,
-      "step": 4705
-    },
-    {
-      "epoch": 0.8023656658763354,
-      "grad_norm": 1.1835929155349731,
-      "learning_rate": 1.97612958226769e-06,
-      "loss": 0.0636,
-      "step": 4706
-    },
-    {
-      "epoch": 0.802536164317873,
-      "grad_norm": 1.0732511281967163,
-      "learning_rate": 1.9744245524296674e-06,
-      "loss": 0.0785,
-      "step": 4707
-    },
-    {
-      "epoch": 0.8027066627594107,
-      "grad_norm": 1.892913818359375,
-      "learning_rate": 1.9727195225916454e-06,
-      "loss": 0.1303,
-      "step": 4708
-    },
-    {
-      "epoch": 0.8028771612009484,
-      "grad_norm": 1.480954885482788,
-      "learning_rate": 1.9710144927536233e-06,
-      "loss": 0.1414,
-      "step": 4709
-    },
-    {
-      "epoch": 0.803047659642486,
-      "grad_norm": 1.3224931955337524,
-      "learning_rate": 1.9693094629156013e-06,
-      "loss": 0.112,
-      "step": 4710
-    },
-    {
-      "epoch": 0.8032181580840237,
-      "grad_norm": 1.0266205072402954,
-      "learning_rate": 1.967604433077579e-06,
-      "loss": 0.0677,
-      "step": 4711
-    },
-    {
-      "epoch": 0.8033886565255615,
-      "grad_norm": 1.4693719148635864,
-      "learning_rate": 1.9658994032395567e-06,
-      "loss": 0.0957,
-      "step": 4712
-    },
-    {
-      "epoch": 0.8035591549670992,
-      "grad_norm": 1.6250672340393066,
-      "learning_rate": 1.9641943734015347e-06,
-      "loss": 0.159,
-      "step": 4713
-    },
-    {
-      "epoch": 0.8037296534086368,
-      "grad_norm": 2.1153995990753174,
-      "learning_rate": 1.9624893435635126e-06,
-      "loss": 0.1968,
-      "step": 4714
-    },
-    {
-      "epoch": 0.8039001518501745,
-      "grad_norm": 2.047391891479492,
-      "learning_rate": 1.96078431372549e-06,
-      "loss": 0.1825,
-      "step": 4715
-    },
-    {
-      "epoch": 0.8040706502917122,
-      "grad_norm": 1.5375449657440186,
-      "learning_rate": 1.959079283887468e-06,
-      "loss": 0.124,
-      "step": 4716
-    },
-    {
-      "epoch": 0.8042411487332498,
-      "grad_norm": 1.9991374015808105,
-      "learning_rate": 1.957374254049446e-06,
-      "loss": 0.1525,
-      "step": 4717
-    },
-    {
-      "epoch": 0.8044116471747875,
-      "grad_norm": 1.095762848854065,
-      "learning_rate": 1.955669224211424e-06,
-      "loss": 0.1023,
-      "step": 4718
-    },
-    {
-      "epoch": 0.8045821456163252,
-      "grad_norm": 1.0201056003570557,
-      "learning_rate": 1.9539641943734015e-06,
-      "loss": 0.103,
-      "step": 4719
-    },
-    {
-      "epoch": 0.804752644057863,
-      "grad_norm": 1.3266394138336182,
-      "learning_rate": 1.9522591645353795e-06,
-      "loss": 0.1133,
-      "step": 4720
-    },
-    {
-      "epoch": 0.8049231424994006,
-      "grad_norm": 1.4412189722061157,
-      "learning_rate": 1.9505541346973574e-06,
-      "loss": 0.0988,
-      "step": 4721
-    },
-    {
-      "epoch": 0.8050936409409383,
-      "grad_norm": 1.7581360340118408,
-      "learning_rate": 1.9488491048593354e-06,
-      "loss": 0.2479,
-      "step": 4722
-    },
-    {
-      "epoch": 0.805264139382476,
-      "grad_norm": 1.7024379968643188,
-      "learning_rate": 1.947144075021313e-06,
-      "loss": 0.1456,
-      "step": 4723
-    },
-    {
-      "epoch": 0.8054346378240136,
-      "grad_norm": 1.8094717264175415,
-      "learning_rate": 1.945439045183291e-06,
-      "loss": 0.1349,
-      "step": 4724
-    },
-    {
-      "epoch": 0.8056051362655513,
-      "grad_norm": 2.129314422607422,
-      "learning_rate": 1.9437340153452684e-06,
-      "loss": 0.218,
-      "step": 4725
-    },
-    {
-      "epoch": 0.805775634707089,
-      "grad_norm": 1.45778226852417,
-      "learning_rate": 1.9420289855072467e-06,
-      "loss": 0.1524,
-      "step": 4726
-    },
-    {
-      "epoch": 0.8059461331486267,
-      "grad_norm": 1.0143263339996338,
-      "learning_rate": 1.9403239556692243e-06,
-      "loss": 0.0989,
-      "step": 4727
-    },
-    {
-      "epoch": 0.8061166315901643,
-      "grad_norm": 1.1257439851760864,
-      "learning_rate": 1.9386189258312022e-06,
-      "loss": 0.1012,
-      "step": 4728
-    },
-    {
-      "epoch": 0.806287130031702,
-      "grad_norm": 1.2927134037017822,
-      "learning_rate": 1.9369138959931797e-06,
-      "loss": 0.0731,
-      "step": 4729
-    },
-    {
-      "epoch": 0.8064576284732398,
-      "grad_norm": 1.4720287322998047,
-      "learning_rate": 1.935208866155158e-06,
-      "loss": 0.1371,
-      "step": 4730
-    },
-    {
-      "epoch": 0.8066281269147774,
-      "grad_norm": 1.858988642692566,
-      "learning_rate": 1.9335038363171356e-06,
-      "loss": 0.2001,
-      "step": 4731
-    },
-    {
-      "epoch": 0.8067986253563151,
-      "grad_norm": 1.4337401390075684,
-      "learning_rate": 1.9317988064791136e-06,
-      "loss": 0.1453,
-      "step": 4732
-    },
-    {
-      "epoch": 0.8069691237978528,
-      "grad_norm": 1.4073408842086792,
-      "learning_rate": 1.930093776641091e-06,
-      "loss": 0.0994,
-      "step": 4733
-    },
-    {
-      "epoch": 0.8071396222393905,
-      "grad_norm": 0.98497474193573,
-      "learning_rate": 1.9283887468030695e-06,
-      "loss": 0.0706,
-      "step": 4734
-    },
-    {
-      "epoch": 0.8073101206809281,
-      "grad_norm": 1.5933598279953003,
-      "learning_rate": 1.926683716965047e-06,
-      "loss": 0.1477,
-      "step": 4735
-    },
-    {
-      "epoch": 0.8074806191224658,
-      "grad_norm": 1.8039004802703857,
-      "learning_rate": 1.924978687127025e-06,
-      "loss": 0.1542,
-      "step": 4736
-    },
-    {
-      "epoch": 0.8076511175640035,
-      "grad_norm": 0.9617605805397034,
-      "learning_rate": 1.9232736572890025e-06,
-      "loss": 0.0702,
-      "step": 4737
-    },
-    {
-      "epoch": 0.8078216160055413,
-      "grad_norm": 1.2599680423736572,
-      "learning_rate": 1.921568627450981e-06,
-      "loss": 0.0808,
-      "step": 4738
-    },
-    {
-      "epoch": 0.8079921144470789,
-      "grad_norm": 1.4257581233978271,
-      "learning_rate": 1.9198635976129584e-06,
-      "loss": 0.1185,
-      "step": 4739
-    },
-    {
-      "epoch": 0.8081626128886166,
-      "grad_norm": 1.2117432355880737,
-      "learning_rate": 1.9181585677749363e-06,
-      "loss": 0.1182,
-      "step": 4740
-    },
-    {
-      "epoch": 0.8083331113301543,
-      "grad_norm": 2.570119619369507,
-      "learning_rate": 1.916453537936914e-06,
-      "loss": 0.2357,
-      "step": 4741
-    },
-    {
-      "epoch": 0.8085036097716919,
-      "grad_norm": 1.4911688566207886,
-      "learning_rate": 1.9147485080988922e-06,
-      "loss": 0.1041,
-      "step": 4742
-    },
-    {
-      "epoch": 0.8086741082132296,
-      "grad_norm": 1.1588139533996582,
-      "learning_rate": 1.9130434782608697e-06,
-      "loss": 0.1045,
-      "step": 4743
-    },
-    {
-      "epoch": 0.8088446066547673,
-      "grad_norm": 0.8025126457214355,
-      "learning_rate": 1.9113384484228477e-06,
-      "loss": 0.0936,
-      "step": 4744
-    },
-    {
-      "epoch": 0.809015105096305,
-      "grad_norm": 1.339248538017273,
-      "learning_rate": 1.909633418584825e-06,
-      "loss": 0.1248,
-      "step": 4745
-    },
-    {
-      "epoch": 0.8091856035378426,
-      "grad_norm": 0.8242576122283936,
-      "learning_rate": 1.907928388746803e-06,
-      "loss": 0.0414,
-      "step": 4746
-    },
-    {
-      "epoch": 0.8093561019793803,
-      "grad_norm": 1.1313693523406982,
-      "learning_rate": 1.906223358908781e-06,
-      "loss": 0.1032,
-      "step": 4747
-    },
-    {
-      "epoch": 0.8095266004209181,
-      "grad_norm": 1.785373568534851,
-      "learning_rate": 1.9045183290707588e-06,
-      "loss": 0.1188,
-      "step": 4748
-    },
-    {
-      "epoch": 0.8096970988624557,
-      "grad_norm": 1.2123826742172241,
-      "learning_rate": 1.9028132992327366e-06,
-      "loss": 0.1482,
-      "step": 4749
-    },
-    {
-      "epoch": 0.8098675973039934,
-      "grad_norm": 2.0810234546661377,
-      "learning_rate": 1.9011082693947145e-06,
-      "loss": 0.1969,
-      "step": 4750
-    },
-    {
-      "epoch": 0.8100380957455311,
-      "grad_norm": 1.8342708349227905,
-      "learning_rate": 1.8994032395566925e-06,
-      "loss": 0.1687,
-      "step": 4751
-    },
-    {
-      "epoch": 0.8102085941870688,
-      "grad_norm": 1.2007415294647217,
-      "learning_rate": 1.8976982097186702e-06,
-      "loss": 0.1193,
-      "step": 4752
-    },
-    {
-      "epoch": 0.8103790926286064,
-      "grad_norm": 1.5728555917739868,
-      "learning_rate": 1.895993179880648e-06,
-      "loss": 0.1214,
-      "step": 4753
-    },
-    {
-      "epoch": 0.8105495910701441,
-      "grad_norm": 2.4741835594177246,
-      "learning_rate": 1.8942881500426257e-06,
-      "loss": 0.1793,
-      "step": 4754
-    },
-    {
-      "epoch": 0.8107200895116818,
-      "grad_norm": 0.9598491191864014,
-      "learning_rate": 1.8925831202046038e-06,
-      "loss": 0.0617,
-      "step": 4755
-    },
-    {
-      "epoch": 0.8108905879532194,
-      "grad_norm": 1.4126801490783691,
-      "learning_rate": 1.8908780903665816e-06,
-      "loss": 0.1332,
-      "step": 4756
-    },
-    {
-      "epoch": 0.8110610863947572,
-      "grad_norm": 1.359977126121521,
-      "learning_rate": 1.8891730605285593e-06,
-      "loss": 0.1188,
-      "step": 4757
-    },
-    {
-      "epoch": 0.8112315848362949,
-      "grad_norm": 0.970696747303009,
-      "learning_rate": 1.887468030690537e-06,
-      "loss": 0.0669,
-      "step": 4758
-    },
-    {
-      "epoch": 0.8114020832778326,
-      "grad_norm": 1.1032432317733765,
-      "learning_rate": 1.8857630008525152e-06,
-      "loss": 0.0869,
-      "step": 4759
-    },
-    {
-      "epoch": 0.8115725817193702,
-      "grad_norm": 1.0205268859863281,
-      "learning_rate": 1.884057971014493e-06,
-      "loss": 0.0861,
-      "step": 4760
-    },
-    {
-      "epoch": 0.8117430801609079,
-      "grad_norm": 1.1021018028259277,
-      "learning_rate": 1.8823529411764707e-06,
-      "loss": 0.1098,
-      "step": 4761
-    },
-    {
-      "epoch": 0.8119135786024456,
-      "grad_norm": 1.8316097259521484,
-      "learning_rate": 1.8806479113384484e-06,
-      "loss": 0.1278,
-      "step": 4762
-    },
-    {
-      "epoch": 0.8120840770439832,
-      "grad_norm": 1.722632884979248,
-      "learning_rate": 1.8789428815004266e-06,
-      "loss": 0.1295,
-      "step": 4763
-    },
-    {
-      "epoch": 0.8122545754855209,
-      "grad_norm": 1.2179450988769531,
-      "learning_rate": 1.8772378516624043e-06,
-      "loss": 0.0912,
-      "step": 4764
-    },
-    {
-      "epoch": 0.8124250739270586,
-      "grad_norm": 1.0734045505523682,
-      "learning_rate": 1.875532821824382e-06,
-      "loss": 0.1192,
-      "step": 4765
-    },
-    {
-      "epoch": 0.8125955723685964,
-      "grad_norm": 1.4942569732666016,
-      "learning_rate": 1.8738277919863598e-06,
-      "loss": 0.1581,
-      "step": 4766
-    },
-    {
-      "epoch": 0.812766070810134,
-      "grad_norm": 1.9975972175598145,
-      "learning_rate": 1.872122762148338e-06,
-      "loss": 0.1512,
-      "step": 4767
-    },
-    {
-      "epoch": 0.8129365692516717,
-      "grad_norm": 0.8487547039985657,
-      "learning_rate": 1.8704177323103157e-06,
-      "loss": 0.0683,
-      "step": 4768
-    },
-    {
-      "epoch": 0.8131070676932094,
-      "grad_norm": 1.2102348804473877,
-      "learning_rate": 1.8687127024722934e-06,
-      "loss": 0.0873,
-      "step": 4769
-    },
-    {
-      "epoch": 0.813277566134747,
-      "grad_norm": 1.8456752300262451,
-      "learning_rate": 1.8670076726342711e-06,
-      "loss": 0.1804,
-      "step": 4770
-    },
-    {
-      "epoch": 0.8134480645762847,
-      "grad_norm": 1.389062523841858,
-      "learning_rate": 1.8653026427962489e-06,
-      "loss": 0.1356,
-      "step": 4771
-    },
-    {
-      "epoch": 0.8136185630178224,
-      "grad_norm": 0.8592960834503174,
-      "learning_rate": 1.863597612958227e-06,
-      "loss": 0.0853,
-      "step": 4772
-    },
-    {
-      "epoch": 0.8137890614593601,
-      "grad_norm": 1.0502662658691406,
-      "learning_rate": 1.8618925831202048e-06,
-      "loss": 0.0735,
-      "step": 4773
-    },
-    {
-      "epoch": 0.8139595599008977,
-      "grad_norm": 1.4128313064575195,
-      "learning_rate": 1.8601875532821825e-06,
-      "loss": 0.0853,
-      "step": 4774
-    },
-    {
-      "epoch": 0.8141300583424355,
-      "grad_norm": 1.5370444059371948,
-      "learning_rate": 1.8584825234441603e-06,
-      "loss": 0.1043,
-      "step": 4775
-    },
-    {
-      "epoch": 0.8143005567839732,
-      "grad_norm": 1.4921493530273438,
-      "learning_rate": 1.8567774936061384e-06,
-      "loss": 0.1193,
-      "step": 4776
-    },
-    {
-      "epoch": 0.8144710552255108,
-      "grad_norm": 1.5475974082946777,
-      "learning_rate": 1.8550724637681161e-06,
-      "loss": 0.1669,
-      "step": 4777
-    },
-    {
-      "epoch": 0.8146415536670485,
-      "grad_norm": 2.721313953399658,
-      "learning_rate": 1.8533674339300939e-06,
-      "loss": 0.2292,
-      "step": 4778
-    },
-    {
-      "epoch": 0.8148120521085862,
-      "grad_norm": 2.104518175125122,
-      "learning_rate": 1.8516624040920716e-06,
-      "loss": 0.1861,
-      "step": 4779
-    },
-    {
-      "epoch": 0.8149825505501239,
-      "grad_norm": 1.5531257390975952,
-      "learning_rate": 1.8499573742540498e-06,
-      "loss": 0.1452,
-      "step": 4780
-    },
-    {
-      "epoch": 0.8151530489916615,
-      "grad_norm": 0.8269780874252319,
-      "learning_rate": 1.8482523444160275e-06,
-      "loss": 0.0698,
-      "step": 4781
-    },
-    {
-      "epoch": 0.8153235474331992,
-      "grad_norm": 1.4180121421813965,
-      "learning_rate": 1.8465473145780052e-06,
-      "loss": 0.1433,
-      "step": 4782
-    },
-    {
-      "epoch": 0.815494045874737,
-      "grad_norm": 1.6181259155273438,
-      "learning_rate": 1.844842284739983e-06,
-      "loss": 0.1311,
-      "step": 4783
-    },
-    {
-      "epoch": 0.8156645443162747,
-      "grad_norm": 1.1741197109222412,
-      "learning_rate": 1.8431372549019611e-06,
-      "loss": 0.0883,
-      "step": 4784
-    },
-    {
-      "epoch": 0.8158350427578123,
-      "grad_norm": 2.6572444438934326,
-      "learning_rate": 1.8414322250639389e-06,
-      "loss": 0.1646,
-      "step": 4785
-    },
-    {
-      "epoch": 0.81600554119935,
-      "grad_norm": 1.264453649520874,
-      "learning_rate": 1.8397271952259166e-06,
-      "loss": 0.0843,
-      "step": 4786
-    },
-    {
-      "epoch": 0.8161760396408877,
-      "grad_norm": 1.3252006769180298,
-      "learning_rate": 1.8380221653878944e-06,
-      "loss": 0.1508,
-      "step": 4787
-    },
-    {
-      "epoch": 0.8163465380824253,
-      "grad_norm": 1.1810276508331299,
-      "learning_rate": 1.8363171355498723e-06,
-      "loss": 0.0991,
-      "step": 4788
-    },
-    {
-      "epoch": 0.816517036523963,
-      "grad_norm": 1.8916993141174316,
-      "learning_rate": 1.8346121057118502e-06,
-      "loss": 0.1708,
-      "step": 4789
-    },
-    {
-      "epoch": 0.8166875349655007,
-      "grad_norm": 2.088521957397461,
-      "learning_rate": 1.832907075873828e-06,
-      "loss": 0.2239,
-      "step": 4790
-    },
-    {
-      "epoch": 0.8168580334070384,
-      "grad_norm": 1.321691870689392,
-      "learning_rate": 1.8312020460358057e-06,
-      "loss": 0.1131,
-      "step": 4791
-    },
-    {
-      "epoch": 0.817028531848576,
-      "grad_norm": 1.0528753995895386,
-      "learning_rate": 1.8294970161977837e-06,
-      "loss": 0.1082,
-      "step": 4792
-    },
-    {
-      "epoch": 0.8171990302901138,
-      "grad_norm": 1.9226561784744263,
-      "learning_rate": 1.8277919863597616e-06,
-      "loss": 0.1226,
-      "step": 4793
-    },
-    {
-      "epoch": 0.8173695287316515,
-      "grad_norm": 1.830688238143921,
-      "learning_rate": 1.8260869565217394e-06,
-      "loss": 0.1374,
-      "step": 4794
-    },
-    {
-      "epoch": 0.8175400271731891,
-      "grad_norm": 1.4284980297088623,
-      "learning_rate": 1.824381926683717e-06,
-      "loss": 0.1419,
-      "step": 4795
-    },
-    {
-      "epoch": 0.8177105256147268,
-      "grad_norm": 0.8486391305923462,
-      "learning_rate": 1.8226768968456948e-06,
-      "loss": 0.0416,
-      "step": 4796
-    },
-    {
-      "epoch": 0.8178810240562645,
-      "grad_norm": 1.925445795059204,
-      "learning_rate": 1.8209718670076728e-06,
-      "loss": 0.184,
-      "step": 4797
-    },
-    {
-      "epoch": 0.8180515224978022,
-      "grad_norm": 1.1646342277526855,
-      "learning_rate": 1.8192668371696507e-06,
-      "loss": 0.1031,
-      "step": 4798
-    },
-    {
-      "epoch": 0.8182220209393398,
-      "grad_norm": 1.6515798568725586,
-      "learning_rate": 1.8175618073316285e-06,
-      "loss": 0.1654,
-      "step": 4799
-    },
-    {
-      "epoch": 0.8183925193808775,
-      "grad_norm": 1.2121893167495728,
-      "learning_rate": 1.8158567774936062e-06,
-      "loss": 0.0971,
-      "step": 4800
-    },
-    {
-      "epoch": 0.8183925193808775,
-      "eval_f1_score": 0.42,
-      "eval_loss": 0.13059745728969574,
-      "eval_runtime": 183.3571,
-      "eval_samples_per_second": 54.538,
-      "eval_steps_per_second": 3.409,
-      "step": 4800
-    },
-    {
-      "epoch": 0.8185630178224153,
-      "grad_norm": 1.1018650531768799,
-      "learning_rate": 1.8141517476555841e-06,
-      "loss": 0.0962,
-      "step": 4801
-    },
-    {
-      "epoch": 0.8187335162639529,
-      "grad_norm": 1.2527377605438232,
-      "learning_rate": 1.812446717817562e-06,
-      "loss": 0.1066,
-      "step": 4802
-    },
-    {
-      "epoch": 0.8189040147054906,
-      "grad_norm": 1.6081165075302124,
-      "learning_rate": 1.8107416879795398e-06,
-      "loss": 0.1714,
-      "step": 4803
-    },
-    {
-      "epoch": 0.8190745131470283,
-      "grad_norm": 1.5928921699523926,
-      "learning_rate": 1.8090366581415176e-06,
-      "loss": 0.0999,
-      "step": 4804
-    },
-    {
-      "epoch": 0.819245011588566,
-      "grad_norm": 1.7792342901229858,
-      "learning_rate": 1.8073316283034955e-06,
-      "loss": 0.0932,
-      "step": 4805
-    },
-    {
-      "epoch": 0.8194155100301036,
-      "grad_norm": 1.6293425559997559,
-      "learning_rate": 1.8056265984654732e-06,
-      "loss": 0.1317,
-      "step": 4806
-    },
-    {
-      "epoch": 0.8195860084716413,
-      "grad_norm": 1.1104508638381958,
-      "learning_rate": 1.8039215686274512e-06,
-      "loss": 0.1182,
-      "step": 4807
-    },
-    {
-      "epoch": 0.819756506913179,
-      "grad_norm": 1.0992122888565063,
-      "learning_rate": 1.802216538789429e-06,
-      "loss": 0.1093,
-      "step": 4808
-    },
-    {
-      "epoch": 0.8199270053547166,
-      "grad_norm": 1.6125050783157349,
-      "learning_rate": 1.8005115089514069e-06,
-      "loss": 0.1031,
-      "step": 4809
-    },
-    {
-      "epoch": 0.8200975037962543,
-      "grad_norm": 0.8295161724090576,
-      "learning_rate": 1.7988064791133846e-06,
-      "loss": 0.0621,
-      "step": 4810
-    },
-    {
-      "epoch": 0.8202680022377921,
-      "grad_norm": 1.5850706100463867,
-      "learning_rate": 1.7971014492753626e-06,
-      "loss": 0.1237,
-      "step": 4811
-    },
-    {
-      "epoch": 0.8204385006793298,
-      "grad_norm": 1.3965903520584106,
-      "learning_rate": 1.7953964194373403e-06,
-      "loss": 0.1801,
-      "step": 4812
-    },
-    {
-      "epoch": 0.8206089991208674,
-      "grad_norm": 1.6549614667892456,
-      "learning_rate": 1.7936913895993182e-06,
-      "loss": 0.1196,
-      "step": 4813
-    },
-    {
-      "epoch": 0.8207794975624051,
-      "grad_norm": 0.9478707909584045,
-      "learning_rate": 1.791986359761296e-06,
-      "loss": 0.0902,
-      "step": 4814
-    },
-    {
-      "epoch": 0.8209499960039428,
-      "grad_norm": 1.6688967943191528,
-      "learning_rate": 1.7902813299232737e-06,
-      "loss": 0.1263,
-      "step": 4815
-    },
-    {
-      "epoch": 0.8211204944454804,
-      "grad_norm": 1.9953707456588745,
-      "learning_rate": 1.7885763000852517e-06,
-      "loss": 0.1458,
-      "step": 4816
-    },
-    {
-      "epoch": 0.8212909928870181,
-      "grad_norm": 1.5312587022781372,
-      "learning_rate": 1.7868712702472294e-06,
-      "loss": 0.1163,
-      "step": 4817
-    },
-    {
-      "epoch": 0.8214614913285558,
-      "grad_norm": 1.563798189163208,
-      "learning_rate": 1.7851662404092073e-06,
-      "loss": 0.1001,
-      "step": 4818
-    },
-    {
-      "epoch": 0.8216319897700936,
-      "grad_norm": 1.5185989141464233,
-      "learning_rate": 1.783461210571185e-06,
-      "loss": 0.141,
-      "step": 4819
-    },
-    {
-      "epoch": 0.8218024882116312,
-      "grad_norm": 1.8926496505737305,
-      "learning_rate": 1.7817561807331628e-06,
-      "loss": 0.2122,
-      "step": 4820
-    },
-    {
-      "epoch": 0.8219729866531689,
-      "grad_norm": 1.165590763092041,
-      "learning_rate": 1.7800511508951408e-06,
-      "loss": 0.1083,
-      "step": 4821
-    },
-    {
-      "epoch": 0.8221434850947066,
-      "grad_norm": 1.5912805795669556,
-      "learning_rate": 1.7783461210571187e-06,
-      "loss": 0.1377,
-      "step": 4822
-    },
-    {
-      "epoch": 0.8223139835362442,
-      "grad_norm": 2.399815082550049,
-      "learning_rate": 1.7766410912190964e-06,
-      "loss": 0.2404,
-      "step": 4823
-    },
-    {
-      "epoch": 0.8224844819777819,
-      "grad_norm": 1.1800510883331299,
-      "learning_rate": 1.7749360613810742e-06,
-      "loss": 0.0652,
-      "step": 4824
-    },
-    {
-      "epoch": 0.8226549804193196,
-      "grad_norm": 1.0360442399978638,
-      "learning_rate": 1.7732310315430521e-06,
-      "loss": 0.0917,
-      "step": 4825
-    },
-    {
-      "epoch": 0.8228254788608573,
-      "grad_norm": 2.043694257736206,
-      "learning_rate": 1.77152600170503e-06,
-      "loss": 0.1947,
-      "step": 4826
-    },
-    {
-      "epoch": 0.8229959773023949,
-      "grad_norm": 1.7305378913879395,
-      "learning_rate": 1.7698209718670078e-06,
-      "loss": 0.1791,
-      "step": 4827
-    },
-    {
-      "epoch": 0.8231664757439326,
-      "grad_norm": 1.4891148805618286,
-      "learning_rate": 1.7681159420289855e-06,
-      "loss": 0.111,
-      "step": 4828
-    },
-    {
-      "epoch": 0.8233369741854704,
-      "grad_norm": 1.1548901796340942,
-      "learning_rate": 1.7664109121909633e-06,
-      "loss": 0.115,
-      "step": 4829
-    },
-    {
-      "epoch": 0.823507472627008,
-      "grad_norm": 1.0407037734985352,
-      "learning_rate": 1.7647058823529414e-06,
-      "loss": 0.0966,
-      "step": 4830
-    },
-    {
-      "epoch": 0.8236779710685457,
-      "grad_norm": 1.8325929641723633,
-      "learning_rate": 1.7630008525149192e-06,
-      "loss": 0.1428,
-      "step": 4831
-    },
-    {
-      "epoch": 0.8238484695100834,
-      "grad_norm": 1.2216639518737793,
-      "learning_rate": 1.761295822676897e-06,
-      "loss": 0.1084,
-      "step": 4832
-    },
-    {
-      "epoch": 0.8240189679516211,
-      "grad_norm": 1.1613625288009644,
-      "learning_rate": 1.7595907928388747e-06,
-      "loss": 0.1026,
-      "step": 4833
-    },
-    {
-      "epoch": 0.8241894663931587,
-      "grad_norm": 1.3345240354537964,
-      "learning_rate": 1.7578857630008528e-06,
-      "loss": 0.0683,
-      "step": 4834
-    },
-    {
-      "epoch": 0.8243599648346964,
-      "grad_norm": 1.459480881690979,
-      "learning_rate": 1.7561807331628305e-06,
-      "loss": 0.1128,
-      "step": 4835
-    },
-    {
-      "epoch": 0.8245304632762341,
-      "grad_norm": 1.513303518295288,
-      "learning_rate": 1.7544757033248083e-06,
-      "loss": 0.1847,
-      "step": 4836
-    },
-    {
-      "epoch": 0.8247009617177719,
-      "grad_norm": 1.2072933912277222,
-      "learning_rate": 1.752770673486786e-06,
-      "loss": 0.0592,
-      "step": 4837
-    },
-    {
-      "epoch": 0.8248714601593095,
-      "grad_norm": 1.6384389400482178,
-      "learning_rate": 1.7510656436487642e-06,
-      "loss": 0.1571,
-      "step": 4838
-    },
-    {
-      "epoch": 0.8250419586008472,
-      "grad_norm": 0.8648350834846497,
-      "learning_rate": 1.749360613810742e-06,
-      "loss": 0.0887,
-      "step": 4839
-    },
-    {
-      "epoch": 0.8252124570423849,
-      "grad_norm": 2.6184184551239014,
-      "learning_rate": 1.7476555839727197e-06,
-      "loss": 0.2689,
-      "step": 4840
-    },
-    {
-      "epoch": 0.8253829554839225,
-      "grad_norm": 1.637653112411499,
-      "learning_rate": 1.7459505541346974e-06,
-      "loss": 0.0831,
-      "step": 4841
-    },
-    {
-      "epoch": 0.8255534539254602,
-      "grad_norm": 0.9104223847389221,
-      "learning_rate": 1.7442455242966751e-06,
-      "loss": 0.0679,
-      "step": 4842
-    },
-    {
-      "epoch": 0.8257239523669979,
-      "grad_norm": 1.3634427785873413,
-      "learning_rate": 1.7425404944586533e-06,
-      "loss": 0.1786,
-      "step": 4843
-    },
-    {
-      "epoch": 0.8258944508085356,
-      "grad_norm": 2.4873688220977783,
-      "learning_rate": 1.740835464620631e-06,
-      "loss": 0.1614,
-      "step": 4844
-    },
-    {
-      "epoch": 0.8260649492500732,
-      "grad_norm": 1.1272398233413696,
-      "learning_rate": 1.7391304347826088e-06,
-      "loss": 0.0912,
-      "step": 4845
-    },
-    {
-      "epoch": 0.826235447691611,
-      "grad_norm": 1.254799246788025,
-      "learning_rate": 1.7374254049445865e-06,
-      "loss": 0.1297,
-      "step": 4846
-    },
-    {
-      "epoch": 0.8264059461331487,
-      "grad_norm": 1.8196457624435425,
-      "learning_rate": 1.7357203751065647e-06,
-      "loss": 0.1737,
-      "step": 4847
-    },
-    {
-      "epoch": 0.8265764445746863,
-      "grad_norm": 0.8753282427787781,
-      "learning_rate": 1.7340153452685424e-06,
-      "loss": 0.0688,
-      "step": 4848
-    },
-    {
-      "epoch": 0.826746943016224,
-      "grad_norm": 1.0449436902999878,
-      "learning_rate": 1.7323103154305201e-06,
-      "loss": 0.0753,
-      "step": 4849
-    },
-    {
-      "epoch": 0.8269174414577617,
-      "grad_norm": 1.1351038217544556,
-      "learning_rate": 1.7306052855924979e-06,
-      "loss": 0.101,
-      "step": 4850
-    },
-    {
-      "epoch": 0.8270879398992994,
-      "grad_norm": 1.4630578756332397,
-      "learning_rate": 1.728900255754476e-06,
-      "loss": 0.1175,
-      "step": 4851
-    },
-    {
-      "epoch": 0.827258438340837,
-      "grad_norm": 1.8749983310699463,
-      "learning_rate": 1.7271952259164538e-06,
-      "loss": 0.148,
-      "step": 4852
-    },
-    {
-      "epoch": 0.8274289367823747,
-      "grad_norm": 1.2372411489486694,
-      "learning_rate": 1.7254901960784315e-06,
-      "loss": 0.0844,
-      "step": 4853
-    },
-    {
-      "epoch": 0.8275994352239124,
-      "grad_norm": 1.1746691465377808,
-      "learning_rate": 1.7237851662404092e-06,
-      "loss": 0.0951,
-      "step": 4854
-    },
-    {
-      "epoch": 0.82776993366545,
-      "grad_norm": 0.9853083491325378,
-      "learning_rate": 1.7220801364023874e-06,
-      "loss": 0.0903,
-      "step": 4855
-    },
-    {
-      "epoch": 0.8279404321069878,
-      "grad_norm": 2.1901590824127197,
-      "learning_rate": 1.7203751065643651e-06,
-      "loss": 0.1728,
-      "step": 4856
-    },
-    {
-      "epoch": 0.8281109305485255,
-      "grad_norm": 1.0862927436828613,
-      "learning_rate": 1.7186700767263429e-06,
-      "loss": 0.1114,
-      "step": 4857
-    },
-    {
-      "epoch": 0.8282814289900632,
-      "grad_norm": 1.617577075958252,
-      "learning_rate": 1.7169650468883206e-06,
-      "loss": 0.1074,
-      "step": 4858
-    },
-    {
-      "epoch": 0.8284519274316008,
-      "grad_norm": 1.824747085571289,
-      "learning_rate": 1.7152600170502988e-06,
-      "loss": 0.081,
-      "step": 4859
-    },
-    {
-      "epoch": 0.8286224258731385,
-      "grad_norm": 1.7742489576339722,
-      "learning_rate": 1.7135549872122765e-06,
-      "loss": 0.1525,
-      "step": 4860
-    },
-    {
-      "epoch": 0.8287929243146762,
-      "grad_norm": 2.076169967651367,
-      "learning_rate": 1.7118499573742542e-06,
-      "loss": 0.118,
-      "step": 4861
-    },
-    {
-      "epoch": 0.8289634227562138,
-      "grad_norm": 1.0002825260162354,
-      "learning_rate": 1.710144927536232e-06,
-      "loss": 0.0538,
-      "step": 4862
-    },
-    {
-      "epoch": 0.8291339211977515,
-      "grad_norm": 1.6570539474487305,
-      "learning_rate": 1.7084398976982097e-06,
-      "loss": 0.1468,
-      "step": 4863
-    },
-    {
-      "epoch": 0.8293044196392892,
-      "grad_norm": 1.037410020828247,
-      "learning_rate": 1.7067348678601879e-06,
-      "loss": 0.1069,
-      "step": 4864
-    },
-    {
-      "epoch": 0.829474918080827,
-      "grad_norm": 1.7056080102920532,
-      "learning_rate": 1.7050298380221656e-06,
-      "loss": 0.1525,
-      "step": 4865
-    },
-    {
-      "epoch": 0.8296454165223646,
-      "grad_norm": 1.6007308959960938,
-      "learning_rate": 1.7033248081841433e-06,
-      "loss": 0.132,
-      "step": 4866
-    },
-    {
-      "epoch": 0.8298159149639023,
-      "grad_norm": 1.514782190322876,
-      "learning_rate": 1.701619778346121e-06,
-      "loss": 0.1442,
-      "step": 4867
-    },
-    {
-      "epoch": 0.82998641340544,
-      "grad_norm": 0.7461420297622681,
-      "learning_rate": 1.6999147485080992e-06,
-      "loss": 0.0602,
-      "step": 4868
-    },
-    {
-      "epoch": 0.8301569118469776,
-      "grad_norm": 1.7397735118865967,
-      "learning_rate": 1.698209718670077e-06,
-      "loss": 0.1756,
-      "step": 4869
-    },
-    {
-      "epoch": 0.8303274102885153,
-      "grad_norm": 1.6185688972473145,
-      "learning_rate": 1.6965046888320547e-06,
-      "loss": 0.1193,
-      "step": 4870
-    },
-    {
-      "epoch": 0.830497908730053,
-      "grad_norm": 1.5054775476455688,
-      "learning_rate": 1.6947996589940324e-06,
-      "loss": 0.1949,
-      "step": 4871
-    },
-    {
-      "epoch": 0.8306684071715907,
-      "grad_norm": 1.8738244771957397,
-      "learning_rate": 1.6930946291560104e-06,
-      "loss": 0.0839,
-      "step": 4872
-    },
-    {
-      "epoch": 0.8308389056131283,
-      "grad_norm": 1.9042476415634155,
-      "learning_rate": 1.6913895993179883e-06,
-      "loss": 0.1257,
-      "step": 4873
-    },
-    {
-      "epoch": 0.8310094040546661,
-      "grad_norm": 1.4473137855529785,
-      "learning_rate": 1.689684569479966e-06,
-      "loss": 0.1606,
-      "step": 4874
-    },
-    {
-      "epoch": 0.8311799024962038,
-      "grad_norm": 1.8061612844467163,
-      "learning_rate": 1.6879795396419438e-06,
-      "loss": 0.1655,
-      "step": 4875
-    },
-    {
-      "epoch": 0.8313504009377414,
-      "grad_norm": 1.2056777477264404,
-      "learning_rate": 1.6862745098039217e-06,
-      "loss": 0.095,
-      "step": 4876
-    },
-    {
-      "epoch": 0.8315208993792791,
-      "grad_norm": 2.325047254562378,
-      "learning_rate": 1.6845694799658997e-06,
-      "loss": 0.1422,
-      "step": 4877
-    },
-    {
-      "epoch": 0.8316913978208168,
-      "grad_norm": 1.2801249027252197,
-      "learning_rate": 1.6828644501278774e-06,
-      "loss": 0.1503,
-      "step": 4878
-    },
-    {
-      "epoch": 0.8318618962623545,
-      "grad_norm": 2.1016643047332764,
-      "learning_rate": 1.6811594202898552e-06,
-      "loss": 0.1834,
-      "step": 4879
-    },
-    {
-      "epoch": 0.8320323947038921,
-      "grad_norm": 1.1316357851028442,
-      "learning_rate": 1.6794543904518331e-06,
-      "loss": 0.1201,
-      "step": 4880
-    },
-    {
-      "epoch": 0.8322028931454298,
-      "grad_norm": 1.2594853639602661,
-      "learning_rate": 1.6777493606138108e-06,
-      "loss": 0.1242,
-      "step": 4881
-    },
-    {
-      "epoch": 0.8323733915869675,
-      "grad_norm": 1.3186732530593872,
-      "learning_rate": 1.6760443307757888e-06,
-      "loss": 0.1331,
-      "step": 4882
-    },
-    {
-      "epoch": 0.8325438900285053,
-      "grad_norm": 1.2607057094573975,
-      "learning_rate": 1.6743393009377665e-06,
-      "loss": 0.1387,
-      "step": 4883
-    },
-    {
-      "epoch": 0.8327143884700429,
-      "grad_norm": 1.6016281843185425,
-      "learning_rate": 1.6726342710997443e-06,
-      "loss": 0.1603,
-      "step": 4884
-    },
-    {
-      "epoch": 0.8328848869115806,
-      "grad_norm": 1.093407154083252,
-      "learning_rate": 1.6709292412617222e-06,
-      "loss": 0.0765,
-      "step": 4885
-    },
-    {
-      "epoch": 0.8330553853531183,
-      "grad_norm": 1.4803746938705444,
-      "learning_rate": 1.6692242114237002e-06,
-      "loss": 0.1681,
-      "step": 4886
-    },
-    {
-      "epoch": 0.8332258837946559,
-      "grad_norm": 1.254752516746521,
-      "learning_rate": 1.667519181585678e-06,
-      "loss": 0.1092,
-      "step": 4887
-    },
-    {
-      "epoch": 0.8333963822361936,
-      "grad_norm": 1.5853029489517212,
-      "learning_rate": 1.6658141517476556e-06,
-      "loss": 0.1387,
-      "step": 4888
-    },
-    {
-      "epoch": 0.8335668806777313,
-      "grad_norm": 1.0304608345031738,
-      "learning_rate": 1.6641091219096336e-06,
-      "loss": 0.0431,
-      "step": 4889
-    },
-    {
-      "epoch": 0.833737379119269,
-      "grad_norm": 1.44893479347229,
-      "learning_rate": 1.6624040920716113e-06,
-      "loss": 0.115,
-      "step": 4890
-    },
-    {
-      "epoch": 0.8339078775608066,
-      "grad_norm": 0.9381269216537476,
-      "learning_rate": 1.6606990622335893e-06,
-      "loss": 0.0736,
-      "step": 4891
-    },
-    {
-      "epoch": 0.8340783760023444,
-      "grad_norm": 1.6925040483474731,
-      "learning_rate": 1.658994032395567e-06,
-      "loss": 0.1446,
-      "step": 4892
-    },
-    {
-      "epoch": 0.8342488744438821,
-      "grad_norm": 1.8539586067199707,
-      "learning_rate": 1.657289002557545e-06,
-      "loss": 0.1497,
-      "step": 4893
-    },
-    {
-      "epoch": 0.8344193728854197,
-      "grad_norm": 1.285607099533081,
-      "learning_rate": 1.6555839727195227e-06,
-      "loss": 0.1212,
-      "step": 4894
-    },
-    {
-      "epoch": 0.8345898713269574,
-      "grad_norm": 1.8159408569335938,
-      "learning_rate": 1.6538789428815004e-06,
-      "loss": 0.1543,
-      "step": 4895
-    },
-    {
-      "epoch": 0.8347603697684951,
-      "grad_norm": 1.9982542991638184,
-      "learning_rate": 1.6521739130434784e-06,
-      "loss": 0.1245,
-      "step": 4896
-    },
-    {
-      "epoch": 0.8349308682100328,
-      "grad_norm": 1.0288341045379639,
-      "learning_rate": 1.6504688832054563e-06,
-      "loss": 0.1175,
-      "step": 4897
-    },
-    {
-      "epoch": 0.8351013666515704,
-      "grad_norm": 1.4282264709472656,
-      "learning_rate": 1.648763853367434e-06,
-      "loss": 0.0739,
-      "step": 4898
-    },
-    {
-      "epoch": 0.8352718650931081,
-      "grad_norm": 1.336698293685913,
-      "learning_rate": 1.6470588235294118e-06,
-      "loss": 0.1349,
-      "step": 4899
-    },
-    {
-      "epoch": 0.8354423635346458,
-      "grad_norm": 1.2105053663253784,
-      "learning_rate": 1.6453537936913897e-06,
-      "loss": 0.1025,
-      "step": 4900
-    },
-    {
-      "epoch": 0.8354423635346458,
-      "eval_f1_score": 0.4364508393285372,
-      "eval_loss": 0.1303476244211197,
-      "eval_runtime": 183.3875,
-      "eval_samples_per_second": 54.529,
-      "eval_steps_per_second": 3.408,
-      "step": 4900
-    },
-    {
-      "epoch": 0.8356128619761835,
-      "grad_norm": 0.8056730031967163,
-      "learning_rate": 1.6436487638533677e-06,
-      "loss": 0.0576,
-      "step": 4901
-    },
-    {
-      "epoch": 0.8357833604177212,
-      "grad_norm": 1.9519644975662231,
-      "learning_rate": 1.6419437340153454e-06,
-      "loss": 0.2069,
-      "step": 4902
-    },
-    {
-      "epoch": 0.8359538588592589,
-      "grad_norm": 1.4039374589920044,
-      "learning_rate": 1.6402387041773232e-06,
-      "loss": 0.1316,
-      "step": 4903
-    },
-    {
-      "epoch": 0.8361243573007966,
-      "grad_norm": 1.4302421808242798,
-      "learning_rate": 1.6385336743393009e-06,
-      "loss": 0.1208,
-      "step": 4904
-    },
-    {
-      "epoch": 0.8362948557423342,
-      "grad_norm": 3.204106569290161,
-      "learning_rate": 1.636828644501279e-06,
-      "loss": 0.0977,
-      "step": 4905
-    },
-    {
-      "epoch": 0.8364653541838719,
-      "grad_norm": 0.9173206686973572,
-      "learning_rate": 1.6351236146632568e-06,
-      "loss": 0.0584,
-      "step": 4906
-    },
-    {
-      "epoch": 0.8366358526254096,
-      "grad_norm": 0.9852831363677979,
-      "learning_rate": 1.6334185848252345e-06,
-      "loss": 0.1155,
-      "step": 4907
-    },
-    {
-      "epoch": 0.8368063510669472,
-      "grad_norm": 1.2431775331497192,
-      "learning_rate": 1.6317135549872123e-06,
-      "loss": 0.1246,
-      "step": 4908
-    },
-    {
-      "epoch": 0.8369768495084849,
-      "grad_norm": 1.100155234336853,
-      "learning_rate": 1.6300085251491902e-06,
-      "loss": 0.0895,
-      "step": 4909
-    },
-    {
-      "epoch": 0.8371473479500227,
-      "grad_norm": 1.8383277654647827,
-      "learning_rate": 1.6283034953111682e-06,
-      "loss": 0.1726,
-      "step": 4910
-    },
-    {
-      "epoch": 0.8373178463915604,
-      "grad_norm": 1.2940951585769653,
-      "learning_rate": 1.6265984654731459e-06,
-      "loss": 0.1096,
-      "step": 4911
-    },
-    {
-      "epoch": 0.837488344833098,
-      "grad_norm": 1.2588635683059692,
-      "learning_rate": 1.6248934356351236e-06,
-      "loss": 0.0808,
-      "step": 4912
-    },
-    {
-      "epoch": 0.8376588432746357,
-      "grad_norm": 2.5419719219207764,
-      "learning_rate": 1.6231884057971014e-06,
-      "loss": 0.1538,
-      "step": 4913
-    },
-    {
-      "epoch": 0.8378293417161734,
-      "grad_norm": 1.7912168502807617,
-      "learning_rate": 1.6214833759590795e-06,
-      "loss": 0.1429,
-      "step": 4914
-    },
-    {
-      "epoch": 0.837999840157711,
-      "grad_norm": 1.7084680795669556,
-      "learning_rate": 1.6197783461210573e-06,
-      "loss": 0.1547,
-      "step": 4915
-    },
-    {
-      "epoch": 0.8381703385992487,
-      "grad_norm": 1.299729824066162,
-      "learning_rate": 1.618073316283035e-06,
-      "loss": 0.0982,
-      "step": 4916
-    },
-    {
-      "epoch": 0.8383408370407864,
-      "grad_norm": 1.2041939496994019,
-      "learning_rate": 1.6163682864450127e-06,
-      "loss": 0.1329,
-      "step": 4917
-    },
-    {
-      "epoch": 0.8385113354823241,
-      "grad_norm": 1.9802684783935547,
-      "learning_rate": 1.6146632566069909e-06,
-      "loss": 0.1781,
-      "step": 4918
-    },
-    {
-      "epoch": 0.8386818339238618,
-      "grad_norm": 1.0352423191070557,
-      "learning_rate": 1.6129582267689686e-06,
-      "loss": 0.0929,
-      "step": 4919
-    },
-    {
-      "epoch": 0.8388523323653995,
-      "grad_norm": 1.1618430614471436,
-      "learning_rate": 1.6112531969309464e-06,
-      "loss": 0.078,
-      "step": 4920
-    },
-    {
-      "epoch": 0.8390228308069372,
-      "grad_norm": 1.4600962400436401,
-      "learning_rate": 1.609548167092924e-06,
-      "loss": 0.1336,
-      "step": 4921
-    },
-    {
-      "epoch": 0.8391933292484748,
-      "grad_norm": 0.9181307554244995,
-      "learning_rate": 1.6078431372549023e-06,
-      "loss": 0.0863,
-      "step": 4922
-    },
-    {
-      "epoch": 0.8393638276900125,
-      "grad_norm": 1.2522655725479126,
-      "learning_rate": 1.60613810741688e-06,
-      "loss": 0.1183,
-      "step": 4923
-    },
-    {
-      "epoch": 0.8395343261315502,
-      "grad_norm": 1.0354984998703003,
-      "learning_rate": 1.6044330775788577e-06,
-      "loss": 0.0542,
-      "step": 4924
-    },
-    {
-      "epoch": 0.8397048245730879,
-      "grad_norm": 1.0398370027542114,
-      "learning_rate": 1.6027280477408355e-06,
-      "loss": 0.1141,
-      "step": 4925
-    },
-    {
-      "epoch": 0.8398753230146255,
-      "grad_norm": 0.7802371382713318,
-      "learning_rate": 1.6010230179028136e-06,
-      "loss": 0.0647,
-      "step": 4926
-    },
-    {
-      "epoch": 0.8400458214561632,
-      "grad_norm": 0.9803279042243958,
-      "learning_rate": 1.5993179880647914e-06,
-      "loss": 0.0742,
-      "step": 4927
-    },
-    {
-      "epoch": 0.840216319897701,
-      "grad_norm": 0.9054097533226013,
-      "learning_rate": 1.597612958226769e-06,
-      "loss": 0.0417,
-      "step": 4928
-    },
-    {
-      "epoch": 0.8403868183392387,
-      "grad_norm": 1.2059639692306519,
-      "learning_rate": 1.5959079283887468e-06,
-      "loss": 0.1149,
-      "step": 4929
-    },
-    {
-      "epoch": 0.8405573167807763,
-      "grad_norm": 1.4713133573532104,
-      "learning_rate": 1.5942028985507246e-06,
-      "loss": 0.1514,
-      "step": 4930
-    },
-    {
-      "epoch": 0.840727815222314,
-      "grad_norm": 1.3940950632095337,
-      "learning_rate": 1.5924978687127027e-06,
-      "loss": 0.1448,
-      "step": 4931
-    },
-    {
-      "epoch": 0.8408983136638517,
-      "grad_norm": 1.4991226196289062,
-      "learning_rate": 1.5907928388746805e-06,
-      "loss": 0.1628,
-      "step": 4932
-    },
-    {
-      "epoch": 0.8410688121053893,
-      "grad_norm": 0.8531244993209839,
-      "learning_rate": 1.5890878090366582e-06,
-      "loss": 0.089,
-      "step": 4933
-    },
-    {
-      "epoch": 0.841239310546927,
-      "grad_norm": 1.2540137767791748,
-      "learning_rate": 1.587382779198636e-06,
-      "loss": 0.0978,
-      "step": 4934
-    },
-    {
-      "epoch": 0.8414098089884647,
-      "grad_norm": 1.934168815612793,
-      "learning_rate": 1.585677749360614e-06,
-      "loss": 0.1578,
-      "step": 4935
-    },
-    {
-      "epoch": 0.8415803074300024,
-      "grad_norm": 1.7683275938034058,
-      "learning_rate": 1.5839727195225918e-06,
-      "loss": 0.1453,
-      "step": 4936
-    },
-    {
-      "epoch": 0.84175080587154,
-      "grad_norm": 1.2101832628250122,
-      "learning_rate": 1.5822676896845696e-06,
-      "loss": 0.1185,
-      "step": 4937
-    },
-    {
-      "epoch": 0.8419213043130778,
-      "grad_norm": 1.5000008344650269,
-      "learning_rate": 1.5805626598465473e-06,
-      "loss": 0.1189,
-      "step": 4938
-    },
-    {
-      "epoch": 0.8420918027546155,
-      "grad_norm": 1.8698530197143555,
-      "learning_rate": 1.5788576300085255e-06,
-      "loss": 0.1658,
-      "step": 4939
-    },
-    {
-      "epoch": 0.8422623011961531,
-      "grad_norm": 1.5301642417907715,
-      "learning_rate": 1.5771526001705032e-06,
-      "loss": 0.1542,
-      "step": 4940
-    },
-    {
-      "epoch": 0.8424327996376908,
-      "grad_norm": 1.2595562934875488,
-      "learning_rate": 1.575447570332481e-06,
-      "loss": 0.1349,
-      "step": 4941
-    },
-    {
-      "epoch": 0.8426032980792285,
-      "grad_norm": 1.5561221837997437,
-      "learning_rate": 1.5737425404944587e-06,
-      "loss": 0.1042,
-      "step": 4942
-    },
-    {
-      "epoch": 0.8427737965207662,
-      "grad_norm": 1.0200626850128174,
-      "learning_rate": 1.5720375106564368e-06,
-      "loss": 0.0825,
-      "step": 4943
-    },
-    {
-      "epoch": 0.8429442949623038,
-      "grad_norm": 1.5128827095031738,
-      "learning_rate": 1.5703324808184146e-06,
-      "loss": 0.1609,
-      "step": 4944
-    },
-    {
-      "epoch": 0.8431147934038415,
-      "grad_norm": 1.8137482404708862,
-      "learning_rate": 1.5686274509803923e-06,
-      "loss": 0.2222,
-      "step": 4945
-    },
-    {
-      "epoch": 0.8432852918453793,
-      "grad_norm": 1.5162135362625122,
-      "learning_rate": 1.56692242114237e-06,
-      "loss": 0.128,
-      "step": 4946
-    },
-    {
-      "epoch": 0.8434557902869169,
-      "grad_norm": 1.5846889019012451,
-      "learning_rate": 1.565217391304348e-06,
-      "loss": 0.1441,
-      "step": 4947
-    },
-    {
-      "epoch": 0.8436262887284546,
-      "grad_norm": 1.188979983329773,
-      "learning_rate": 1.563512361466326e-06,
-      "loss": 0.0862,
-      "step": 4948
-    },
-    {
-      "epoch": 0.8437967871699923,
-      "grad_norm": 1.2255123853683472,
-      "learning_rate": 1.5618073316283037e-06,
-      "loss": 0.1131,
-      "step": 4949
-    },
-    {
-      "epoch": 0.84396728561153,
-      "grad_norm": 0.8165611028671265,
-      "learning_rate": 1.5601023017902814e-06,
-      "loss": 0.069,
-      "step": 4950
-    },
-    {
-      "epoch": 0.8441377840530676,
-      "grad_norm": 1.3417201042175293,
-      "learning_rate": 1.5583972719522594e-06,
-      "loss": 0.139,
-      "step": 4951
-    },
-    {
-      "epoch": 0.8443082824946053,
-      "grad_norm": 0.9354560375213623,
-      "learning_rate": 1.5566922421142373e-06,
-      "loss": 0.0778,
-      "step": 4952
-    },
-    {
-      "epoch": 0.844478780936143,
-      "grad_norm": 1.1510093212127686,
-      "learning_rate": 1.554987212276215e-06,
-      "loss": 0.1106,
-      "step": 4953
-    },
-    {
-      "epoch": 0.8446492793776806,
-      "grad_norm": 1.5288021564483643,
-      "learning_rate": 1.5532821824381928e-06,
-      "loss": 0.1585,
-      "step": 4954
-    },
-    {
-      "epoch": 0.8448197778192184,
-      "grad_norm": 0.8060934543609619,
-      "learning_rate": 1.5515771526001705e-06,
-      "loss": 0.0384,
-      "step": 4955
-    },
-    {
-      "epoch": 0.8449902762607561,
-      "grad_norm": 1.1227192878723145,
-      "learning_rate": 1.5498721227621485e-06,
-      "loss": 0.0937,
-      "step": 4956
-    },
-    {
-      "epoch": 0.8451607747022938,
-      "grad_norm": 1.2658599615097046,
-      "learning_rate": 1.5481670929241264e-06,
-      "loss": 0.1081,
-      "step": 4957
-    },
-    {
-      "epoch": 0.8453312731438314,
-      "grad_norm": 1.8658208847045898,
-      "learning_rate": 1.5464620630861041e-06,
-      "loss": 0.1758,
-      "step": 4958
-    },
-    {
-      "epoch": 0.8455017715853691,
-      "grad_norm": 1.0130605697631836,
-      "learning_rate": 1.5447570332480819e-06,
-      "loss": 0.0907,
-      "step": 4959
-    },
-    {
-      "epoch": 0.8456722700269068,
-      "grad_norm": 1.6483023166656494,
-      "learning_rate": 1.5430520034100598e-06,
-      "loss": 0.0851,
-      "step": 4960
-    },
-    {
-      "epoch": 0.8458427684684444,
-      "grad_norm": 1.4492061138153076,
-      "learning_rate": 1.5413469735720376e-06,
-      "loss": 0.1319,
-      "step": 4961
-    },
-    {
-      "epoch": 0.8460132669099821,
-      "grad_norm": 1.1907610893249512,
-      "learning_rate": 1.5396419437340155e-06,
-      "loss": 0.115,
-      "step": 4962
-    },
-    {
-      "epoch": 0.8461837653515198,
-      "grad_norm": 1.2362143993377686,
-      "learning_rate": 1.5379369138959932e-06,
-      "loss": 0.1142,
-      "step": 4963
-    },
-    {
-      "epoch": 0.8463542637930576,
-      "grad_norm": 1.0403637886047363,
-      "learning_rate": 1.5362318840579712e-06,
-      "loss": 0.0966,
-      "step": 4964
-    },
-    {
-      "epoch": 0.8465247622345952,
-      "grad_norm": 1.4426369667053223,
-      "learning_rate": 1.534526854219949e-06,
-      "loss": 0.1402,
-      "step": 4965
-    },
-    {
-      "epoch": 0.8466952606761329,
-      "grad_norm": 1.1102651357650757,
-      "learning_rate": 1.5328218243819269e-06,
-      "loss": 0.1067,
-      "step": 4966
-    },
-    {
-      "epoch": 0.8468657591176706,
-      "grad_norm": 1.952067255973816,
-      "learning_rate": 1.5311167945439046e-06,
-      "loss": 0.1652,
-      "step": 4967
-    },
-    {
-      "epoch": 0.8470362575592082,
-      "grad_norm": 0.5843350291252136,
-      "learning_rate": 1.5294117647058826e-06,
-      "loss": 0.0278,
-      "step": 4968
-    },
-    {
-      "epoch": 0.8472067560007459,
-      "grad_norm": 1.8096089363098145,
-      "learning_rate": 1.5277067348678603e-06,
-      "loss": 0.2164,
-      "step": 4969
-    },
-    {
-      "epoch": 0.8473772544422836,
-      "grad_norm": 1.3740298748016357,
-      "learning_rate": 1.526001705029838e-06,
-      "loss": 0.1416,
-      "step": 4970
-    },
-    {
-      "epoch": 0.8475477528838213,
-      "grad_norm": 1.931396245956421,
-      "learning_rate": 1.524296675191816e-06,
-      "loss": 0.1531,
-      "step": 4971
-    },
-    {
-      "epoch": 0.8477182513253589,
-      "grad_norm": 1.6420286893844604,
-      "learning_rate": 1.522591645353794e-06,
-      "loss": 0.2084,
-      "step": 4972
-    },
-    {
-      "epoch": 0.8478887497668967,
-      "grad_norm": 0.8120304942131042,
-      "learning_rate": 1.5208866155157717e-06,
-      "loss": 0.0732,
-      "step": 4973
-    },
-    {
-      "epoch": 0.8480592482084344,
-      "grad_norm": 1.8183575868606567,
-      "learning_rate": 1.5191815856777494e-06,
-      "loss": 0.1322,
-      "step": 4974
-    },
-    {
-      "epoch": 0.8482297466499721,
-      "grad_norm": 1.8287473917007446,
-      "learning_rate": 1.5174765558397273e-06,
-      "loss": 0.1104,
-      "step": 4975
-    },
-    {
-      "epoch": 0.8484002450915097,
-      "grad_norm": 1.7065812349319458,
-      "learning_rate": 1.515771526001705e-06,
-      "loss": 0.0899,
-      "step": 4976
-    },
-    {
-      "epoch": 0.8485707435330474,
-      "grad_norm": 2.2731716632843018,
-      "learning_rate": 1.514066496163683e-06,
-      "loss": 0.1503,
-      "step": 4977
-    },
-    {
-      "epoch": 0.8487412419745851,
-      "grad_norm": 1.0893604755401611,
-      "learning_rate": 1.5123614663256608e-06,
-      "loss": 0.0547,
-      "step": 4978
-    },
-    {
-      "epoch": 0.8489117404161227,
-      "grad_norm": 1.4843307733535767,
-      "learning_rate": 1.5106564364876385e-06,
-      "loss": 0.0802,
-      "step": 4979
-    },
-    {
-      "epoch": 0.8490822388576604,
-      "grad_norm": 1.09200119972229,
-      "learning_rate": 1.5089514066496164e-06,
-      "loss": 0.0775,
-      "step": 4980
-    },
-    {
-      "epoch": 0.8492527372991981,
-      "grad_norm": 1.0041292905807495,
-      "learning_rate": 1.5072463768115944e-06,
-      "loss": 0.0594,
-      "step": 4981
-    },
-    {
-      "epoch": 0.8494232357407359,
-      "grad_norm": 1.7999296188354492,
-      "learning_rate": 1.5055413469735721e-06,
-      "loss": 0.1413,
-      "step": 4982
-    },
-    {
-      "epoch": 0.8495937341822735,
-      "grad_norm": 2.3291573524475098,
-      "learning_rate": 1.5038363171355499e-06,
-      "loss": 0.1984,
-      "step": 4983
-    },
-    {
-      "epoch": 0.8497642326238112,
-      "grad_norm": 0.9111186265945435,
-      "learning_rate": 1.5021312872975278e-06,
-      "loss": 0.0399,
-      "step": 4984
-    },
-    {
-      "epoch": 0.8499347310653489,
-      "grad_norm": 1.2384296655654907,
-      "learning_rate": 1.5004262574595058e-06,
-      "loss": 0.085,
-      "step": 4985
-    },
-    {
-      "epoch": 0.8501052295068865,
-      "grad_norm": 1.2116646766662598,
-      "learning_rate": 1.4987212276214835e-06,
-      "loss": 0.0481,
-      "step": 4986
-    },
-    {
-      "epoch": 0.8502757279484242,
-      "grad_norm": 1.6571450233459473,
-      "learning_rate": 1.4970161977834612e-06,
-      "loss": 0.2027,
-      "step": 4987
-    },
-    {
-      "epoch": 0.8504462263899619,
-      "grad_norm": 3.2682974338531494,
-      "learning_rate": 1.495311167945439e-06,
-      "loss": 0.339,
-      "step": 4988
-    },
-    {
-      "epoch": 0.8506167248314996,
-      "grad_norm": 1.6870797872543335,
-      "learning_rate": 1.4936061381074171e-06,
-      "loss": 0.1268,
-      "step": 4989
-    },
-    {
-      "epoch": 0.8507872232730372,
-      "grad_norm": 1.3081997632980347,
-      "learning_rate": 1.4919011082693949e-06,
-      "loss": 0.0856,
-      "step": 4990
-    },
-    {
-      "epoch": 0.850957721714575,
-      "grad_norm": 1.368513584136963,
-      "learning_rate": 1.4901960784313726e-06,
-      "loss": 0.1201,
-      "step": 4991
-    },
-    {
-      "epoch": 0.8511282201561127,
-      "grad_norm": 1.4982609748840332,
-      "learning_rate": 1.4884910485933503e-06,
-      "loss": 0.1889,
-      "step": 4992
-    },
-    {
-      "epoch": 0.8512987185976503,
-      "grad_norm": 1.5080363750457764,
-      "learning_rate": 1.4867860187553285e-06,
-      "loss": 0.1494,
-      "step": 4993
-    },
-    {
-      "epoch": 0.851469217039188,
-      "grad_norm": 1.3966141939163208,
-      "learning_rate": 1.4850809889173062e-06,
-      "loss": 0.1247,
-      "step": 4994
-    },
-    {
-      "epoch": 0.8516397154807257,
-      "grad_norm": 1.3656100034713745,
-      "learning_rate": 1.483375959079284e-06,
-      "loss": 0.1148,
-      "step": 4995
-    },
-    {
-      "epoch": 0.8518102139222634,
-      "grad_norm": 1.749886393547058,
-      "learning_rate": 1.4816709292412617e-06,
-      "loss": 0.158,
-      "step": 4996
-    },
-    {
-      "epoch": 0.851980712363801,
-      "grad_norm": 0.9646751284599304,
-      "learning_rate": 1.4799658994032399e-06,
-      "loss": 0.0809,
-      "step": 4997
-    },
-    {
-      "epoch": 0.8521512108053387,
-      "grad_norm": 1.7525877952575684,
-      "learning_rate": 1.4782608695652176e-06,
-      "loss": 0.0756,
-      "step": 4998
-    },
-    {
-      "epoch": 0.8523217092468764,
-      "grad_norm": 1.801682472229004,
-      "learning_rate": 1.4765558397271953e-06,
-      "loss": 0.1975,
-      "step": 4999
-    },
-    {
-      "epoch": 0.852492207688414,
-      "grad_norm": 1.4232743978500366,
-      "learning_rate": 1.474850809889173e-06,
-      "loss": 0.0964,
-      "step": 5000
-    },
-    {
-      "epoch": 0.852492207688414,
-      "eval_f1_score": 0.40703517587939697,
-      "eval_loss": 0.1314370483160019,
-      "eval_runtime": 183.3181,
-      "eval_samples_per_second": 54.55,
-      "eval_steps_per_second": 3.409,
-      "step": 5000
-    }
-  ],
-  "logging_steps": 1,
-  "max_steps": 5865,
-  "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
-  "save_steps": 500,
-  "stateful_callbacks": {
-    "TrainerControl": {
-      "args": {
-        "should_epoch_stop": false,
-        "should_evaluate": false,
-        "should_log": false,
-        "should_save": true,
-        "should_training_stop": false
-      },
-      "attributes": {}
-    }
-  },
-  "total_flos": 8.9352014266368e+17,
-  "train_batch_size": 8,
-  "trial_name": null,
-  "trial_params": null
-}