diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -11,3873 +11,3873 @@ "log_history": [ { "epoch": 0.0018066847335140017, - "grad_norm": 2.6597820350894357, + "grad_norm": 2.906479789331899, "learning_rate": 0.0, "loss": 0.8251, "step": 1 }, { "epoch": 0.0036133694670280035, - "grad_norm": 2.7266510344811494, + "grad_norm": 2.946569058957928, "learning_rate": 1.0714285714285714e-06, "loss": 0.8284, "step": 2 }, { "epoch": 0.005420054200542005, - "grad_norm": 2.7200407943823968, + "grad_norm": 2.96702642089792, "learning_rate": 2.1428571428571427e-06, - "loss": 0.8429, + "loss": 0.8428, "step": 3 }, { "epoch": 0.007226738934056007, - "grad_norm": 2.6398267415203076, + "grad_norm": 2.9084398756854015, "learning_rate": 3.2142857142857143e-06, - "loss": 0.8381, + "loss": 0.8378, "step": 4 }, { "epoch": 0.009033423667570008, - "grad_norm": 2.450058379558589, + "grad_norm": 2.621568341450903, "learning_rate": 4.2857142857142855e-06, - "loss": 0.8245, + "loss": 0.8229, "step": 5 }, { "epoch": 0.01084010840108401, - "grad_norm": 1.9536840033460428, + "grad_norm": 2.1930464956581446, "learning_rate": 5.357142857142857e-06, - "loss": 0.814, + "loss": 0.8097, "step": 6 }, { "epoch": 0.012646793134598013, - "grad_norm": 1.2180340367347295, + "grad_norm": 1.4933984713931483, "learning_rate": 6.428571428571429e-06, - "loss": 0.7845, + "loss": 0.7742, "step": 7 }, { "epoch": 0.014453477868112014, - "grad_norm": 1.1514973199014618, + "grad_norm": 1.397985264834512, "learning_rate": 7.5e-06, - "loss": 0.7874, + "loss": 0.7739, "step": 8 }, { "epoch": 0.016260162601626018, - "grad_norm": 1.1427251731602237, + "grad_norm": 1.3545488436805673, "learning_rate": 8.571428571428571e-06, - "loss": 0.7793, + "loss": 0.7663, "step": 9 }, { "epoch": 0.018066847335140017, - "grad_norm": 2.4537001853666673, + "grad_norm": 2.316903712014158, "learning_rate": 9.642857142857144e-06, - "loss": 0.7624, + "loss": 0.7505, "step": 10 }, { "epoch": 0.01987353206865402, - "grad_norm": 2.474412722550665, + "grad_norm": 2.3724937294713033, "learning_rate": 1.0714285714285714e-05, - "loss": 0.7491, + "loss": 0.7368, "step": 11 }, { "epoch": 0.02168021680216802, - "grad_norm": 2.333456673377541, + "grad_norm": 2.1948974279154383, "learning_rate": 1.1785714285714286e-05, - "loss": 0.762, + "loss": 0.7501, "step": 12 }, { "epoch": 0.023486901535682024, - "grad_norm": 1.6274020508818596, + "grad_norm": 1.7769826234474873, "learning_rate": 1.2857142857142857e-05, - "loss": 0.7368, + "loss": 0.7263, "step": 13 }, { "epoch": 0.025293586269196026, - "grad_norm": 1.563869927563279, + "grad_norm": 1.7537734276506007, "learning_rate": 1.3928571428571429e-05, - "loss": 0.7378, + "loss": 0.7297, "step": 14 }, { "epoch": 0.02710027100271003, - "grad_norm": 1.253626680340628, + "grad_norm": 1.4150131497465723, "learning_rate": 1.5e-05, - "loss": 0.7271, + "loss": 0.7181, "step": 15 }, { "epoch": 0.028906955736224028, - "grad_norm": 0.9176330164857595, + "grad_norm": 1.388901011985526, "learning_rate": 1.6071428571428572e-05, - "loss": 0.7131, + "loss": 0.7029, "step": 16 }, { "epoch": 0.03071364046973803, - "grad_norm": 0.7185343991261351, + "grad_norm": 0.9533064996374424, "learning_rate": 1.7142857142857142e-05, - "loss": 0.7131, + "loss": 0.7021, "step": 17 }, { "epoch": 0.032520325203252036, - "grad_norm": 0.7884609645105337, + "grad_norm": 0.7772119179800466, "learning_rate": 1.8214285714285712e-05, - "loss": 0.7045, + "loss": 0.6907, "step": 18 }, { "epoch": 0.03432700993676603, - "grad_norm": 0.7399830971167167, + "grad_norm": 0.7719134453183193, "learning_rate": 1.928571428571429e-05, - "loss": 0.6995, + "loss": 0.6849, "step": 19 }, { "epoch": 0.036133694670280034, - "grad_norm": 0.7376328500886337, + "grad_norm": 0.6865260996169726, "learning_rate": 2.0357142857142858e-05, - "loss": 0.6779, + "loss": 0.661, "step": 20 }, { "epoch": 0.037940379403794036, - "grad_norm": 0.6439873145292982, + "grad_norm": 0.6261375647490827, "learning_rate": 2.1428571428571428e-05, - "loss": 0.6834, + "loss": 0.6666, "step": 21 }, { "epoch": 0.03974706413730804, - "grad_norm": 0.5790542088592184, + "grad_norm": 0.6378186989443875, "learning_rate": 2.25e-05, - "loss": 0.6721, + "loss": 0.6558, "step": 22 }, { "epoch": 0.04155374887082204, - "grad_norm": 0.5191304250088061, + "grad_norm": 0.6713458173721598, "learning_rate": 2.357142857142857e-05, - "loss": 0.6707, + "loss": 0.6564, "step": 23 }, { "epoch": 0.04336043360433604, - "grad_norm": 0.4488946670945292, + "grad_norm": 0.6274625379226225, "learning_rate": 2.464285714285714e-05, - "loss": 0.676, + "loss": 0.6614, "step": 24 }, { "epoch": 0.045167118337850046, - "grad_norm": 0.4798647764154148, + "grad_norm": 0.5715761838454848, "learning_rate": 2.5714285714285714e-05, - "loss": 0.6585, + "loss": 0.6432, "step": 25 }, { "epoch": 0.04697380307136405, - "grad_norm": 0.4204330566493729, + "grad_norm": 0.48739002531211656, "learning_rate": 2.6785714285714288e-05, - "loss": 0.673, + "loss": 0.6574, "step": 26 }, { "epoch": 0.04878048780487805, - "grad_norm": 0.47130112684574876, + "grad_norm": 0.5170424711902095, "learning_rate": 2.7857142857142858e-05, - "loss": 0.6516, + "loss": 0.6355, "step": 27 }, { "epoch": 0.05058717253839205, - "grad_norm": 0.4362815754533529, + "grad_norm": 0.48953368197603536, "learning_rate": 2.892857142857143e-05, - "loss": 0.6545, + "loss": 0.6386, "step": 28 }, { "epoch": 0.052393857271906055, - "grad_norm": 0.43822648560890193, + "grad_norm": 0.5005989792055041, "learning_rate": 3e-05, - "loss": 0.6647, + "loss": 0.6504, "step": 29 }, { "epoch": 0.05420054200542006, - "grad_norm": 0.3853311574389301, + "grad_norm": 0.4544942981478791, "learning_rate": 2.9999731440137413e-05, - "loss": 0.6644, + "loss": 0.6514, "step": 30 }, { "epoch": 0.05600722673893405, - "grad_norm": 0.4221785516753897, + "grad_norm": 0.3974914740221412, "learning_rate": 2.9998925770166232e-05, - "loss": 0.6534, + "loss": 0.6402, "step": 31 }, { "epoch": 0.057813911472448055, - "grad_norm": 0.3980570059327311, + "grad_norm": 0.3878954325744685, "learning_rate": 2.9997583018935875e-05, - "loss": 0.6416, + "loss": 0.6281, "step": 32 }, { "epoch": 0.05962059620596206, - "grad_norm": 0.35571676272289743, + "grad_norm": 0.4561986805367028, "learning_rate": 2.9995703234527553e-05, - "loss": 0.6513, + "loss": 0.6383, "step": 33 }, { "epoch": 0.06142728093947606, - "grad_norm": 0.38024957472516396, + "grad_norm": 0.37737783307889894, "learning_rate": 2.999328648425255e-05, - "loss": 0.6385, + "loss": 0.6262, "step": 34 }, { "epoch": 0.06323396567299007, - "grad_norm": 0.2924858728569485, + "grad_norm": 0.3869373298490812, "learning_rate": 2.999033285464982e-05, - "loss": 0.644, + "loss": 0.6313, "step": 35 }, { "epoch": 0.06504065040650407, - "grad_norm": 0.3482798299572746, + "grad_norm": 0.4272515794130934, "learning_rate": 2.9986842451482876e-05, - "loss": 0.6387, + "loss": 0.6274, "step": 36 }, { "epoch": 0.06684733514001806, - "grad_norm": 0.3705337140725527, + "grad_norm": 0.36158577191319774, "learning_rate": 2.9982815399736008e-05, - "loss": 0.6432, + "loss": 0.632, "step": 37 }, { "epoch": 0.06865401987353206, - "grad_norm": 0.3364721163164359, + "grad_norm": 0.4217112984330933, "learning_rate": 2.9978251843609816e-05, - "loss": 0.6351, + "loss": 0.6248, "step": 38 }, { "epoch": 0.07046070460704607, - "grad_norm": 0.3272631660096153, + "grad_norm": 0.7204159027715281, "learning_rate": 2.9973151946516027e-05, - "loss": 0.6364, + "loss": 0.6269, "step": 39 }, { "epoch": 0.07226738934056007, - "grad_norm": 0.29708127282462954, + "grad_norm": 0.3648244667440142, "learning_rate": 2.996751589107167e-05, - "loss": 0.6252, + "loss": 0.6154, "step": 40 }, { "epoch": 0.07407407407407407, - "grad_norm": 0.24921077158970448, + "grad_norm": 0.37700089354074445, "learning_rate": 2.9961343879092512e-05, - "loss": 0.6308, + "loss": 0.6218, "step": 41 }, { "epoch": 0.07588075880758807, - "grad_norm": 0.3479569257927118, + "grad_norm": 0.32409600863650384, "learning_rate": 2.9954636131585845e-05, - "loss": 0.6234, + "loss": 0.6143, "step": 42 }, { "epoch": 0.07768744354110207, - "grad_norm": 0.28917511643628596, + "grad_norm": 0.3707207340219278, "learning_rate": 2.9947392888742566e-05, - "loss": 0.6233, + "loss": 0.6148, "step": 43 }, { "epoch": 0.07949412827461608, - "grad_norm": 0.34737822266968965, + "grad_norm": 0.39427784806992083, "learning_rate": 2.993961440992859e-05, - "loss": 0.6171, + "loss": 0.609, "step": 44 }, { "epoch": 0.08130081300813008, - "grad_norm": 0.3247689228102166, + "grad_norm": 0.32610161762849305, "learning_rate": 2.993130097367553e-05, - "loss": 0.6238, + "loss": 0.6168, "step": 45 }, { "epoch": 0.08310749774164408, - "grad_norm": 0.30719902123424325, + "grad_norm": 0.35669866117775423, "learning_rate": 2.9922452877670775e-05, - "loss": 0.6252, + "loss": 0.6185, "step": 46 }, { "epoch": 0.08491418247515808, - "grad_norm": 0.3175141153740525, + "grad_norm": 0.3676448513773045, "learning_rate": 2.991307043874677e-05, - "loss": 0.6281, + "loss": 0.6213, "step": 47 }, { "epoch": 0.08672086720867209, - "grad_norm": 0.31244488771705314, + "grad_norm": 0.38768747406718834, "learning_rate": 2.9903153992869734e-05, - "loss": 0.6148, + "loss": 0.6072, "step": 48 }, { "epoch": 0.08852755194218609, - "grad_norm": 0.3429131319420423, + "grad_norm": 0.3564630767927543, "learning_rate": 2.989270389512756e-05, - "loss": 0.6211, + "loss": 0.6154, "step": 49 }, { "epoch": 0.09033423667570009, - "grad_norm": 0.3104062918234721, + "grad_norm": 0.38941497566142114, "learning_rate": 2.988172051971717e-05, - "loss": 0.6204, + "loss": 0.615, "step": 50 }, { "epoch": 0.0921409214092141, - "grad_norm": 0.31461103786584654, + "grad_norm": 0.3584473583813607, "learning_rate": 2.9870204259931062e-05, - "loss": 0.6097, + "loss": 0.6041, "step": 51 }, { "epoch": 0.0939476061427281, - "grad_norm": 0.3151625720082357, + "grad_norm": 0.6346955736685954, "learning_rate": 2.9858155528143256e-05, - "loss": 0.6268, + "loss": 0.6216, "step": 52 }, { "epoch": 0.0957542908762421, - "grad_norm": 0.2860126481493585, + "grad_norm": 0.41395691562612613, "learning_rate": 2.9845574755794522e-05, - "loss": 0.6125, + "loss": 0.6074, "step": 53 }, { "epoch": 0.0975609756097561, - "grad_norm": 0.28841121592431124, + "grad_norm": 0.3325284883699586, "learning_rate": 2.9832462393376926e-05, - "loss": 0.6018, + "loss": 0.5977, "step": 54 }, { "epoch": 0.0993676603432701, - "grad_norm": 0.23534181927847642, + "grad_norm": 0.4006744277449496, "learning_rate": 2.9818818910417706e-05, - "loss": 0.6169, + "loss": 0.6134, "step": 55 }, { "epoch": 0.1011743450767841, - "grad_norm": 0.28661471046317555, + "grad_norm": 0.3503235962002047, "learning_rate": 2.9804644795462437e-05, - "loss": 0.6212, + "loss": 0.6178, "step": 56 }, { "epoch": 0.10298102981029811, - "grad_norm": 0.2443857675436309, + "grad_norm": 0.3596793230653787, "learning_rate": 2.9789940556057574e-05, - "loss": 0.6065, + "loss": 0.6026, "step": 57 }, { "epoch": 0.10478771454381211, - "grad_norm": 0.292418070185767, + "grad_norm": 0.375719190862958, "learning_rate": 2.9774706718732255e-05, - "loss": 0.6152, + "loss": 0.6115, "step": 58 }, { "epoch": 0.10659439927732611, - "grad_norm": 0.24742182455944323, + "grad_norm": 0.40865738147424024, "learning_rate": 2.9758943828979444e-05, - "loss": 0.6009, + "loss": 0.5972, "step": 59 }, { "epoch": 0.10840108401084012, - "grad_norm": 0.25007211477600044, + "grad_norm": 0.459336272208224, "learning_rate": 2.9742652451236414e-05, - "loss": 0.6023, + "loss": 0.5984, "step": 60 }, { "epoch": 0.1102077687443541, - "grad_norm": 0.24539117288225054, + "grad_norm": 0.4395461744005695, "learning_rate": 2.972583316886451e-05, - "loss": 0.6059, + "loss": 0.6027, "step": 61 }, { "epoch": 0.1120144534778681, - "grad_norm": 0.2544907417129908, + "grad_norm": 0.5409276356046155, "learning_rate": 2.9708486584128303e-05, - "loss": 0.6107, + "loss": 0.607, "step": 62 }, { "epoch": 0.11382113821138211, - "grad_norm": 0.25269030392560393, + "grad_norm": 0.4127832467579958, "learning_rate": 2.9690613318173966e-05, - "loss": 0.611, + "loss": 0.6072, "step": 63 }, { "epoch": 0.11562782294489611, - "grad_norm": 0.20651691272156247, + "grad_norm": 0.3295124749236948, "learning_rate": 2.9672214011007087e-05, - "loss": 0.6091, + "loss": 0.606, "step": 64 }, { "epoch": 0.11743450767841011, - "grad_norm": 0.25759031036445135, + "grad_norm": 0.38489242897645604, "learning_rate": 2.9653289321469715e-05, - "loss": 0.6031, + "loss": 0.5998, "step": 65 }, { "epoch": 0.11924119241192412, - "grad_norm": 0.27004948462624506, + "grad_norm": 0.35627109239374033, "learning_rate": 2.9633839927216793e-05, - "loss": 0.6061, + "loss": 0.6026, "step": 66 }, { "epoch": 0.12104787714543812, - "grad_norm": 0.23899477664304233, + "grad_norm": 0.2956644369843949, "learning_rate": 2.9613866524691867e-05, - "loss": 0.5958, + "loss": 0.593, "step": 67 }, { "epoch": 0.12285456187895212, - "grad_norm": 0.3109356502387666, + "grad_norm": 0.3766640456122109, "learning_rate": 2.9593369829102173e-05, - "loss": 0.6119, + "loss": 0.6093, "step": 68 }, { "epoch": 0.12466124661246612, - "grad_norm": 0.23973937288759506, + "grad_norm": 0.3452101448577207, "learning_rate": 2.957235057439301e-05, - "loss": 0.6135, + "loss": 0.6105, "step": 69 }, { "epoch": 0.12646793134598014, - "grad_norm": 0.29402451520285133, + "grad_norm": 0.3069425861118418, "learning_rate": 2.955080951322147e-05, - "loss": 0.6178, + "loss": 0.6148, "step": 70 }, { "epoch": 0.12827461607949414, - "grad_norm": 0.29085498534508775, + "grad_norm": 0.41522461019104656, "learning_rate": 2.9528747416929467e-05, - "loss": 0.5907, + "loss": 0.5875, "step": 71 }, { "epoch": 0.13008130081300814, - "grad_norm": 0.25477514255751904, + "grad_norm": 0.35892642901243405, "learning_rate": 2.9506165075516148e-05, - "loss": 0.6073, + "loss": 0.6043, "step": 72 }, { "epoch": 0.13188798554652212, - "grad_norm": 0.28366229733780235, + "grad_norm": 0.3560421570359481, "learning_rate": 2.9483063297609577e-05, - "loss": 0.6063, + "loss": 0.6034, "step": 73 }, { "epoch": 0.13369467028003612, - "grad_norm": 0.26317649422081363, + "grad_norm": 0.3724223713554414, "learning_rate": 2.9459442910437798e-05, - "loss": 0.598, + "loss": 0.5958, "step": 74 }, { "epoch": 0.13550135501355012, - "grad_norm": 0.28039599919104496, + "grad_norm": 0.33938996629688567, "learning_rate": 2.94353047597992e-05, - "loss": 0.5971, + "loss": 0.5944, "step": 75 }, { "epoch": 0.13730803974706413, - "grad_norm": 0.271745198157929, + "grad_norm": 0.3499978376673935, "learning_rate": 2.941064971003224e-05, - "loss": 0.593, + "loss": 0.5908, "step": 76 }, { "epoch": 0.13911472448057813, - "grad_norm": 0.24998502389958926, + "grad_norm": 0.3727209187041342, "learning_rate": 2.9385478643984484e-05, - "loss": 0.5878, + "loss": 0.5866, "step": 77 }, { "epoch": 0.14092140921409213, - "grad_norm": 0.30322559816019845, + "grad_norm": 2.377825424765151, "learning_rate": 2.9359792462981007e-05, - "loss": 0.599, + "loss": 0.5978, "step": 78 }, { "epoch": 0.14272809394760613, - "grad_norm": 0.29351194680487536, + "grad_norm": 0.4456515263682091, "learning_rate": 2.9333592086792113e-05, - "loss": 0.6075, + "loss": 0.6069, "step": 79 }, { "epoch": 0.14453477868112014, - "grad_norm": 0.2670107767188747, + "grad_norm": 0.34569754140915615, "learning_rate": 2.9306878453600382e-05, - "loss": 0.6161, + "loss": 0.6146, "step": 80 }, { "epoch": 0.14634146341463414, - "grad_norm": 0.2927366904918854, + "grad_norm": 0.41523607463693246, "learning_rate": 2.9279652519967105e-05, - "loss": 0.6004, + "loss": 0.5989, "step": 81 }, { "epoch": 0.14814814814814814, - "grad_norm": 0.28777141201555767, + "grad_norm": 0.4045393297763538, "learning_rate": 2.9251915260798024e-05, - "loss": 0.593, + "loss": 0.5913, "step": 82 }, { "epoch": 0.14995483288166214, - "grad_norm": 0.282262421377073, + "grad_norm": 0.4028600904179543, "learning_rate": 2.9223667669308395e-05, - "loss": 0.6038, + "loss": 0.6016, "step": 83 }, { "epoch": 0.15176151761517614, - "grad_norm": 0.24811154699761442, + "grad_norm": 0.34357038583025223, "learning_rate": 2.9194910756987464e-05, - "loss": 0.5976, + "loss": 0.5953, "step": 84 }, { "epoch": 0.15356820234869015, - "grad_norm": 0.27670190484589513, + "grad_norm": 0.37799582892290484, "learning_rate": 2.9165645553562215e-05, - "loss": 0.5997, + "loss": 0.5983, "step": 85 }, { "epoch": 0.15537488708220415, - "grad_norm": 0.24338487756299684, + "grad_norm": 1.4320099283126178, "learning_rate": 2.9135873106960525e-05, - "loss": 0.5994, + "loss": 0.5969, "step": 86 }, { "epoch": 0.15718157181571815, - "grad_norm": 0.3040191021283788, + "grad_norm": 0.35047306772001974, "learning_rate": 2.9105594483273603e-05, - "loss": 0.6012, + "loss": 0.6003, "step": 87 }, { "epoch": 0.15898825654923215, - "grad_norm": 0.26703924304748566, + "grad_norm": 0.4630513209073766, "learning_rate": 2.9074810766717865e-05, - "loss": 0.6058, + "loss": 0.6063, "step": 88 }, { "epoch": 0.16079494128274616, - "grad_norm": 0.2694312104706441, + "grad_norm": 0.38854101508245126, "learning_rate": 2.904352305959606e-05, - "loss": 0.602, + "loss": 0.6016, "step": 89 }, { "epoch": 0.16260162601626016, - "grad_norm": 0.29517928394037635, + "grad_norm": 0.40449057823694373, "learning_rate": 2.9011732482257835e-05, - "loss": 0.5944, + "loss": 0.5934, "step": 90 }, { "epoch": 0.16440831074977416, - "grad_norm": 0.25118801401605734, + "grad_norm": 0.43904838090142206, "learning_rate": 2.89794401730596e-05, - "loss": 0.5923, + "loss": 0.5921, "step": 91 }, { "epoch": 0.16621499548328816, - "grad_norm": 0.3348680034328015, + "grad_norm": 0.4396255315476391, "learning_rate": 2.894664728832377e-05, - "loss": 0.5958, + "loss": 0.5959, "step": 92 }, { "epoch": 0.16802168021680217, - "grad_norm": 0.25945936645525364, + "grad_norm": 0.38031370648914475, "learning_rate": 2.8913355002297367e-05, - "loss": 0.6025, + "loss": 0.6022, "step": 93 }, { "epoch": 0.16982836495031617, - "grad_norm": 0.31672412557275315, + "grad_norm": 0.523279763736358, "learning_rate": 2.887956450710995e-05, - "loss": 0.602, + "loss": 0.6028, "step": 94 }, { "epoch": 0.17163504968383017, - "grad_norm": 0.2631935513788724, + "grad_norm": 0.6704399767230238, "learning_rate": 2.8845277012730963e-05, - "loss": 0.5939, + "loss": 0.5942, "step": 95 }, { "epoch": 0.17344173441734417, - "grad_norm": 0.28309967073656406, + "grad_norm": 0.5066644836900498, "learning_rate": 2.8810493746926364e-05, - "loss": 0.5932, + "loss": 0.5933, "step": 96 }, { "epoch": 0.17524841915085818, - "grad_norm": 0.28939178950900774, + "grad_norm": 0.3646318333476554, "learning_rate": 2.87752159552147e-05, - "loss": 0.5993, + "loss": 0.5997, "step": 97 }, { "epoch": 0.17705510388437218, - "grad_norm": 0.3021445789644071, + "grad_norm": 0.376548770661227, "learning_rate": 2.87394449008225e-05, - "loss": 0.5924, + "loss": 0.5916, "step": 98 }, { "epoch": 0.17886178861788618, - "grad_norm": 0.2648161341085879, + "grad_norm": 0.4285200046228619, "learning_rate": 2.8703181864639013e-05, - "loss": 0.6058, + "loss": 0.6057, "step": 99 }, { "epoch": 0.18066847335140018, - "grad_norm": 0.28484263640616314, + "grad_norm": 0.33939531318747057, "learning_rate": 2.8666428145170385e-05, - "loss": 0.5789, + "loss": 0.5779, "step": 100 }, { "epoch": 0.18247515808491419, - "grad_norm": 0.2611846886724786, + "grad_norm": 0.3969014717910721, "learning_rate": 2.8629185058493116e-05, - "loss": 0.5949, + "loss": 0.5944, "step": 101 }, { "epoch": 0.1842818428184282, - "grad_norm": 0.33149916071305763, + "grad_norm": 0.36815480889738744, "learning_rate": 2.8591453938206985e-05, - "loss": 0.6134, + "loss": 0.6126, "step": 102 }, { "epoch": 0.1860885275519422, - "grad_norm": 0.28402400837841096, + "grad_norm": 0.3820598707301275, "learning_rate": 2.8553236135387247e-05, - "loss": 0.5914, + "loss": 0.5911, "step": 103 }, { "epoch": 0.1878952122854562, - "grad_norm": 0.28825320782588004, + "grad_norm": 0.4452739463450766, "learning_rate": 2.8514533018536286e-05, - "loss": 0.5897, + "loss": 0.5887, "step": 104 }, { "epoch": 0.1897018970189702, - "grad_norm": 0.2901395812522203, + "grad_norm": 0.37087706923300234, "learning_rate": 2.8475345973534605e-05, - "loss": 0.5822, + "loss": 0.5817, "step": 105 }, { "epoch": 0.1915085817524842, - "grad_norm": 0.25034051990006584, + "grad_norm": 0.6988372458559541, "learning_rate": 2.8435676403591193e-05, - "loss": 0.5846, + "loss": 0.584, "step": 106 }, { "epoch": 0.1933152664859982, - "grad_norm": 0.27868357694305945, + "grad_norm": 0.3800814296781821, "learning_rate": 2.8395525729193284e-05, - "loss": 0.5907, + "loss": 0.59, "step": 107 }, { "epoch": 0.1951219512195122, - "grad_norm": 0.257742952703987, + "grad_norm": 0.3993950018851527, "learning_rate": 2.835489538805548e-05, - "loss": 0.5888, + "loss": 0.589, "step": 108 }, { "epoch": 0.1969286359530262, - "grad_norm": 0.26397326166674545, + "grad_norm": 0.34149177022116556, "learning_rate": 2.8313786835068314e-05, - "loss": 0.5976, + "loss": 0.5973, "step": 109 }, { "epoch": 0.1987353206865402, - "grad_norm": 0.2721484510112116, + "grad_norm": 0.3199472139502249, "learning_rate": 2.8272201542246077e-05, - "loss": 0.5792, + "loss": 0.5791, "step": 110 }, { "epoch": 0.2005420054200542, - "grad_norm": 0.2374425057236497, + "grad_norm": 0.3608234766567561, "learning_rate": 2.8230140998674185e-05, - "loss": 0.599, + "loss": 0.5989, "step": 111 }, { "epoch": 0.2023486901535682, - "grad_norm": 0.24242218904343407, + "grad_norm": 0.3371671357477985, "learning_rate": 2.8187606710455807e-05, - "loss": 0.5921, + "loss": 0.5922, "step": 112 }, { "epoch": 0.2041553748870822, - "grad_norm": 0.2742052397132371, + "grad_norm": 0.36397935298555917, "learning_rate": 2.8144600200657953e-05, - "loss": 0.5965, + "loss": 0.5962, "step": 113 }, { "epoch": 0.20596205962059622, - "grad_norm": 0.2240931438132582, + "grad_norm": 0.44229892857297654, "learning_rate": 2.8101123009256946e-05, - "loss": 0.5792, + "loss": 0.5803, "step": 114 }, { "epoch": 0.20776874435411022, - "grad_norm": 0.2763456467888191, + "grad_norm": 0.3933110247729101, "learning_rate": 2.8057176693083253e-05, - "loss": 0.5958, + "loss": 0.5971, "step": 115 }, { "epoch": 0.20957542908762422, - "grad_norm": 0.2435434093260049, + "grad_norm": 0.35133090195312044, "learning_rate": 2.8012762825765763e-05, - "loss": 0.5891, + "loss": 0.59, "step": 116 }, { "epoch": 0.21138211382113822, - "grad_norm": 0.2769370185736096, + "grad_norm": 0.42414767478573207, "learning_rate": 2.7967882997675424e-05, - "loss": 0.6012, + "loss": 0.602, "step": 117 }, { "epoch": 0.21318879855465223, - "grad_norm": 0.2556991024086139, + "grad_norm": 0.3435364783148156, "learning_rate": 2.7922538815868287e-05, - "loss": 0.5879, + "loss": 0.5881, "step": 118 }, { "epoch": 0.21499548328816623, - "grad_norm": 0.2913165244476914, + "grad_norm": 0.33943051644040134, "learning_rate": 2.7876731904027994e-05, - "loss": 0.5784, + "loss": 0.5788, "step": 119 }, { "epoch": 0.21680216802168023, - "grad_norm": 0.29037318268167045, + "grad_norm": 0.3221999031967881, "learning_rate": 2.78304639024076e-05, - "loss": 0.5968, + "loss": 0.597, "step": 120 }, { "epoch": 0.2186088527551942, - "grad_norm": 0.24340049924612678, + "grad_norm": 0.7107243183485162, "learning_rate": 2.7783736467770863e-05, "loss": 0.5927, "step": 121 }, { "epoch": 0.2204155374887082, - "grad_norm": 0.29825692316659325, + "grad_norm": 0.4275366085321626, "learning_rate": 2.7736551273332908e-05, - "loss": 0.5972, + "loss": 0.5977, "step": 122 }, { "epoch": 0.2222222222222222, - "grad_norm": 0.2593832259683122, + "grad_norm": 0.34325044467207816, "learning_rate": 2.7688910008700305e-05, - "loss": 0.5876, + "loss": 0.5883, "step": 123 }, { "epoch": 0.2240289069557362, - "grad_norm": 0.27547771011759425, + "grad_norm": 0.3503193328962726, "learning_rate": 2.764081437981059e-05, - "loss": 0.5742, + "loss": 0.5747, "step": 124 }, { "epoch": 0.22583559168925021, - "grad_norm": 0.27158658907185634, + "grad_norm": 0.39807179938908716, "learning_rate": 2.7592266108871158e-05, - "loss": 0.5812, + "loss": 0.5816, "step": 125 }, { "epoch": 0.22764227642276422, - "grad_norm": 0.2715497634460973, + "grad_norm": 0.36593993813666903, "learning_rate": 2.754326693429761e-05, - "loss": 0.5931, + "loss": 0.5936, "step": 126 }, { "epoch": 0.22944896115627822, - "grad_norm": 0.24142985625388147, + "grad_norm": 0.3871443516927523, "learning_rate": 2.7493818610651493e-05, - "loss": 0.5859, + "loss": 0.5875, "step": 127 }, { "epoch": 0.23125564588979222, - "grad_norm": 0.2530474955317245, + "grad_norm": 0.3606586360294697, "learning_rate": 2.744392290857747e-05, - "loss": 0.5857, + "loss": 0.5866, "step": 128 }, { "epoch": 0.23306233062330622, - "grad_norm": 0.2567731311347638, + "grad_norm": 0.38342216198991025, "learning_rate": 2.7393581614739924e-05, - "loss": 0.5803, + "loss": 0.5813, "step": 129 }, { "epoch": 0.23486901535682023, - "grad_norm": 0.2628073110459229, + "grad_norm": 0.37465039657790306, "learning_rate": 2.7342796531758984e-05, - "loss": 0.5987, + "loss": 0.6004, "step": 130 }, { "epoch": 0.23667570009033423, - "grad_norm": 0.23510376935644756, + "grad_norm": 0.3764086380678446, "learning_rate": 2.729156947814598e-05, - "loss": 0.5926, + "loss": 0.5943, "step": 131 }, { "epoch": 0.23848238482384823, - "grad_norm": 0.23576954058784624, + "grad_norm": 0.7637683242409493, "learning_rate": 2.7239902288238297e-05, - "loss": 0.5721, + "loss": 0.5737, "step": 132 }, { "epoch": 0.24028906955736223, - "grad_norm": 0.24964231366212492, + "grad_norm": 0.3560589890890579, "learning_rate": 2.7187796812133733e-05, - "loss": 0.5873, + "loss": 0.5887, "step": 133 }, { "epoch": 0.24209575429087624, - "grad_norm": 0.23911643013915115, + "grad_norm": 0.3825941152602638, "learning_rate": 2.7135254915624213e-05, - "loss": 0.5878, + "loss": 0.59, "step": 134 }, { "epoch": 0.24390243902439024, - "grad_norm": 0.23601189225311667, + "grad_norm": 0.36389498688783184, "learning_rate": 2.708227848012901e-05, - "loss": 0.585, + "loss": 0.5876, "step": 135 }, { "epoch": 0.24570912375790424, - "grad_norm": 0.28061149806949875, + "grad_norm": 0.37878840510853407, "learning_rate": 2.7028869402627357e-05, - "loss": 0.5889, + "loss": 0.591, "step": 136 }, { "epoch": 0.24751580849141824, - "grad_norm": 0.2537605160128264, + "grad_norm": 0.3899552482148334, "learning_rate": 2.6975029595590523e-05, - "loss": 0.5726, + "loss": 0.5742, "step": 137 }, { "epoch": 0.24932249322493225, - "grad_norm": 0.2631279788827005, + "grad_norm": 0.3807471778199868, "learning_rate": 2.6920760986913332e-05, - "loss": 0.5901, + "loss": 0.592, "step": 138 }, { "epoch": 0.25112917795844625, - "grad_norm": 0.24684546421512338, + "grad_norm": 0.41701742101917744, "learning_rate": 2.6866065519845124e-05, - "loss": 0.5932, + "loss": 0.5953, "step": 139 }, { "epoch": 0.2529358626919603, - "grad_norm": 0.25108587954115535, + "grad_norm": 0.39189298513945836, "learning_rate": 2.681094515292018e-05, - "loss": 0.5927, + "loss": 0.5943, "step": 140 }, { "epoch": 0.25474254742547425, - "grad_norm": 0.2759732073497199, + "grad_norm": 0.5240993515099046, "learning_rate": 2.6755401859887598e-05, - "loss": 0.573, + "loss": 0.5755, "step": 141 }, { "epoch": 0.2565492321589883, - "grad_norm": 0.2927986426856055, + "grad_norm": 0.37512137985409255, "learning_rate": 2.6699437629640595e-05, - "loss": 0.5934, + "loss": 0.5956, "step": 142 }, { "epoch": 0.25835591689250226, - "grad_norm": 0.2915811926744858, + "grad_norm": 0.3707219843953511, "learning_rate": 2.6643054466145297e-05, - "loss": 0.5846, + "loss": 0.5869, "step": 143 }, { "epoch": 0.2601626016260163, - "grad_norm": 0.25639674407942853, + "grad_norm": 0.3555027971231102, "learning_rate": 2.6586254388368995e-05, - "loss": 0.5733, + "loss": 0.5752, "step": 144 }, { "epoch": 0.26196928635953026, - "grad_norm": 0.25466281147535325, + "grad_norm": 0.3026146177016955, "learning_rate": 2.652903943020783e-05, - "loss": 0.5862, + "loss": 0.5883, "step": 145 }, { "epoch": 0.26377597109304424, - "grad_norm": 0.24151461009286013, + "grad_norm": 0.41020681896565947, "learning_rate": 2.647141164041398e-05, - "loss": 0.5904, + "loss": 0.5937, "step": 146 }, { "epoch": 0.26558265582655827, - "grad_norm": 0.259822033974557, + "grad_norm": 0.36661807722028844, "learning_rate": 2.641337308252228e-05, - "loss": 0.581, + "loss": 0.583, "step": 147 }, { "epoch": 0.26738934056007224, - "grad_norm": 0.24532475377494284, + "grad_norm": 0.32357318798570317, "learning_rate": 2.6354925834776346e-05, - "loss": 0.5817, + "loss": 0.5841, "step": 148 }, { "epoch": 0.26919602529358627, - "grad_norm": 0.2555788655192275, + "grad_norm": 0.4240562481643986, "learning_rate": 2.6296071990054167e-05, - "loss": 0.579, + "loss": 0.5812, "step": 149 }, { "epoch": 0.27100271002710025, - "grad_norm": 0.2534339164884998, + "grad_norm": 0.35429923851701156, "learning_rate": 2.6236813655793123e-05, - "loss": 0.5886, + "loss": 0.5905, "step": 150 }, { "epoch": 0.2728093947606143, - "grad_norm": 0.25366511050468843, + "grad_norm": 0.3981067430459517, "learning_rate": 2.617715295391457e-05, - "loss": 0.5707, + "loss": 0.572, "step": 151 }, { "epoch": 0.27461607949412825, - "grad_norm": 0.26118391438335414, + "grad_norm": 0.516641806924778, "learning_rate": 2.6117092020747824e-05, - "loss": 0.5779, + "loss": 0.5802, "step": 152 }, { "epoch": 0.2764227642276423, - "grad_norm": 0.31049055796662123, + "grad_norm": 0.3349831187986836, "learning_rate": 2.6056633006953677e-05, - "loss": 0.5783, + "loss": 0.581, "step": 153 }, { "epoch": 0.27822944896115626, - "grad_norm": 0.2611788471953917, + "grad_norm": 0.32079723720771636, "learning_rate": 2.5995778077447393e-05, - "loss": 0.5823, + "loss": 0.5847, "step": 154 }, { "epoch": 0.2800361336946703, - "grad_norm": 0.3259190805002546, + "grad_norm": 0.4491293903796211, "learning_rate": 2.5934529411321174e-05, - "loss": 0.5743, + "loss": 0.577, "step": 155 }, { "epoch": 0.28184281842818426, - "grad_norm": 0.2222204544644696, + "grad_norm": 0.3338078899554527, "learning_rate": 2.587288920176613e-05, - "loss": 0.575, + "loss": 0.577, "step": 156 }, { "epoch": 0.2836495031616983, - "grad_norm": 0.29042132848143276, + "grad_norm": 0.337095193899371, "learning_rate": 2.581085965599375e-05, - "loss": 0.5877, + "loss": 0.5905, "step": 157 }, { "epoch": 0.28545618789521227, - "grad_norm": 0.2577824036794116, + "grad_norm": 0.3241526231317586, "learning_rate": 2.5748442995156882e-05, - "loss": 0.5854, + "loss": 0.5872, "step": 158 }, { "epoch": 0.2872628726287263, - "grad_norm": 0.24041412490024394, + "grad_norm": 0.32809237233012317, "learning_rate": 2.5685641454270172e-05, - "loss": 0.566, + "loss": 0.5683, "step": 159 }, { "epoch": 0.28906955736224027, - "grad_norm": 0.33137613211077116, + "grad_norm": 0.33268577227395, "learning_rate": 2.5622457282130046e-05, - "loss": 0.5753, + "loss": 0.5778, "step": 160 }, { "epoch": 0.2908762420957543, - "grad_norm": 0.2252657452973701, + "grad_norm": 0.4132273947780692, "learning_rate": 2.5558892741234173e-05, - "loss": 0.5827, + "loss": 0.5861, "step": 161 }, { "epoch": 0.2926829268292683, - "grad_norm": 0.353225035311219, + "grad_norm": 0.33902072712237397, "learning_rate": 2.5494950107700482e-05, - "loss": 0.5692, + "loss": 0.5709, "step": 162 }, { "epoch": 0.2944896115627823, - "grad_norm": 0.2549710140850951, + "grad_norm": 0.3587450885360311, "learning_rate": 2.5430631671185616e-05, - "loss": 0.5775, + "loss": 0.5791, "step": 163 }, { "epoch": 0.2962962962962963, - "grad_norm": 0.320108207545845, + "grad_norm": 0.32948214359444944, "learning_rate": 2.5365939734802973e-05, - "loss": 0.5871, + "loss": 0.5892, "step": 164 }, { "epoch": 0.2981029810298103, - "grad_norm": 0.27118178617195304, + "grad_norm": 0.36655920600202585, "learning_rate": 2.5300876615040223e-05, - "loss": 0.5777, + "loss": 0.5802, "step": 165 }, { "epoch": 0.2999096657633243, - "grad_norm": 0.2732065296628375, + "grad_norm": 0.36077426576274396, "learning_rate": 2.523544464167637e-05, - "loss": 0.5842, + "loss": 0.5868, "step": 166 }, { "epoch": 0.3017163504968383, - "grad_norm": 0.2440418264084237, + "grad_norm": 0.35802150689356144, "learning_rate": 2.5169646157698313e-05, - "loss": 0.583, + "loss": 0.5857, "step": 167 }, { "epoch": 0.3035230352303523, - "grad_norm": 0.2400624724182379, + "grad_norm": 0.3262888729605983, "learning_rate": 2.5103483519216964e-05, - "loss": 0.5813, + "loss": 0.5838, "step": 168 }, { "epoch": 0.3053297199638663, - "grad_norm": 0.25905283588713374, + "grad_norm": 0.3362096574051899, "learning_rate": 2.5036959095382875e-05, - "loss": 0.574, + "loss": 0.5755, "step": 169 }, { "epoch": 0.3071364046973803, - "grad_norm": 0.2591290671871167, + "grad_norm": 0.4060212080395927, "learning_rate": 2.4970075268301388e-05, - "loss": 0.5877, + "loss": 0.5893, "step": 170 }, { "epoch": 0.3089430894308943, - "grad_norm": 0.2235510897395504, + "grad_norm": 0.2886329167833011, "learning_rate": 2.4902834432947353e-05, - "loss": 0.5706, + "loss": 0.5733, "step": 171 }, { "epoch": 0.3107497741644083, - "grad_norm": 0.23760061910249497, + "grad_norm": 0.3808717667450013, "learning_rate": 2.4835238997079382e-05, - "loss": 0.5771, + "loss": 0.5795, "step": 172 }, { "epoch": 0.31255645889792233, - "grad_norm": 0.2617951363502666, + "grad_norm": 0.3072237334054852, "learning_rate": 2.4767291381153603e-05, - "loss": 0.5807, + "loss": 0.5832, "step": 173 }, { "epoch": 0.3143631436314363, - "grad_norm": 0.23091611449448254, + "grad_norm": 0.2965837108398263, "learning_rate": 2.4698994018236994e-05, - "loss": 0.5721, + "loss": 0.5747, "step": 174 }, { "epoch": 0.31616982836495033, - "grad_norm": 0.25262572503013164, + "grad_norm": 0.3384696066427267, "learning_rate": 2.4630349353920284e-05, - "loss": 0.5767, + "loss": 0.5784, "step": 175 }, { "epoch": 0.3179765130984643, - "grad_norm": 0.23055115622552902, + "grad_norm": 0.30615079028521514, "learning_rate": 2.4561359846230346e-05, - "loss": 0.5906, + "loss": 0.5923, "step": 176 }, { "epoch": 0.31978319783197834, - "grad_norm": 0.21968694104114944, + "grad_norm": 0.2888166567825809, "learning_rate": 2.4492027965542217e-05, - "loss": 0.5657, + "loss": 0.5679, "step": 177 }, { "epoch": 0.3215898825654923, - "grad_norm": 0.2525014182918365, + "grad_norm": 0.30385385604927617, "learning_rate": 2.44223561944906e-05, - "loss": 0.5724, + "loss": 0.5744, "step": 178 }, { "epoch": 0.32339656729900634, - "grad_norm": 0.2407328065060556, + "grad_norm": 0.28021720457684524, "learning_rate": 2.4352347027881003e-05, - "loss": 0.5825, + "loss": 0.5846, "step": 179 }, { "epoch": 0.3252032520325203, - "grad_norm": 0.2547387307288275, + "grad_norm": 0.30261488499331, "learning_rate": 2.4282002972600382e-05, - "loss": 0.5908, + "loss": 0.5932, "step": 180 }, { "epoch": 0.32700993676603435, - "grad_norm": 0.2263495601766316, + "grad_norm": 0.2851073248221503, "learning_rate": 2.4211326547527377e-05, - "loss": 0.5688, + "loss": 0.5706, "step": 181 }, { "epoch": 0.3288166214995483, - "grad_norm": 0.2789402543570071, + "grad_norm": 0.3111115931546976, "learning_rate": 2.4140320283442125e-05, - "loss": 0.579, + "loss": 0.5813, "step": 182 }, { "epoch": 0.33062330623306235, - "grad_norm": 0.2567309540458022, + "grad_norm": 0.3143186202532528, "learning_rate": 2.4068986722935625e-05, - "loss": 0.5893, + "loss": 0.5924, "step": 183 }, { "epoch": 0.3324299909665763, - "grad_norm": 0.24182513399667138, + "grad_norm": 0.35464611022112247, "learning_rate": 2.3997328420318705e-05, - "loss": 0.5858, + "loss": 0.5885, "step": 184 }, { "epoch": 0.33423667570009036, - "grad_norm": 0.2675738732389291, + "grad_norm": 0.3119101078690158, "learning_rate": 2.3925347941530556e-05, - "loss": 0.5822, + "loss": 0.585, "step": 185 }, { "epoch": 0.33604336043360433, - "grad_norm": 0.23428952659343405, + "grad_norm": 0.3362124015760118, "learning_rate": 2.3853047864046843e-05, - "loss": 0.5779, + "loss": 0.5805, "step": 186 }, { "epoch": 0.33785004516711836, - "grad_norm": 0.26339154899807604, + "grad_norm": 0.31155910265392833, "learning_rate": 2.3780430776787413e-05, - "loss": 0.5747, + "loss": 0.5775, "step": 187 }, { "epoch": 0.33965672990063234, - "grad_norm": 0.21349849308750005, + "grad_norm": 0.30082608710665315, "learning_rate": 2.3707499280023604e-05, - "loss": 0.5747, + "loss": 0.5764, "step": 188 }, { "epoch": 0.34146341463414637, - "grad_norm": 0.26554041110868815, + "grad_norm": 0.3051156622197956, "learning_rate": 2.3634255985285104e-05, - "loss": 0.5825, + "loss": 0.5845, "step": 189 }, { "epoch": 0.34327009936766034, - "grad_norm": 0.23538164196617103, + "grad_norm": 0.37884369981203814, "learning_rate": 2.356070351526648e-05, - "loss": 0.5708, + "loss": 0.5736, "step": 190 }, { "epoch": 0.34507678410117437, - "grad_norm": 0.24516795740455555, + "grad_norm": 0.32478176061158476, "learning_rate": 2.348684450373322e-05, - "loss": 0.5691, + "loss": 0.5721, "step": 191 }, { "epoch": 0.34688346883468835, - "grad_norm": 0.2320939063694949, + "grad_norm": 0.8776768851811568, "learning_rate": 2.3412681595427467e-05, - "loss": 0.5761, + "loss": 0.579, "step": 192 }, { "epoch": 0.3486901535682023, - "grad_norm": 0.24367157013907606, + "grad_norm": 0.4610655284279262, "learning_rate": 2.3338217445973268e-05, - "loss": 0.5776, + "loss": 0.5806, "step": 193 }, { "epoch": 0.35049683830171635, - "grad_norm": 0.2383952654545809, + "grad_norm": 0.32423892822372186, "learning_rate": 2.3263454721781537e-05, - "loss": 0.5793, + "loss": 0.5824, "step": 194 }, { "epoch": 0.3523035230352303, - "grad_norm": 0.21970860044862534, + "grad_norm": 0.3305614412713182, "learning_rate": 2.318839609995453e-05, - "loss": 0.5625, + "loss": 0.5655, "step": 195 }, { "epoch": 0.35411020776874436, - "grad_norm": 0.22393266445368756, + "grad_norm": 0.3088383542670599, "learning_rate": 2.3113044268189995e-05, - "loss": 0.5758, + "loss": 0.579, "step": 196 }, { "epoch": 0.35591689250225833, - "grad_norm": 0.20965253283535495, + "grad_norm": 0.3087829310275476, "learning_rate": 2.303740192468495e-05, - "loss": 0.5754, + "loss": 0.5782, "step": 197 }, { "epoch": 0.35772357723577236, - "grad_norm": 0.2134213495454822, + "grad_norm": 0.5804440405125408, "learning_rate": 2.2961471778039045e-05, - "loss": 0.5521, + "loss": 0.5552, "step": 198 }, { "epoch": 0.35953026196928634, - "grad_norm": 0.2077804399624621, + "grad_norm": 0.3215405528760461, "learning_rate": 2.288525654715757e-05, - "loss": 0.5758, + "loss": 0.5797, "step": 199 }, { "epoch": 0.36133694670280037, - "grad_norm": 0.21089298206975676, + "grad_norm": 0.34958938754043156, "learning_rate": 2.280875896115413e-05, - "loss": 0.5645, + "loss": 0.5676, "step": 200 }, { "epoch": 0.36314363143631434, - "grad_norm": 0.22751400705060473, + "grad_norm": 0.33248083840336357, "learning_rate": 2.2731981759252876e-05, - "loss": 0.5595, + "loss": 0.5624, "step": 201 }, { "epoch": 0.36495031616982837, - "grad_norm": 0.22071355320317226, + "grad_norm": 0.26750582916819504, "learning_rate": 2.2654927690690445e-05, - "loss": 0.5858, + "loss": 0.5887, "step": 202 }, { "epoch": 0.36675700090334235, - "grad_norm": 0.2269131246517602, + "grad_norm": 0.3306138445776805, "learning_rate": 2.257759951461752e-05, - "loss": 0.5733, + "loss": 0.5765, "step": 203 }, { "epoch": 0.3685636856368564, - "grad_norm": 0.27195813360203375, + "grad_norm": 0.3309054658853368, "learning_rate": 2.25e-05, - "loss": 0.5855, + "loss": 0.588, "step": 204 }, { "epoch": 0.37037037037037035, - "grad_norm": 0.24074584104894936, + "grad_norm": 0.3069934073052549, "learning_rate": 2.24221319255199e-05, - "loss": 0.5761, + "loss": 0.5801, "step": 205 }, { "epoch": 0.3721770551038844, - "grad_norm": 0.2643169067368919, + "grad_norm": 0.36963521242301206, "learning_rate": 2.234399807947579e-05, - "loss": 0.5632, + "loss": 0.5669, "step": 206 }, { "epoch": 0.37398373983739835, - "grad_norm": 0.2602452192335381, + "grad_norm": 0.2842636472789742, "learning_rate": 2.2265601259683e-05, - "loss": 0.5713, + "loss": 0.5746, "step": 207 }, { "epoch": 0.3757904245709124, - "grad_norm": 0.22491768260975722, + "grad_norm": 0.2967300023743323, "learning_rate": 2.2186944273373426e-05, - "loss": 0.5589, + "loss": 0.5621, "step": 208 }, { "epoch": 0.37759710930442636, - "grad_norm": 0.23899265916988002, + "grad_norm": 0.30164540888223945, "learning_rate": 2.210802993709498e-05, - "loss": 0.5799, + "loss": 0.5841, "step": 209 }, { "epoch": 0.3794037940379404, - "grad_norm": 0.28185732149789655, + "grad_norm": 0.3020950057384688, "learning_rate": 2.202886107661078e-05, - "loss": 0.576, + "loss": 0.5803, "step": 210 }, { "epoch": 0.38121047877145436, - "grad_norm": 0.20671343969870537, + "grad_norm": 0.30358643623890824, "learning_rate": 2.1949440526797928e-05, - "loss": 0.5761, + "loss": 0.5798, "step": 211 }, { "epoch": 0.3830171635049684, - "grad_norm": 0.245945113158813, + "grad_norm": 0.35616750454359836, "learning_rate": 2.1869771131546015e-05, - "loss": 0.5834, + "loss": 0.587, "step": 212 }, { "epoch": 0.38482384823848237, - "grad_norm": 0.2051130167210825, + "grad_norm": 0.27878546685216377, "learning_rate": 2.178985574365529e-05, - "loss": 0.5837, + "loss": 0.5879, "step": 213 }, { "epoch": 0.3866305329719964, - "grad_norm": 0.25952305018947347, + "grad_norm": 0.32699494080692665, "learning_rate": 2.170969722473449e-05, - "loss": 0.57, + "loss": 0.5738, "step": 214 }, { "epoch": 0.3884372177055104, - "grad_norm": 0.24249808093673167, + "grad_norm": 0.31103539696417715, "learning_rate": 2.1629298445098403e-05, - "loss": 0.5966, + "loss": 0.6008, "step": 215 }, { "epoch": 0.3902439024390244, - "grad_norm": 0.237872387820411, + "grad_norm": 0.32794619585958185, "learning_rate": 2.154866228366505e-05, - "loss": 0.5856, + "loss": 0.5886, "step": 216 }, { "epoch": 0.3920505871725384, - "grad_norm": 0.20572539636470738, + "grad_norm": 0.36007383306015284, "learning_rate": 2.146779162785263e-05, - "loss": 0.5857, + "loss": 0.5896, "step": 217 }, { "epoch": 0.3938572719060524, - "grad_norm": 0.22849509908427684, + "grad_norm": 0.3059835677015186, "learning_rate": 2.138668937347609e-05, - "loss": 0.5775, + "loss": 0.5808, "step": 218 }, { "epoch": 0.3956639566395664, - "grad_norm": 0.21171703597428448, + "grad_norm": 0.33519811127594074, "learning_rate": 2.1305358424643484e-05, - "loss": 0.572, + "loss": 0.5758, "step": 219 }, { "epoch": 0.3974706413730804, - "grad_norm": 0.2333174990121186, + "grad_norm": 0.3173280187406549, "learning_rate": 2.1223801693651927e-05, - "loss": 0.5746, + "loss": 0.5778, "step": 220 }, { "epoch": 0.3992773261065944, - "grad_norm": 0.21728105831361583, + "grad_norm": 0.3249654059010307, "learning_rate": 2.114202210088336e-05, - "loss": 0.5633, + "loss": 0.5665, "step": 221 }, { "epoch": 0.4010840108401084, - "grad_norm": 0.2232469073657137, + "grad_norm": 0.315983288828985, "learning_rate": 2.106002257469993e-05, - "loss": 0.5805, + "loss": 0.5841, "step": 222 }, { "epoch": 0.4028906955736224, - "grad_norm": 0.23039919622981786, + "grad_norm": 0.317439517236036, "learning_rate": 2.0977806051339172e-05, - "loss": 0.5631, + "loss": 0.5661, "step": 223 }, { "epoch": 0.4046973803071364, - "grad_norm": 0.24767592985180426, + "grad_norm": 0.31483264135931766, "learning_rate": 2.0895375474808857e-05, - "loss": 0.5681, + "loss": 0.5714, "step": 224 }, { "epoch": 0.4065040650406504, - "grad_norm": 0.22933111893390376, + "grad_norm": 0.2948623418665935, "learning_rate": 2.0812733796781544e-05, - "loss": 0.5728, + "loss": 0.5765, "step": 225 }, { "epoch": 0.4083107497741644, - "grad_norm": 0.24263450877516124, + "grad_norm": 0.2966240742398221, "learning_rate": 2.0729883976488936e-05, - "loss": 0.5792, + "loss": 0.5831, "step": 226 }, { "epoch": 0.4101174345076784, - "grad_norm": 0.24644389391459087, + "grad_norm": 0.3938546298314616, "learning_rate": 2.064682898061588e-05, - "loss": 0.5836, + "loss": 0.5872, "step": 227 }, { "epoch": 0.41192411924119243, - "grad_norm": 0.23999077805752708, + "grad_norm": 0.2823560010807599, "learning_rate": 2.0563571783194146e-05, - "loss": 0.5697, + "loss": 0.573, "step": 228 }, { "epoch": 0.4137308039747064, - "grad_norm": 0.20338232613186608, + "grad_norm": 0.3226289122791917, "learning_rate": 2.0480115365495928e-05, - "loss": 0.5723, + "loss": 0.5759, "step": 229 }, { "epoch": 0.41553748870822044, - "grad_norm": 0.21451640836421845, + "grad_norm": 0.3116964943536504, "learning_rate": 2.0396462715927107e-05, - "loss": 0.5748, + "loss": 0.5782, "step": 230 }, { "epoch": 0.4173441734417344, - "grad_norm": 0.23674092702209837, + "grad_norm": 0.33810521611187627, "learning_rate": 2.0312616829920222e-05, - "loss": 0.5737, + "loss": 0.5771, "step": 231 }, { "epoch": 0.41915085817524844, - "grad_norm": 0.2040271869891673, + "grad_norm": 0.3046098546864739, "learning_rate": 2.022858070982723e-05, - "loss": 0.5768, + "loss": 0.5802, "step": 232 }, { "epoch": 0.4209575429087624, - "grad_norm": 0.23013110791835603, + "grad_norm": 0.2740022073595928, "learning_rate": 2.0144357364811973e-05, - "loss": 0.5586, + "loss": 0.5615, "step": 233 }, { "epoch": 0.42276422764227645, - "grad_norm": 0.23251714747465213, + "grad_norm": 0.2913067130533732, "learning_rate": 2.0059949810742452e-05, - "loss": 0.5744, + "loss": 0.5771, "step": 234 }, { "epoch": 0.4245709123757904, - "grad_norm": 0.24113338282908972, + "grad_norm": 0.2762685594049124, "learning_rate": 1.997536107008281e-05, - "loss": 0.5696, + "loss": 0.5725, "step": 235 }, { "epoch": 0.42637759710930445, - "grad_norm": 0.217275693069977, + "grad_norm": 0.5121383274688709, "learning_rate": 1.9890594171785128e-05, - "loss": 0.5807, + "loss": 0.5843, "step": 236 }, { "epoch": 0.4281842818428184, - "grad_norm": 0.20589216064741256, + "grad_norm": 0.3344358767381985, "learning_rate": 1.9805652151180945e-05, - "loss": 0.5804, + "loss": 0.5848, "step": 237 }, { "epoch": 0.42999096657633246, - "grad_norm": 0.22359306351640823, + "grad_norm": 0.31383841038930804, "learning_rate": 1.972053804987258e-05, - "loss": 0.5889, + "loss": 0.5924, "step": 238 }, { "epoch": 0.43179765130984643, - "grad_norm": 0.20375725610745427, + "grad_norm": 0.32911204189458476, "learning_rate": 1.963525491562421e-05, - "loss": 0.5701, + "loss": 0.573, "step": 239 }, { "epoch": 0.43360433604336046, - "grad_norm": 0.20454609097134233, + "grad_norm": 0.3543469488051476, "learning_rate": 1.954980580225275e-05, - "loss": 0.5702, + "loss": 0.5741, "step": 240 }, { "epoch": 0.43541102077687444, - "grad_norm": 0.19497163244699564, + "grad_norm": 0.2769864406337721, "learning_rate": 1.946419376951848e-05, - "loss": 0.5693, + "loss": 0.5736, "step": 241 }, { "epoch": 0.4372177055103884, - "grad_norm": 0.21551459772294188, + "grad_norm": 0.2772504692036322, "learning_rate": 1.9378421883015505e-05, - "loss": 0.57, + "loss": 0.5737, "step": 242 }, { "epoch": 0.43902439024390244, - "grad_norm": 0.23200634978654613, + "grad_norm": 0.27865104121234274, "learning_rate": 1.9292493214061953e-05, - "loss": 0.5741, + "loss": 0.5787, "step": 243 }, { "epoch": 0.4408310749774164, - "grad_norm": 0.1884963011054949, + "grad_norm": 0.33872737991020524, "learning_rate": 1.9206410839590042e-05, - "loss": 0.5804, + "loss": 0.5849, "step": 244 }, { "epoch": 0.44263775971093045, - "grad_norm": 0.25404075829436523, + "grad_norm": 0.29138500205569745, "learning_rate": 1.9120177842035853e-05, - "loss": 0.5726, + "loss": 0.5769, "step": 245 }, { "epoch": 0.4444444444444444, - "grad_norm": 0.22097319274710464, + "grad_norm": 0.33436184779749656, "learning_rate": 1.9033797309228984e-05, - "loss": 0.5794, + "loss": 0.584, "step": 246 }, { "epoch": 0.44625112917795845, - "grad_norm": 0.19246576672187318, + "grad_norm": 0.3402673798099814, "learning_rate": 1.8947272334281977e-05, - "loss": 0.5616, + "loss": 0.5654, "step": 247 }, { "epoch": 0.4480578139114724, - "grad_norm": 0.22700593999033858, + "grad_norm": 0.30708428061855253, "learning_rate": 1.8860606015479537e-05, - "loss": 0.5801, + "loss": 0.5846, "step": 248 }, { "epoch": 0.44986449864498645, - "grad_norm": 0.1934681513313933, + "grad_norm": 0.31233742569187306, "learning_rate": 1.877380145616763e-05, - "loss": 0.5632, + "loss": 0.5679, "step": 249 }, { "epoch": 0.45167118337850043, - "grad_norm": 0.21167453440737605, + "grad_norm": 0.3506675404562718, "learning_rate": 1.868686176464232e-05, - "loss": 0.5781, + "loss": 0.5836, "step": 250 }, { "epoch": 0.45347786811201446, - "grad_norm": 0.23469761452053498, + "grad_norm": 0.29769387160985, "learning_rate": 1.8599790054038487e-05, - "loss": 0.5653, + "loss": 0.57, "step": 251 }, { "epoch": 0.45528455284552843, - "grad_norm": 0.20579485783023402, + "grad_norm": 0.3272145932882665, "learning_rate": 1.8512589442218358e-05, - "loss": 0.5472, + "loss": 0.5523, "step": 252 }, { "epoch": 0.45709123757904246, - "grad_norm": 0.20450811714095307, + "grad_norm": 0.3089325909787182, "learning_rate": 1.8425263051659838e-05, - "loss": 0.5701, + "loss": 0.5751, "step": 253 }, { "epoch": 0.45889792231255644, - "grad_norm": 0.2795902989622695, + "grad_norm": 0.36927293242924614, "learning_rate": 1.8337814009344716e-05, - "loss": 0.5692, + "loss": 0.5736, "step": 254 }, { "epoch": 0.46070460704607047, - "grad_norm": 0.26043149982302555, + "grad_norm": 0.31333756675521174, "learning_rate": 1.8250245446646707e-05, - "loss": 0.5627, + "loss": 0.5667, "step": 255 }, { "epoch": 0.46251129177958444, - "grad_norm": 0.2115149201329313, + "grad_norm": 0.3194088697708899, "learning_rate": 1.8162560499219286e-05, - "loss": 0.563, + "loss": 0.567, "step": 256 }, { "epoch": 0.4643179765130985, - "grad_norm": 0.2863165296718786, + "grad_norm": 0.3358076681829595, "learning_rate": 1.807476230688346e-05, - "loss": 0.5726, + "loss": 0.5768, "step": 257 }, { "epoch": 0.46612466124661245, - "grad_norm": 0.22723135404479125, + "grad_norm": 0.3507131609472825, "learning_rate": 1.7986854013515274e-05, - "loss": 0.5682, + "loss": 0.573, "step": 258 }, { "epoch": 0.4679313459801265, - "grad_norm": 0.21269365868540174, + "grad_norm": 0.333864958674876, "learning_rate": 1.78988387669333e-05, - "loss": 0.5718, + "loss": 0.5759, "step": 259 }, { "epoch": 0.46973803071364045, - "grad_norm": 0.27706961689827647, + "grad_norm": 0.33675575200680075, "learning_rate": 1.781071971878587e-05, - "loss": 0.5852, + "loss": 0.5908, "step": 260 }, { "epoch": 0.4715447154471545, - "grad_norm": 0.21919053031721478, + "grad_norm": 0.32765202872639676, "learning_rate": 1.7722500024438244e-05, - "loss": 0.5766, + "loss": 0.5812, "step": 261 }, { "epoch": 0.47335140018066846, - "grad_norm": 0.2458759125768864, + "grad_norm": 0.36381544719415326, "learning_rate": 1.7634182842859628e-05, - "loss": 0.565, + "loss": 0.5693, "step": 262 }, { "epoch": 0.4751580849141825, - "grad_norm": 0.2489757112909537, + "grad_norm": 0.4851712003668751, "learning_rate": 1.7545771336510033e-05, - "loss": 0.5695, + "loss": 0.5736, "step": 263 }, { "epoch": 0.47696476964769646, - "grad_norm": 0.2101979687684982, + "grad_norm": 0.41291312521131196, "learning_rate": 1.7457268671227067e-05, - "loss": 0.5759, + "loss": 0.5784, "step": 264 }, { "epoch": 0.4787714543812105, - "grad_norm": 0.2500578266252668, + "grad_norm": 0.2955323853549614, "learning_rate": 1.736867801611254e-05, - "loss": 0.5569, + "loss": 0.5607, "step": 265 }, { "epoch": 0.48057813911472447, - "grad_norm": 0.19262415552045406, + "grad_norm": 0.37616325609332063, "learning_rate": 1.728000254341901e-05, - "loss": 0.5813, + "loss": 0.5861, "step": 266 }, { "epoch": 0.4823848238482385, - "grad_norm": 0.19478984136741012, + "grad_norm": 0.29381479811591366, "learning_rate": 1.7191245428436175e-05, - "loss": 0.5653, + "loss": 0.5699, "step": 267 }, { "epoch": 0.48419150858175247, - "grad_norm": 0.23871243123226799, + "grad_norm": 0.32495443011128766, "learning_rate": 1.7102409849377188e-05, - "loss": 0.5661, + "loss": 0.5697, "step": 268 }, { "epoch": 0.4859981933152665, - "grad_norm": 0.17665577180387385, + "grad_norm": 0.2624608390186021, "learning_rate": 1.7013498987264832e-05, - "loss": 0.5727, + "loss": 0.5769, "step": 269 }, { "epoch": 0.4878048780487805, - "grad_norm": 0.21091637968705407, + "grad_norm": 0.30513009228861504, "learning_rate": 1.6924516025817636e-05, - "loss": 0.5544, + "loss": 0.5586, "step": 270 }, { "epoch": 0.4896115627822945, - "grad_norm": 0.25253202754252524, + "grad_norm": 0.2900302075914806, "learning_rate": 1.683546415133584e-05, - "loss": 0.5761, + "loss": 0.5808, "step": 271 }, { "epoch": 0.4914182475158085, - "grad_norm": 0.2150474071339765, + "grad_norm": 0.30132501927469585, "learning_rate": 1.6746346552587342e-05, - "loss": 0.5704, + "loss": 0.5759, "step": 272 }, { "epoch": 0.4932249322493225, - "grad_norm": 0.24422291161795256, + "grad_norm": 0.3040014918275698, "learning_rate": 1.665716642069349e-05, - "loss": 0.5785, + "loss": 0.5823, "step": 273 }, { "epoch": 0.4950316169828365, - "grad_norm": 0.19465875800542706, + "grad_norm": 0.2581541065065194, "learning_rate": 1.6567926949014805e-05, - "loss": 0.5658, + "loss": 0.5705, "step": 274 }, { "epoch": 0.4968383017163505, - "grad_norm": 0.20059784342394038, + "grad_norm": 0.3506532731278841, "learning_rate": 1.6478631333036655e-05, - "loss": 0.5607, + "loss": 0.5654, "step": 275 }, { "epoch": 0.4986449864498645, - "grad_norm": 0.19428493006954836, + "grad_norm": 0.2841799020533544, "learning_rate": 1.638928277025482e-05, - "loss": 0.5523, + "loss": 0.5569, "step": 276 }, { "epoch": 0.5004516711833785, - "grad_norm": 0.190112158192132, + "grad_norm": 0.3147444372936614, "learning_rate": 1.6299884460061005e-05, - "loss": 0.5643, + "loss": 0.5683, "step": 277 }, { "epoch": 0.5022583559168925, - "grad_norm": 0.18317648770164868, + "grad_norm": 0.28903827689512285, "learning_rate": 1.621043960362826e-05, - "loss": 0.5509, + "loss": 0.5549, "step": 278 }, { "epoch": 0.5040650406504065, - "grad_norm": 0.20777757920037251, + "grad_norm": 0.31036698418346015, "learning_rate": 1.6120951403796367e-05, - "loss": 0.5659, + "loss": 0.5704, "step": 279 }, { "epoch": 0.5058717253839206, - "grad_norm": 0.24714269404949812, + "grad_norm": 0.31137986469945766, "learning_rate": 1.603142306495714e-05, - "loss": 0.5625, + "loss": 0.5661, "step": 280 }, { "epoch": 0.5076784101174345, - "grad_norm": 0.1771707758636966, + "grad_norm": 0.2971208151403228, "learning_rate": 1.5941857792939702e-05, - "loss": 0.5629, + "loss": 0.5673, "step": 281 }, { "epoch": 0.5094850948509485, - "grad_norm": 0.19209684745372535, + "grad_norm": 0.27859789351234376, "learning_rate": 1.585225879489567e-05, - "loss": 0.5577, + "loss": 0.562, "step": 282 }, { "epoch": 0.5112917795844625, - "grad_norm": 0.22597515112664288, + "grad_norm": 0.30768743394999065, "learning_rate": 1.5762629279184326e-05, - "loss": 0.5728, + "loss": 0.5779, "step": 283 }, { "epoch": 0.5130984643179766, - "grad_norm": 0.17250544743705307, + "grad_norm": 0.29234292940585527, "learning_rate": 1.5672972455257726e-05, - "loss": 0.5611, + "loss": 0.5662, "step": 284 }, { "epoch": 0.5149051490514905, - "grad_norm": 0.20410988336988065, + "grad_norm": 0.32303845185987906, "learning_rate": 1.5583291533545775e-05, - "loss": 0.5557, + "loss": 0.5603, "step": 285 }, { "epoch": 0.5167118337850045, - "grad_norm": 0.20391700905101098, + "grad_norm": 0.2994101822443779, "learning_rate": 1.549358972534128e-05, - "loss": 0.5538, + "loss": 0.5577, "step": 286 }, { "epoch": 0.5185185185185185, - "grad_norm": 0.18259503875294242, + "grad_norm": 0.3225727799664951, "learning_rate": 1.5403870242684942e-05, - "loss": 0.5754, + "loss": 0.5794, "step": 287 }, { "epoch": 0.5203252032520326, - "grad_norm": 0.18696008552172258, + "grad_norm": 0.2923618155390653, "learning_rate": 1.5314136298250355e-05, - "loss": 0.5723, + "loss": 0.577, "step": 288 }, { "epoch": 0.5221318879855466, - "grad_norm": 0.19653948540679103, + "grad_norm": 0.29720244961677544, "learning_rate": 1.5224391105228956e-05, - "loss": 0.5565, + "loss": 0.5603, "step": 289 }, { "epoch": 0.5239385727190605, - "grad_norm": 0.17253613754365116, + "grad_norm": 0.2989180878702361, "learning_rate": 1.5134637877214968e-05, - "loss": 0.5682, + "loss": 0.5727, "step": 290 }, { "epoch": 0.5257452574525745, - "grad_norm": 0.2056061343083173, + "grad_norm": 0.25519500575209336, "learning_rate": 1.5044879828090346e-05, - "loss": 0.5727, + "loss": 0.5773, "step": 291 }, { "epoch": 0.5275519421860885, - "grad_norm": 0.19894424925026757, + "grad_norm": 0.2830864583495515, "learning_rate": 1.4955120171909658e-05, - "loss": 0.5763, + "loss": 0.5816, "step": 292 }, { "epoch": 0.5293586269196026, - "grad_norm": 0.1659275906965013, + "grad_norm": 0.2764753853248595, "learning_rate": 1.4865362122785031e-05, - "loss": 0.5689, + "loss": 0.5744, "step": 293 }, { "epoch": 0.5311653116531165, - "grad_norm": 0.17986609793129524, + "grad_norm": 0.3547980154145159, "learning_rate": 1.4775608894771048e-05, - "loss": 0.5612, + "loss": 0.5664, "step": 294 }, { "epoch": 0.5329719963866305, - "grad_norm": 0.232250278692599, + "grad_norm": 0.29803300652072573, "learning_rate": 1.4685863701749648e-05, - "loss": 0.5589, + "loss": 0.5639, "step": 295 }, { "epoch": 0.5347786811201445, - "grad_norm": 0.22464497097731725, + "grad_norm": 0.290491851084967, "learning_rate": 1.4596129757315062e-05, - "loss": 0.5628, + "loss": 0.5681, "step": 296 }, { "epoch": 0.5365853658536586, - "grad_norm": 0.1952420747386217, + "grad_norm": 0.3389739871118521, "learning_rate": 1.4506410274658718e-05, - "loss": 0.5682, + "loss": 0.5726, "step": 297 }, { "epoch": 0.5383920505871725, - "grad_norm": 0.2106666884744506, + "grad_norm": 0.3246943570187164, "learning_rate": 1.441670846645423e-05, - "loss": 0.5627, + "loss": 0.5676, "step": 298 }, { "epoch": 0.5401987353206865, - "grad_norm": 0.21061181918065086, + "grad_norm": 0.3242408286020586, "learning_rate": 1.4327027544742281e-05, - "loss": 0.5783, + "loss": 0.5839, "step": 299 }, { "epoch": 0.5420054200542005, - "grad_norm": 0.1861290798519661, + "grad_norm": 0.28501826222047294, "learning_rate": 1.4237370720815675e-05, - "loss": 0.5664, + "loss": 0.5717, "step": 300 }, { "epoch": 0.5438121047877146, - "grad_norm": 0.18862693026886343, + "grad_norm": 0.2801125597531159, "learning_rate": 1.4147741205104336e-05, - "loss": 0.5683, + "loss": 0.5742, "step": 301 }, { "epoch": 0.5456187895212286, - "grad_norm": 0.18676460565472633, + "grad_norm": 0.2865662680741476, "learning_rate": 1.40581422070603e-05, - "loss": 0.5796, + "loss": 0.5859, "step": 302 }, { "epoch": 0.5474254742547425, - "grad_norm": 0.1752400383590249, + "grad_norm": 0.2880283177460633, "learning_rate": 1.3968576935042864e-05, - "loss": 0.5608, + "loss": 0.5666, "step": 303 }, { "epoch": 0.5492321589882565, - "grad_norm": 0.18836688433228918, + "grad_norm": 0.29776279014261486, "learning_rate": 1.3879048596203637e-05, - "loss": 0.5561, + "loss": 0.5614, "step": 304 }, { "epoch": 0.5510388437217706, - "grad_norm": 0.19530506408089982, + "grad_norm": 0.3046214139843183, "learning_rate": 1.3789560396371742e-05, - "loss": 0.5622, + "loss": 0.5679, "step": 305 }, { "epoch": 0.5528455284552846, - "grad_norm": 0.17965863699921675, + "grad_norm": 0.31091017665288007, "learning_rate": 1.3700115539938995e-05, - "loss": 0.5605, + "loss": 0.5652, "step": 306 }, { "epoch": 0.5546522131887985, - "grad_norm": 0.21750349150527712, + "grad_norm": 0.3342055750377691, "learning_rate": 1.3610717229745183e-05, - "loss": 0.578, + "loss": 0.5831, "step": 307 }, { "epoch": 0.5564588979223125, - "grad_norm": 0.19147158863524186, + "grad_norm": 0.2825943487551394, "learning_rate": 1.3521368666963348e-05, - "loss": 0.5572, + "loss": 0.5622, "step": 308 }, { "epoch": 0.5582655826558266, - "grad_norm": 0.18608236035974007, + "grad_norm": 0.42463521342108723, "learning_rate": 1.3432073050985201e-05, - "loss": 0.5669, + "loss": 0.5719, "step": 309 }, { "epoch": 0.5600722673893406, - "grad_norm": 0.18440593641158914, + "grad_norm": 0.24690027892733366, "learning_rate": 1.3342833579306511e-05, - "loss": 0.5492, + "loss": 0.5537, "step": 310 }, { "epoch": 0.5618789521228545, - "grad_norm": 0.1843799787511743, + "grad_norm": 0.2782245278873188, "learning_rate": 1.3253653447412658e-05, - "loss": 0.5685, + "loss": 0.5738, "step": 311 }, { "epoch": 0.5636856368563685, - "grad_norm": 0.20261725696919525, + "grad_norm": 0.269514288271004, "learning_rate": 1.3164535848664167e-05, - "loss": 0.5674, + "loss": 0.5728, "step": 312 }, { "epoch": 0.5654923215898826, - "grad_norm": 0.17558184672847507, + "grad_norm": 0.28864669964933554, "learning_rate": 1.3075483974182368e-05, - "loss": 0.5684, + "loss": 0.5731, "step": 313 }, { "epoch": 0.5672990063233966, - "grad_norm": 0.17907331587608558, + "grad_norm": 0.2612891164500954, "learning_rate": 1.2986501012735174e-05, - "loss": 0.5609, + "loss": 0.5663, "step": 314 }, { "epoch": 0.5691056910569106, - "grad_norm": 0.1934736256775333, + "grad_norm": 0.2551741665283283, "learning_rate": 1.2897590150622812e-05, - "loss": 0.5613, + "loss": 0.5662, "step": 315 }, { "epoch": 0.5709123757904245, - "grad_norm": 0.17567517580021857, + "grad_norm": 0.28291261096839593, "learning_rate": 1.2808754571563827e-05, - "loss": 0.5573, + "loss": 0.5623, "step": 316 }, { "epoch": 0.5727190605239386, - "grad_norm": 0.19690748080324788, + "grad_norm": 0.33675663529353467, "learning_rate": 1.2719997456580993e-05, - "loss": 0.5525, + "loss": 0.5573, "step": 317 }, { "epoch": 0.5745257452574526, - "grad_norm": 0.20212907631156332, + "grad_norm": 0.242227937824476, "learning_rate": 1.2631321983887464e-05, - "loss": 0.5654, + "loss": 0.5701, "step": 318 }, { "epoch": 0.5763324299909666, - "grad_norm": 0.19608541723414, + "grad_norm": 1.7361889228979208, "learning_rate": 1.2542731328772936e-05, - "loss": 0.5744, + "loss": 0.5796, "step": 319 }, { "epoch": 0.5781391147244805, - "grad_norm": 0.20857428025225674, + "grad_norm": 0.29371094872544395, "learning_rate": 1.2454228663489967e-05, - "loss": 0.5569, + "loss": 0.5622, "step": 320 }, { "epoch": 0.5799457994579946, - "grad_norm": 0.17609191697173338, + "grad_norm": 0.275215385186721, "learning_rate": 1.2365817157140373e-05, - "loss": 0.5541, + "loss": 0.5595, "step": 321 }, { "epoch": 0.5817524841915086, - "grad_norm": 0.18308421719879198, + "grad_norm": 0.2925560707999663, "learning_rate": 1.2277499975561755e-05, - "loss": 0.5753, + "loss": 0.5807, "step": 322 }, { "epoch": 0.5835591689250226, - "grad_norm": 0.17484046660309832, + "grad_norm": 0.4068710625406492, "learning_rate": 1.2189280281214128e-05, - "loss": 0.5609, + "loss": 0.566, "step": 323 }, { "epoch": 0.5853658536585366, - "grad_norm": 0.17252551757672757, + "grad_norm": 0.2628331351797529, "learning_rate": 1.2101161233066704e-05, - "loss": 0.5419, + "loss": 0.5469, "step": 324 }, { "epoch": 0.5871725383920506, - "grad_norm": 0.17553463607522976, + "grad_norm": 0.2911285026218736, "learning_rate": 1.201314598648473e-05, - "loss": 0.5591, + "loss": 0.565, "step": 325 }, { "epoch": 0.5889792231255646, - "grad_norm": 0.17326333720847636, + "grad_norm": 0.2831006227507552, "learning_rate": 1.1925237693116546e-05, - "loss": 0.5689, + "loss": 0.5747, "step": 326 }, { "epoch": 0.5907859078590786, - "grad_norm": 0.18687445963344007, + "grad_norm": 0.26598868778075935, "learning_rate": 1.1837439500780718e-05, - "loss": 0.5553, + "loss": 0.5602, "step": 327 }, { "epoch": 0.5925925925925926, - "grad_norm": 0.19428514170213504, + "grad_norm": 0.3558168004733384, "learning_rate": 1.1749754553353297e-05, - "loss": 0.562, + "loss": 0.5676, "step": 328 }, { "epoch": 0.5943992773261066, - "grad_norm": 0.17110021647920834, + "grad_norm": 0.25067605853110625, "learning_rate": 1.1662185990655285e-05, - "loss": 0.5518, + "loss": 0.5569, "step": 329 }, { "epoch": 0.5962059620596206, - "grad_norm": 0.18984185404031334, + "grad_norm": 0.3465570264289409, "learning_rate": 1.1574736948340163e-05, - "loss": 0.559, + "loss": 0.5644, "step": 330 }, { "epoch": 0.5980126467931346, - "grad_norm": 0.16874712774452827, + "grad_norm": 0.2555984652347325, "learning_rate": 1.1487410557781642e-05, - "loss": 0.5632, + "loss": 0.5689, "step": 331 }, { "epoch": 0.5998193315266486, - "grad_norm": 0.1806801244410073, + "grad_norm": 1.3045885103794765, "learning_rate": 1.140020994596151e-05, - "loss": 0.5467, + "loss": 0.5525, "step": 332 }, { "epoch": 0.6016260162601627, - "grad_norm": 0.17626693416571312, + "grad_norm": 0.269714982616486, "learning_rate": 1.1313138235357683e-05, - "loss": 0.5674, + "loss": 0.5735, "step": 333 }, { "epoch": 0.6034327009936766, - "grad_norm": 0.1770136401046488, + "grad_norm": 0.2468409802180684, "learning_rate": 1.1226198543832373e-05, - "loss": 0.5798, + "loss": 0.5864, "step": 334 }, { "epoch": 0.6052393857271906, - "grad_norm": 0.1910726991764081, + "grad_norm": 0.2359964626102903, "learning_rate": 1.1139393984520467e-05, - "loss": 0.563, + "loss": 0.5688, "step": 335 }, { "epoch": 0.6070460704607046, - "grad_norm": 0.1718176664612323, + "grad_norm": 0.2792218994212215, "learning_rate": 1.1052727665718027e-05, - "loss": 0.5596, + "loss": 0.5651, "step": 336 }, { "epoch": 0.6088527551942186, - "grad_norm": 0.17250346288280585, + "grad_norm": 0.24660150003449235, "learning_rate": 1.0966202690771015e-05, - "loss": 0.5611, + "loss": 0.5673, "step": 337 }, { "epoch": 0.6106594399277326, - "grad_norm": 0.19111207535840807, + "grad_norm": 0.28553688739699007, "learning_rate": 1.087982215796415e-05, - "loss": 0.5572, + "loss": 0.5628, "step": 338 }, { "epoch": 0.6124661246612466, - "grad_norm": 0.1686795998299278, + "grad_norm": 0.24555542198671096, "learning_rate": 1.079358916040996e-05, - "loss": 0.5657, + "loss": 0.5717, "step": 339 }, { "epoch": 0.6142728093947606, - "grad_norm": 0.16862020455628596, + "grad_norm": 0.25213878756115177, "learning_rate": 1.0707506785938052e-05, - "loss": 0.5611, + "loss": 0.567, "step": 340 }, { "epoch": 0.6160794941282746, - "grad_norm": 0.1637656583385122, + "grad_norm": 0.2316189652392717, "learning_rate": 1.0621578116984501e-05, - "loss": 0.5593, + "loss": 0.5649, "step": 341 }, { "epoch": 0.6178861788617886, - "grad_norm": 0.16170288648566758, + "grad_norm": 0.2642230456418394, "learning_rate": 1.0535806230481525e-05, - "loss": 0.5628, + "loss": 0.5682, "step": 342 }, { "epoch": 0.6196928635953026, - "grad_norm": 0.1717924247456263, + "grad_norm": 0.25607403892541075, "learning_rate": 1.0450194197747252e-05, - "loss": 0.5567, + "loss": 0.5626, "step": 343 }, { "epoch": 0.6214995483288166, - "grad_norm": 0.16174500558159677, + "grad_norm": 0.2454366513142169, "learning_rate": 1.036474508437579e-05, - "loss": 0.5464, + "loss": 0.5524, "step": 344 }, { "epoch": 0.6233062330623306, - "grad_norm": 0.16381172759058904, + "grad_norm": 0.25692513164041986, "learning_rate": 1.0279461950127422e-05, - "loss": 0.5608, + "loss": 0.5666, "step": 345 }, { "epoch": 0.6251129177958447, - "grad_norm": 0.17001972937010654, + "grad_norm": 0.2567223339593557, "learning_rate": 1.0194347848819055e-05, - "loss": 0.5623, + "loss": 0.5675, "step": 346 }, { "epoch": 0.6269196025293586, - "grad_norm": 0.17303311757989492, + "grad_norm": 0.24303756149752093, "learning_rate": 1.0109405828214872e-05, - "loss": 0.5482, + "loss": 0.5534, "step": 347 }, { "epoch": 0.6287262872628726, - "grad_norm": 0.1882537905657933, + "grad_norm": 0.3469962933198629, "learning_rate": 1.0024638929917192e-05, - "loss": 0.5492, + "loss": 0.5543, "step": 348 }, { "epoch": 0.6305329719963866, - "grad_norm": 0.15457360070696674, + "grad_norm": 0.29482139739377156, "learning_rate": 9.940050189257552e-06, - "loss": 0.5521, + "loss": 0.5582, "step": 349 }, { "epoch": 0.6323396567299007, - "grad_norm": 0.15615688544922562, + "grad_norm": 0.2543061317007475, "learning_rate": 9.85564263518803e-06, - "loss": 0.5643, + "loss": 0.5703, "step": 350 }, { "epoch": 0.6341463414634146, - "grad_norm": 0.16319062879774318, + "grad_norm": 0.2598563996097049, "learning_rate": 9.771419290172776e-06, - "loss": 0.554, + "loss": 0.5594, "step": 351 }, { "epoch": 0.6359530261969286, - "grad_norm": 0.15770060432248675, + "grad_norm": 0.28480208198026236, "learning_rate": 9.68738317007978e-06, - "loss": 0.5682, + "loss": 0.5738, "step": 352 }, { "epoch": 0.6377597109304426, - "grad_norm": 0.16898427412267347, + "grad_norm": 0.24709090613155346, "learning_rate": 9.6035372840729e-06, - "loss": 0.568, + "loss": 0.5738, "step": 353 }, { "epoch": 0.6395663956639567, - "grad_norm": 0.15918955141776597, + "grad_norm": 0.2671031729169573, "learning_rate": 9.519884634504074e-06, - "loss": 0.5631, + "loss": 0.5691, "step": 354 }, { "epoch": 0.6413730803974707, - "grad_norm": 0.1635896927311437, + "grad_norm": 0.22856827107644548, "learning_rate": 9.436428216805862e-06, - "loss": 0.5524, + "loss": 0.5579, "step": 355 }, { "epoch": 0.6431797651309846, - "grad_norm": 0.15329421947992738, + "grad_norm": 0.2544739414287776, "learning_rate": 9.35317101938412e-06, - "loss": 0.566, + "loss": 0.5723, "step": 356 }, { "epoch": 0.6449864498644986, - "grad_norm": 0.18290183186305234, + "grad_norm": 0.25959041144887784, "learning_rate": 9.270116023511063e-06, - "loss": 0.5676, + "loss": 0.5736, "step": 357 }, { "epoch": 0.6467931345980127, - "grad_norm": 0.18253658487556157, + "grad_norm": 0.4509629870171336, "learning_rate": 9.187266203218457e-06, - "loss": 0.5729, + "loss": 0.5788, "step": 358 }, { "epoch": 0.6485998193315267, - "grad_norm": 0.1791341463101353, + "grad_norm": 0.2554658264431559, "learning_rate": 9.104624525191147e-06, - "loss": 0.546, + "loss": 0.5516, "step": 359 }, { "epoch": 0.6504065040650406, - "grad_norm": 0.14674620232896526, + "grad_norm": 0.22730131545870966, "learning_rate": 9.022193948660825e-06, - "loss": 0.5548, + "loss": 0.5601, "step": 360 }, { "epoch": 0.6522131887985546, - "grad_norm": 0.19660949623094373, + "grad_norm": 0.2705934074611756, "learning_rate": 8.939977425300072e-06, - "loss": 0.56, + "loss": 0.5657, "step": 361 }, { "epoch": 0.6540198735320687, - "grad_norm": 0.1876501178479123, + "grad_norm": 0.23508126204947832, "learning_rate": 8.857977899116645e-06, - "loss": 0.5717, + "loss": 0.5772, "step": 362 }, { "epoch": 0.6558265582655827, - "grad_norm": 0.21507796351172995, + "grad_norm": 0.2670484472522681, "learning_rate": 8.776198306348072e-06, - "loss": 0.5618, + "loss": 0.5679, "step": 363 }, { "epoch": 0.6576332429990966, - "grad_norm": 0.18228320215554383, + "grad_norm": 0.25807775800350113, "learning_rate": 8.69464157535652e-06, - "loss": 0.5612, + "loss": 0.5678, "step": 364 }, { "epoch": 0.6594399277326106, - "grad_norm": 0.1757206345144353, + "grad_norm": 0.4143373428623339, "learning_rate": 8.61331062652391e-06, - "loss": 0.5534, + "loss": 0.5593, "step": 365 }, { "epoch": 0.6612466124661247, - "grad_norm": 0.19026770352731234, + "grad_norm": 0.25432160506360196, "learning_rate": 8.532208372147376e-06, - "loss": 0.5746, + "loss": 0.5811, "step": 366 }, { "epoch": 0.6630532971996387, - "grad_norm": 0.1865886995183675, + "grad_norm": 0.2218223608724009, "learning_rate": 8.451337716334953e-06, - "loss": 0.5556, + "loss": 0.5612, "step": 367 }, { "epoch": 0.6648599819331527, - "grad_norm": 0.1797695927669861, + "grad_norm": 0.26338098583116515, "learning_rate": 8.370701554901603e-06, - "loss": 0.5675, + "loss": 0.573, "step": 368 }, { "epoch": 0.6666666666666666, - "grad_norm": 0.17670482413719021, + "grad_norm": 0.27305656638372205, "learning_rate": 8.290302775265509e-06, - "loss": 0.5686, + "loss": 0.5743, "step": 369 }, { "epoch": 0.6684733514001807, - "grad_norm": 0.1701577263486839, + "grad_norm": 0.28354813004240265, "learning_rate": 8.210144256344714e-06, - "loss": 0.5668, + "loss": 0.5732, "step": 370 }, { "epoch": 0.6702800361336947, - "grad_norm": 0.2061129978805257, + "grad_norm": 0.2540985448161381, "learning_rate": 8.130228868453989e-06, - "loss": 0.555, + "loss": 0.5604, "step": 371 }, { "epoch": 0.6720867208672087, - "grad_norm": 0.1969046824239301, + "grad_norm": 0.23334255114709357, "learning_rate": 8.050559473202078e-06, - "loss": 0.5727, + "loss": 0.578, "step": 372 }, { "epoch": 0.6738934056007226, - "grad_norm": 0.15311615793545202, + "grad_norm": 0.23245111955419623, "learning_rate": 7.971138923389221e-06, - "loss": 0.5583, + "loss": 0.5634, "step": 373 }, { "epoch": 0.6757000903342367, - "grad_norm": 0.18976350186526125, + "grad_norm": 0.33850041377146883, "learning_rate": 7.89197006290502e-06, - "loss": 0.5473, + "loss": 0.5532, "step": 374 }, { "epoch": 0.6775067750677507, - "grad_norm": 0.1879352111528028, + "grad_norm": 0.24306579520579513, "learning_rate": 7.813055726626579e-06, - "loss": 0.5619, + "loss": 0.567, "step": 375 }, { "epoch": 0.6793134598012647, - "grad_norm": 0.20499323771840253, + "grad_norm": 0.23084325043799506, "learning_rate": 7.734398740316999e-06, - "loss": 0.5467, + "loss": 0.552, "step": 376 }, { "epoch": 0.6811201445347786, - "grad_norm": 0.17137028091933101, + "grad_norm": 0.24330485181839703, "learning_rate": 7.65600192052422e-06, - "loss": 0.563, + "loss": 0.5686, "step": 377 }, { "epoch": 0.6829268292682927, - "grad_norm": 0.18099761803792097, + "grad_norm": 0.2537299633849379, "learning_rate": 7.577868074480107e-06, - "loss": 0.5634, + "loss": 0.5695, "step": 378 }, { "epoch": 0.6847335140018067, - "grad_norm": 0.19896866364484633, + "grad_norm": 0.2324650195741195, "learning_rate": 7.500000000000004e-06, - "loss": 0.5664, + "loss": 0.5723, "step": 379 }, { "epoch": 0.6865401987353207, - "grad_norm": 0.1884102435991128, + "grad_norm": 0.2559577756429273, "learning_rate": 7.422400485382484e-06, - "loss": 0.5473, + "loss": 0.553, "step": 380 }, { "epoch": 0.6883468834688347, - "grad_norm": 0.17513345385135934, + "grad_norm": 0.25311375180891915, "learning_rate": 7.345072309309557e-06, - "loss": 0.5638, + "loss": 0.5694, "step": 381 }, { "epoch": 0.6901535682023487, - "grad_norm": 0.1733806383727063, + "grad_norm": 0.2344778183121317, "learning_rate": 7.268018240747129e-06, - "loss": 0.5515, + "loss": 0.5564, "step": 382 }, { "epoch": 0.6919602529358627, - "grad_norm": 0.18213237487884423, + "grad_norm": 0.21551852417528822, "learning_rate": 7.191241038845876e-06, - "loss": 0.5704, + "loss": 0.5767, "step": 383 }, { "epoch": 0.6937669376693767, - "grad_norm": 0.20160376104741037, + "grad_norm": 0.2715893346415364, "learning_rate": 7.114743452842428e-06, - "loss": 0.5681, + "loss": 0.5738, "step": 384 }, { "epoch": 0.6955736224028907, - "grad_norm": 0.14133588164420194, + "grad_norm": 0.23073802447177677, "learning_rate": 7.038528221960959e-06, - "loss": 0.5504, + "loss": 0.5558, "step": 385 }, { "epoch": 0.6973803071364046, - "grad_norm": 0.1554959410199801, + "grad_norm": 0.240908436382785, "learning_rate": 6.962598075315047e-06, - "loss": 0.5622, + "loss": 0.5677, "step": 386 }, { "epoch": 0.6991869918699187, - "grad_norm": 0.17804707625227728, + "grad_norm": 0.256554182250642, "learning_rate": 6.886955731810009e-06, - "loss": 0.5622, + "loss": 0.5679, "step": 387 }, { "epoch": 0.7009936766034327, - "grad_norm": 0.1770700659186734, + "grad_norm": 0.39527336236612054, "learning_rate": 6.811603900045477e-06, - "loss": 0.5612, + "loss": 0.5663, "step": 388 }, { "epoch": 0.7028003613369467, - "grad_norm": 0.15981371678518486, + "grad_norm": 0.25185271897899014, "learning_rate": 6.736545278218464e-06, - "loss": 0.5642, + "loss": 0.5698, "step": 389 }, { "epoch": 0.7046070460704607, - "grad_norm": 0.1530064162827602, + "grad_norm": 0.3166682567846798, "learning_rate": 6.661782554026735e-06, - "loss": 0.5584, + "loss": 0.5638, "step": 390 }, { "epoch": 0.7064137308039747, - "grad_norm": 0.16596305105845469, + "grad_norm": 0.2319838260541366, "learning_rate": 6.587318404572537e-06, - "loss": 0.5583, + "loss": 0.564, "step": 391 }, { "epoch": 0.7082204155374887, - "grad_norm": 0.16327470933567317, + "grad_norm": 0.22505306970140945, "learning_rate": 6.513155496266783e-06, - "loss": 0.5641, + "loss": 0.5703, "step": 392 }, { "epoch": 0.7100271002710027, - "grad_norm": 0.1574337263337296, + "grad_norm": 0.24527359143209262, "learning_rate": 6.439296484733526e-06, - "loss": 0.5689, + "loss": 0.575, "step": 393 }, { "epoch": 0.7118337850045167, - "grad_norm": 0.16018951967211098, + "grad_norm": 0.21842675068411357, "learning_rate": 6.3657440147149e-06, - "loss": 0.5716, + "loss": 0.5776, "step": 394 }, { "epoch": 0.7136404697380307, - "grad_norm": 0.14974112027258443, + "grad_norm": 0.22526457972223854, "learning_rate": 6.292500719976397e-06, - "loss": 0.5659, + "loss": 0.5722, "step": 395 }, { "epoch": 0.7154471544715447, - "grad_norm": 0.16158709365351387, + "grad_norm": 0.2386933725648013, "learning_rate": 6.219569223212585e-06, - "loss": 0.56, + "loss": 0.5659, "step": 396 }, { "epoch": 0.7172538392050587, - "grad_norm": 0.15600729659089102, + "grad_norm": 0.2797910217583912, "learning_rate": 6.146952135953158e-06, - "loss": 0.5598, + "loss": 0.5663, "step": 397 }, { "epoch": 0.7190605239385727, - "grad_norm": 0.15620491253414515, + "grad_norm": 0.222440302196531, "learning_rate": 6.074652058469447e-06, - "loss": 0.5616, + "loss": 0.5677, "step": 398 }, { "epoch": 0.7208672086720868, - "grad_norm": 0.16380045583881614, + "grad_norm": 0.2244823119289259, "learning_rate": 6.0026715796812945e-06, - "loss": 0.5644, + "loss": 0.5694, "step": 399 }, { "epoch": 0.7226738934056007, - "grad_norm": 0.14967907113452944, + "grad_norm": 0.22028730249011422, "learning_rate": 5.931013277064377e-06, - "loss": 0.5609, + "loss": 0.567, "step": 400 }, { "epoch": 0.7244805781391147, - "grad_norm": 0.15280123889646421, + "grad_norm": 0.21677756210322688, "learning_rate": 5.859679716557879e-06, - "loss": 0.5626, + "loss": 0.5688, "step": 401 }, { "epoch": 0.7262872628726287, - "grad_norm": 0.16341975790372415, + "grad_norm": 0.4344517020406158, "learning_rate": 5.788673452472622e-06, - "loss": 0.5696, + "loss": 0.5752, "step": 402 }, { "epoch": 0.7280939476061428, - "grad_norm": 0.1562147019531761, + "grad_norm": 0.24869038865191093, "learning_rate": 5.717997027399624e-06, - "loss": 0.5516, + "loss": 0.5567, "step": 403 }, { "epoch": 0.7299006323396567, - "grad_norm": 0.15660229823809, + "grad_norm": 0.23064985584415565, "learning_rate": 5.647652972118998e-06, - "loss": 0.5569, + "loss": 0.5626, "step": 404 }, { "epoch": 0.7317073170731707, - "grad_norm": 0.16555666302974795, + "grad_norm": 0.22343320849286405, "learning_rate": 5.577643805509404e-06, - "loss": 0.5549, + "loss": 0.5606, "step": 405 }, { "epoch": 0.7335140018066847, - "grad_norm": 0.1507767089868411, + "grad_norm": 0.2410416469241811, "learning_rate": 5.507972034457786e-06, - "loss": 0.5615, + "loss": 0.567, "step": 406 }, { "epoch": 0.7353206865401988, - "grad_norm": 0.1557734692497816, + "grad_norm": 0.21682033645252197, "learning_rate": 5.438640153769654e-06, - "loss": 0.5534, + "loss": 0.5594, "step": 407 }, { "epoch": 0.7371273712737128, - "grad_norm": 0.14782949914277185, + "grad_norm": 0.21362885977787827, "learning_rate": 5.369650646079719e-06, - "loss": 0.547, + "loss": 0.5527, "step": 408 }, { "epoch": 0.7389340560072267, - "grad_norm": 0.17200870547934033, + "grad_norm": 0.2501581124737507, "learning_rate": 5.301005981763007e-06, - "loss": 0.5669, + "loss": 0.5723, "step": 409 }, { "epoch": 0.7407407407407407, - "grad_norm": 0.16956590081345058, + "grad_norm": 0.21950559446814377, "learning_rate": 5.2327086188463964e-06, - "loss": 0.5672, + "loss": 0.5733, "step": 410 }, { "epoch": 0.7425474254742548, - "grad_norm": 0.17288972695084148, + "grad_norm": 0.22874967519790235, "learning_rate": 5.164761002920615e-06, - "loss": 0.5695, + "loss": 0.5757, "step": 411 }, { "epoch": 0.7443541102077688, - "grad_norm": 0.15302551358387134, + "grad_norm": 0.23529791112541182, "learning_rate": 5.097165567052646e-06, - "loss": 0.543, + "loss": 0.5478, "step": 412 }, { "epoch": 0.7461607949412827, - "grad_norm": 0.1659332169274283, + "grad_norm": 0.24186795740106914, "learning_rate": 5.029924731698619e-06, - "loss": 0.5586, + "loss": 0.5641, "step": 413 }, { "epoch": 0.7479674796747967, - "grad_norm": 0.14771275975107256, + "grad_norm": 0.22241278304703085, "learning_rate": 4.963040904617131e-06, - "loss": 0.5466, + "loss": 0.5528, "step": 414 }, { "epoch": 0.7497741644083108, - "grad_norm": 0.14811614613987537, + "grad_norm": 0.23230777162136979, "learning_rate": 4.8965164807830355e-06, - "loss": 0.568, + "loss": 0.5735, "step": 415 }, { "epoch": 0.7515808491418248, - "grad_norm": 0.1448610651899535, + "grad_norm": 0.222363917086626, "learning_rate": 4.83035384230169e-06, - "loss": 0.5537, + "loss": 0.5598, "step": 416 }, { "epoch": 0.7533875338753387, - "grad_norm": 0.1705391112022685, + "grad_norm": 0.23822096081299107, "learning_rate": 4.764555358323632e-06, - "loss": 0.5756, + "loss": 0.5815, "step": 417 }, { "epoch": 0.7551942186088527, - "grad_norm": 0.17264531117869172, + "grad_norm": 0.2262092084376523, "learning_rate": 4.699123384959783e-06, - "loss": 0.5576, + "loss": 0.5629, "step": 418 }, { "epoch": 0.7570009033423668, - "grad_norm": 0.15278513072031447, + "grad_norm": 0.23223952114679952, "learning_rate": 4.6340602651970304e-06, - "loss": 0.559, + "loss": 0.5644, "step": 419 }, { "epoch": 0.7588075880758808, - "grad_norm": 0.15921165848077162, + "grad_norm": 0.23091124005107544, "learning_rate": 4.5693683288143884e-06, - "loss": 0.5727, + "loss": 0.5794, "step": 420 }, { "epoch": 0.7606142728093948, - "grad_norm": 0.14827263328330628, + "grad_norm": 0.22340107077805454, "learning_rate": 4.505049892299517e-06, - "loss": 0.5544, + "loss": 0.5611, "step": 421 }, { "epoch": 0.7624209575429087, - "grad_norm": 0.13887513696660272, + "grad_norm": 0.20827372625102367, "learning_rate": 4.441107258765827e-06, - "loss": 0.5578, + "loss": 0.5643, "step": 422 }, { "epoch": 0.7642276422764228, - "grad_norm": 0.15863025954427973, + "grad_norm": 0.8594750186394492, "learning_rate": 4.377542717869959e-06, - "loss": 0.5669, + "loss": 0.5727, "step": 423 }, { "epoch": 0.7660343270099368, - "grad_norm": 0.16888324911277486, + "grad_norm": 0.22572761637619165, "learning_rate": 4.31435854572983e-06, - "loss": 0.5701, + "loss": 0.576, "step": 424 }, { "epoch": 0.7678410117434508, - "grad_norm": 0.1543410763323266, + "grad_norm": 0.22288386378508676, "learning_rate": 4.251557004843116e-06, - "loss": 0.5489, + "loss": 0.5545, "step": 425 }, { "epoch": 0.7696476964769647, - "grad_norm": 0.1535360923186684, + "grad_norm": 0.22039211202209727, "learning_rate": 4.1891403440062495e-06, - "loss": 0.5696, + "loss": 0.576, "step": 426 }, { "epoch": 0.7714543812104788, - "grad_norm": 0.1436847387652519, + "grad_norm": 0.24740997697332476, "learning_rate": 4.1271107982338745e-06, - "loss": 0.541, + "loss": 0.5463, "step": 427 }, { "epoch": 0.7732610659439928, - "grad_norm": 0.14133645853071267, + "grad_norm": 0.21571373017824447, "learning_rate": 4.06547058867883e-06, - "loss": 0.5616, + "loss": 0.5673, "step": 428 }, { "epoch": 0.7750677506775068, - "grad_norm": 0.14833855735489168, + "grad_norm": 0.21461808803522264, "learning_rate": 4.004221922552608e-06, - "loss": 0.5567, + "loss": 0.5626, "step": 429 }, { "epoch": 0.7768744354110207, - "grad_norm": 0.13504301863448828, + "grad_norm": 0.21534749337458045, "learning_rate": 3.943366993046323e-06, - "loss": 0.5573, + "loss": 0.5632, "step": 430 }, { "epoch": 0.7786811201445348, - "grad_norm": 0.15479562167447294, + "grad_norm": 0.22705388420183456, "learning_rate": 3.88290797925218e-06, - "loss": 0.5665, + "loss": 0.5722, "step": 431 }, { "epoch": 0.7804878048780488, - "grad_norm": 0.14971545791291554, + "grad_norm": 0.20634048423367268, "learning_rate": 3.82284704608543e-06, - "loss": 0.5618, + "loss": 0.5677, "step": 432 }, { "epoch": 0.7822944896115628, - "grad_norm": 0.15163059932702447, + "grad_norm": 0.22924008526331077, "learning_rate": 3.763186344206877e-06, - "loss": 0.5694, + "loss": 0.5754, "step": 433 }, { "epoch": 0.7841011743450768, - "grad_norm": 0.14913417270860077, + "grad_norm": 0.19705482767836247, "learning_rate": 3.7039280099458373e-06, - "loss": 0.5678, + "loss": 0.573, "step": 434 }, { "epoch": 0.7859078590785907, - "grad_norm": 0.13874898886438794, + "grad_norm": 0.24321807481345778, "learning_rate": 3.645074165223656e-06, - "loss": 0.5457, + "loss": 0.552, "step": 435 }, { "epoch": 0.7877145438121048, - "grad_norm": 0.15341138739380675, + "grad_norm": 0.2557673843285419, "learning_rate": 3.586626917477724e-06, - "loss": 0.5599, + "loss": 0.5663, "step": 436 }, { "epoch": 0.7895212285456188, - "grad_norm": 0.1396125925965087, + "grad_norm": 0.20741635557070642, "learning_rate": 3.528588359586024e-06, - "loss": 0.5586, + "loss": 0.5646, "step": 437 }, { "epoch": 0.7913279132791328, - "grad_norm": 0.1724528227666086, + "grad_norm": 0.2688295910485162, "learning_rate": 3.4709605697921735e-06, - "loss": 0.5499, + "loss": 0.5556, "step": 438 }, { "epoch": 0.7931345980126467, - "grad_norm": 0.15954102946516008, + "grad_norm": 0.46975916458279, "learning_rate": 3.413745611631009e-06, - "loss": 0.5672, + "loss": 0.5736, "step": 439 }, { "epoch": 0.7949412827461608, - "grad_norm": 0.13471275186125906, + "grad_norm": 0.20860411045588065, "learning_rate": 3.356945533854708e-06, - "loss": 0.5582, + "loss": 0.565, "step": 440 }, { "epoch": 0.7967479674796748, - "grad_norm": 0.1277139415264683, + "grad_norm": 0.20156415753013815, "learning_rate": 3.3005623703594082e-06, - "loss": 0.544, + "loss": 0.5498, "step": 441 }, { "epoch": 0.7985546522131888, - "grad_norm": 0.13269007698962448, + "grad_norm": 0.25558258979943804, "learning_rate": 3.244598140112404e-06, - "loss": 0.5531, + "loss": 0.5598, "step": 442 }, { "epoch": 0.8003613369467028, - "grad_norm": 0.16876313389901235, + "grad_norm": 0.2223887090687036, "learning_rate": 3.1890548470798196e-06, - "loss": 0.5603, + "loss": 0.5666, "step": 443 }, { "epoch": 0.8021680216802168, - "grad_norm": 0.14703779234405664, + "grad_norm": 0.2805594085283592, "learning_rate": 3.1339344801548854e-06, - "loss": 0.5662, + "loss": 0.5722, "step": 444 }, { "epoch": 0.8039747064137308, - "grad_norm": 0.16211778059337975, + "grad_norm": 0.22063158100276353, "learning_rate": 3.079239013086673e-06, - "loss": 0.5541, + "loss": 0.5598, "step": 445 }, { "epoch": 0.8057813911472448, - "grad_norm": 0.13960967725757564, + "grad_norm": 0.20964425176708046, "learning_rate": 3.02497040440948e-06, - "loss": 0.5521, + "loss": 0.5582, "step": 446 }, { "epoch": 0.8075880758807588, - "grad_norm": 0.14347409440257566, + "grad_norm": 0.19879708507272054, "learning_rate": 2.9711305973726404e-06, - "loss": 0.5568, + "loss": 0.5627, "step": 447 }, { "epoch": 0.8093947606142728, - "grad_norm": 0.13685149510690353, + "grad_norm": 0.21371375999064623, "learning_rate": 2.91772151987099e-06, - "loss": 0.5595, + "loss": 0.566, "step": 448 }, { "epoch": 0.8112014453477868, - "grad_norm": 0.14371214973855934, + "grad_norm": 0.21479511072866714, "learning_rate": 2.86474508437579e-06, - "loss": 0.5541, + "loss": 0.5602, "step": 449 }, { "epoch": 0.8130081300813008, - "grad_norm": 0.13902608999443639, + "grad_norm": 0.2507704066850702, "learning_rate": 2.812203187866272e-06, - "loss": 0.5651, + "loss": 0.5702, "step": 450 }, { "epoch": 0.8148148148148148, - "grad_norm": 0.18456478885896846, + "grad_norm": 0.21755447137965794, "learning_rate": 2.7600977117617027e-06, - "loss": 0.566, + "loss": 0.5715, "step": 451 }, { "epoch": 0.8166214995483289, - "grad_norm": 0.13874263584368351, + "grad_norm": 0.20443439016373935, "learning_rate": 2.7084305218540217e-06, - "loss": 0.5626, + "loss": 0.5689, "step": 452 }, { "epoch": 0.8184281842818428, - "grad_norm": 0.13915434189224746, + "grad_norm": 0.22019364090464516, "learning_rate": 2.6572034682410144e-06, - "loss": 0.5547, + "loss": 0.5601, "step": 453 }, { "epoch": 0.8202348690153568, - "grad_norm": 0.14072111536288506, + "grad_norm": 0.23616305038551402, "learning_rate": 2.60641838526008e-06, - "loss": 0.5396, + "loss": 0.5441, "step": 454 }, { "epoch": 0.8220415537488708, - "grad_norm": 0.13282415327633357, + "grad_norm": 0.22116587713371783, "learning_rate": 2.5560770914225356e-06, - "loss": 0.5625, + "loss": 0.5678, "step": 455 }, { "epoch": 0.8238482384823849, - "grad_norm": 0.1472812938618023, + "grad_norm": 0.2246824985093436, "learning_rate": 2.5061813893485087e-06, - "loss": 0.5549, + "loss": 0.5605, "step": 456 }, { "epoch": 0.8256549232158988, - "grad_norm": 0.14223186780386796, + "grad_norm": 0.21936692578708686, "learning_rate": 2.45673306570239e-06, - "loss": 0.5692, + "loss": 0.5749, "step": 457 }, { "epoch": 0.8274616079494128, - "grad_norm": 0.13382708637823343, + "grad_norm": 0.2022912181556414, "learning_rate": 2.407733891128841e-06, - "loss": 0.5612, + "loss": 0.5668, "step": 458 }, { "epoch": 0.8292682926829268, - "grad_norm": 0.13648137063446, + "grad_norm": 0.20494267015114925, "learning_rate": 2.3591856201894124e-06, - "loss": 0.5511, + "loss": 0.5566, "step": 459 }, { "epoch": 0.8310749774164409, - "grad_norm": 0.1547065089258408, + "grad_norm": 0.27106860594647675, "learning_rate": 2.3110899912996993e-06, - "loss": 0.5602, + "loss": 0.5663, "step": 460 }, { "epoch": 0.8328816621499548, - "grad_norm": 0.14643671146676157, + "grad_norm": 0.2046891500263313, "learning_rate": 2.263448726667098e-06, - "loss": 0.548, + "loss": 0.5542, "step": 461 }, { "epoch": 0.8346883468834688, - "grad_norm": 0.13651107636976717, + "grad_norm": 0.23171892826309606, "learning_rate": 2.2162635322291365e-06, - "loss": 0.556, + "loss": 0.5623, "step": 462 }, { "epoch": 0.8364950316169828, - "grad_norm": 0.1368276434381763, + "grad_norm": 0.19018606574785138, "learning_rate": 2.169536097592401e-06, - "loss": 0.5603, + "loss": 0.5668, "step": 463 }, { "epoch": 0.8383017163504969, - "grad_norm": 0.1445464211298085, + "grad_norm": 0.2329448518163566, "learning_rate": 2.1232680959720085e-06, - "loss": 0.5628, + "loss": 0.5692, "step": 464 }, { "epoch": 0.8401084010840109, - "grad_norm": 0.1581727260316551, + "grad_norm": 0.2221258218802644, "learning_rate": 2.0774611841317164e-06, - "loss": 0.5728, + "loss": 0.5789, "step": 465 }, { "epoch": 0.8419150858175248, - "grad_norm": 0.13832300858835891, + "grad_norm": 0.2074746474319042, "learning_rate": 2.0321170023245835e-06, - "loss": 0.5554, + "loss": 0.5612, "step": 466 }, { "epoch": 0.8437217705510388, - "grad_norm": 0.1326854914012355, + "grad_norm": 0.24112788222205025, "learning_rate": 1.9872371742342366e-06, - "loss": 0.5525, + "loss": 0.5588, "step": 467 }, { "epoch": 0.8455284552845529, - "grad_norm": 0.1324315694543524, + "grad_norm": 0.19795185527219575, "learning_rate": 1.942823306916746e-06, - "loss": 0.5629, + "loss": 0.5693, "step": 468 }, { "epoch": 0.8473351400180669, - "grad_norm": 0.14327810150506223, + "grad_norm": 0.1977693992081305, "learning_rate": 1.8988769907430553e-06, - "loss": 0.5522, + "loss": 0.5578, "step": 469 }, { "epoch": 0.8491418247515808, - "grad_norm": 0.1440779207735543, + "grad_norm": 0.2013451884270062, "learning_rate": 1.8553997993420495e-06, - "loss": 0.5485, + "loss": 0.5545, "step": 470 }, { "epoch": 0.8509485094850948, - "grad_norm": 0.1582379305959562, + "grad_norm": 0.21929485106443544, "learning_rate": 1.8123932895441953e-06, - "loss": 0.5777, + "loss": 0.5831, "step": 471 }, { "epoch": 0.8527551942186089, - "grad_norm": 0.13834720570877437, + "grad_norm": 0.2069370487233822, "learning_rate": 1.7698590013258187e-06, - "loss": 0.5486, + "loss": 0.5541, "step": 472 }, { "epoch": 0.8545618789521229, - "grad_norm": 0.13843120298455933, + "grad_norm": 0.21274761276018167, "learning_rate": 1.7277984577539224e-06, - "loss": 0.5676, + "loss": 0.5735, "step": 473 }, { "epoch": 0.8563685636856369, - "grad_norm": 0.1279953110251041, + "grad_norm": 0.20824350625431515, "learning_rate": 1.6862131649316885e-06, - "loss": 0.5629, + "loss": 0.569, "step": 474 }, { "epoch": 0.8581752484191508, - "grad_norm": 0.1585683209243492, + "grad_norm": 0.2118004520239169, "learning_rate": 1.6451046119445174e-06, - "loss": 0.553, + "loss": 0.5585, "step": 475 }, { "epoch": 0.8599819331526649, - "grad_norm": 0.1466497632190738, + "grad_norm": 0.2166189914121313, "learning_rate": 1.6044742708067206e-06, - "loss": 0.5592, + "loss": 0.5655, "step": 476 }, { "epoch": 0.8617886178861789, - "grad_norm": 0.1387550564906503, + "grad_norm": 0.20548635358142273, "learning_rate": 1.5643235964088065e-06, - "loss": 0.5564, + "loss": 0.5625, "step": 477 }, { "epoch": 0.8635953026196929, - "grad_norm": 0.14950588928435643, + "grad_norm": 0.3919523574078014, "learning_rate": 1.524654026465394e-06, - "loss": 0.5671, + "loss": 0.5736, "step": 478 }, { "epoch": 0.8654019873532068, - "grad_norm": 0.14015867212642105, + "grad_norm": 0.2146057433027723, "learning_rate": 1.4854669814637145e-06, - "loss": 0.5532, + "loss": 0.5587, "step": 479 }, { "epoch": 0.8672086720867209, - "grad_norm": 0.1262404297566839, + "grad_norm": 0.2617349364439417, "learning_rate": 1.4467638646127551e-06, - "loss": 0.5513, + "loss": 0.5577, "step": 480 }, { "epoch": 0.8690153568202349, - "grad_norm": 0.12824154810533317, + "grad_norm": 0.22872811738070856, "learning_rate": 1.4085460617930196e-06, - "loss": 0.5622, + "loss": 0.5689, "step": 481 }, { "epoch": 0.8708220415537489, - "grad_norm": 0.1345600337122948, + "grad_norm": 0.2796054869270635, "learning_rate": 1.3708149415068827e-06, - "loss": 0.5661, + "loss": 0.5725, "step": 482 }, { "epoch": 0.8726287262872628, - "grad_norm": 0.13838894371767235, + "grad_norm": 0.19160336089610616, "learning_rate": 1.3335718548296194e-06, - "loss": 0.5504, + "loss": 0.5563, "step": 483 }, { "epoch": 0.8744354110207768, - "grad_norm": 0.13576491587772027, + "grad_norm": 0.2012810744492221, "learning_rate": 1.2968181353609854e-06, - "loss": 0.5472, + "loss": 0.5535, "step": 484 }, { "epoch": 0.8762420957542909, - "grad_norm": 0.14339241553346477, + "grad_norm": 0.21604349922209895, "learning_rate": 1.2605550991775033e-06, - "loss": 0.5576, + "loss": 0.5629, "step": 485 }, { "epoch": 0.8780487804878049, - "grad_norm": 0.131210165981635, + "grad_norm": 0.20846295082414001, "learning_rate": 1.2247840447852993e-06, - "loss": 0.5633, + "loss": 0.5702, "step": 486 }, { "epoch": 0.8798554652213189, - "grad_norm": 0.12636963673731474, + "grad_norm": 0.24538983066246176, "learning_rate": 1.1895062530736412e-06, - "loss": 0.5515, + "loss": 0.5577, "step": 487 }, { "epoch": 0.8816621499548328, - "grad_norm": 0.12776691333292706, + "grad_norm": 0.2027134974758516, "learning_rate": 1.1547229872690407e-06, - "loss": 0.5505, + "loss": 0.5567, "step": 488 }, { "epoch": 0.8834688346883469, - "grad_norm": 0.12702102055757536, + "grad_norm": 0.18716638337399608, "learning_rate": 1.1204354928900496e-06, - "loss": 0.5602, + "loss": 0.5663, "step": 489 }, { "epoch": 0.8852755194218609, - "grad_norm": 0.13494740824410642, + "grad_norm": 0.21298790397676395, "learning_rate": 1.086644997702635e-06, - "loss": 0.5571, + "loss": 0.5625, "step": 490 }, { "epoch": 0.8870822041553749, - "grad_norm": 0.12513086965807382, + "grad_norm": 0.18568745233434789, "learning_rate": 1.0533527116762298e-06, - "loss": 0.5483, + "loss": 0.5551, "step": 491 }, { "epoch": 0.8888888888888888, - "grad_norm": 0.1320750230179298, + "grad_norm": 0.20538517713781662, "learning_rate": 1.0205598269404026e-06, - "loss": 0.55, + "loss": 0.5556, "step": 492 }, { "epoch": 0.8906955736224029, - "grad_norm": 0.12753125629115394, + "grad_norm": 0.21478395379619078, "learning_rate": 9.882675177421662e-07, - "loss": 0.5444, + "loss": 0.5506, "step": 493 }, { "epoch": 0.8925022583559169, - "grad_norm": 0.12278292224063106, + "grad_norm": 0.20451931843479138, "learning_rate": 9.56476940403942e-07, - "loss": 0.5617, + "loss": 0.5672, "step": 494 }, { "epoch": 0.8943089430894309, - "grad_norm": 0.13711394488670506, + "grad_norm": 0.21150501837407576, "learning_rate": 9.251892332821388e-07, - "loss": 0.5584, + "loss": 0.5641, "step": 495 }, { "epoch": 0.8961156278229448, - "grad_norm": 0.13665462597200176, + "grad_norm": 0.20494840081387994, "learning_rate": 8.944055167263982e-07, - "loss": 0.5706, + "loss": 0.5768, "step": 496 }, { "epoch": 0.8979223125564589, - "grad_norm": 0.14101341052067656, + "grad_norm": 0.23208540837310512, "learning_rate": 8.641268930394791e-07, - "loss": 0.5578, + "loss": 0.5637, "step": 497 }, { "epoch": 0.8997289972899729, - "grad_norm": 0.1371088435731868, + "grad_norm": 0.43670103718238795, "learning_rate": 8.34354446437785e-07, - "loss": 0.5571, + "loss": 0.5631, "step": 498 }, { "epoch": 0.9015356820234869, - "grad_norm": 0.15044968278360257, + "grad_norm": 0.22601413932793338, "learning_rate": 8.050892430125362e-07, - "loss": 0.5662, + "loss": 0.5728, "step": 499 }, { "epoch": 0.9033423667570009, - "grad_norm": 0.13190920385566557, + "grad_norm": 0.19668433127008136, "learning_rate": 7.763323306916037e-07, - "loss": 0.565, + "loss": 0.5715, "step": 500 }, { "epoch": 0.9051490514905149, - "grad_norm": 0.1330272682028843, + "grad_norm": 0.2755002021494048, "learning_rate": 7.48084739201978e-07, - "loss": 0.5629, + "loss": 0.5691, "step": 501 }, { "epoch": 0.9069557362240289, - "grad_norm": 0.12596324805796347, + "grad_norm": 0.3171350403732765, "learning_rate": 7.20347480032893e-07, - "loss": 0.5575, + "loss": 0.5636, "step": 502 }, { "epoch": 0.9087624209575429, - "grad_norm": 0.12269256945275395, + "grad_norm": 0.19229422822362555, "learning_rate": 6.931215463996204e-07, - "loss": 0.5536, + "loss": 0.5592, "step": 503 }, { "epoch": 0.9105691056910569, - "grad_norm": 0.12397511410433176, + "grad_norm": 0.27683701338372546, "learning_rate": 6.664079132078881e-07, - "loss": 0.5693, + "loss": 0.5751, "step": 504 }, { "epoch": 0.912375790424571, - "grad_norm": 0.12830241052252903, + "grad_norm": 0.2809384935828692, "learning_rate": 6.402075370189914e-07, - "loss": 0.541, + "loss": 0.5459, "step": 505 }, { "epoch": 0.9141824751580849, - "grad_norm": 0.13190860459526532, + "grad_norm": 0.19763837414051177, "learning_rate": 6.145213560155167e-07, - "loss": 0.5613, + "loss": 0.5665, "step": 506 }, { "epoch": 0.9159891598915989, - "grad_norm": 0.13145720432300395, + "grad_norm": 0.19330852387204572, "learning_rate": 5.893502899677622e-07, - "loss": 0.5597, + "loss": 0.5655, "step": 507 }, { "epoch": 0.9177958446251129, - "grad_norm": 0.1390726475782522, + "grad_norm": 0.18671045701948052, "learning_rate": 5.646952402007993e-07, - "loss": 0.5523, + "loss": 0.5578, "step": 508 }, { "epoch": 0.919602529358627, - "grad_norm": 0.13621784896319666, + "grad_norm": 0.1935404379381202, "learning_rate": 5.405570895622014e-07, - "loss": 0.5716, + "loss": 0.5771, "step": 509 }, { "epoch": 0.9214092140921409, - "grad_norm": 0.1303455229261674, + "grad_norm": 0.2266010335193308, "learning_rate": 5.169367023904226e-07, - "loss": 0.5547, + "loss": 0.5612, "step": 510 }, { "epoch": 0.9232158988256549, - "grad_norm": 0.12456382952524406, + "grad_norm": 0.18970071959766807, "learning_rate": 4.938349244838541e-07, - "loss": 0.5452, + "loss": 0.5503, "step": 511 }, { "epoch": 0.9250225835591689, - "grad_norm": 0.13033855895306248, + "grad_norm": 0.18751450133804787, "learning_rate": 4.7125258307053385e-07, - "loss": 0.5725, + "loss": 0.5788, "step": 512 }, { "epoch": 0.926829268292683, - "grad_norm": 0.12446562507353466, + "grad_norm": 0.20331269139776248, "learning_rate": 4.491904867785329e-07, - "loss": 0.5548, + "loss": 0.5604, "step": 513 }, { "epoch": 0.928635953026197, - "grad_norm": 0.13470648707570285, + "grad_norm": 0.23016504488814535, "learning_rate": 4.276494256069874e-07, - "loss": 0.5512, + "loss": 0.5567, "step": 514 }, { "epoch": 0.9304426377597109, - "grad_norm": 0.12820525405710728, + "grad_norm": 0.19554882132706672, "learning_rate": 4.0663017089782593e-07, - "loss": 0.5567, + "loss": 0.5625, "step": 515 }, { "epoch": 0.9322493224932249, - "grad_norm": 0.1285059094961294, + "grad_norm": 0.2141915434693991, "learning_rate": 3.8613347530813527e-07, - "loss": 0.5649, + "loss": 0.5705, "step": 516 }, { "epoch": 0.934056007226739, - "grad_norm": 0.12804520061809396, + "grad_norm": 0.19325511174233914, "learning_rate": 3.6616007278320985e-07, - "loss": 0.5529, + "loss": 0.5589, "step": 517 }, { "epoch": 0.935862691960253, - "grad_norm": 0.13304365719456168, + "grad_norm": 0.2867842616135943, "learning_rate": 3.467106785302865e-07, - "loss": 0.5545, + "loss": 0.5597, "step": 518 }, { "epoch": 0.9376693766937669, - "grad_norm": 0.1361684924628673, + "grad_norm": 0.1965457021012975, "learning_rate": 3.277859889929147e-07, - "loss": 0.5444, + "loss": 0.5506, "step": 519 }, { "epoch": 0.9394760614272809, - "grad_norm": 0.1308177622907266, + "grad_norm": 0.1983487205786454, "learning_rate": 3.0938668182603656e-07, - "loss": 0.5538, + "loss": 0.5599, "step": 520 }, { "epoch": 0.941282746160795, - "grad_norm": 0.12589685134140113, + "grad_norm": 0.2016483308305746, "learning_rate": 2.9151341587169976e-07, - "loss": 0.5614, + "loss": 0.5676, "step": 521 }, { "epoch": 0.943089430894309, - "grad_norm": 0.1309203120133497, + "grad_norm": 0.18584462016114534, "learning_rate": 2.7416683113548945e-07, - "loss": 0.5568, + "loss": 0.5628, "step": 522 }, { "epoch": 0.9448961156278229, - "grad_norm": 0.11901547198938607, + "grad_norm": 0.1978444406310148, "learning_rate": 2.5734754876358846e-07, - "loss": 0.5478, + "loss": 0.5537, "step": 523 }, { "epoch": 0.9467028003613369, - "grad_norm": 0.13909322877084576, + "grad_norm": 0.36806346840727133, "learning_rate": 2.41056171020555e-07, - "loss": 0.5577, + "loss": 0.5638, "step": 524 }, { "epoch": 0.948509485094851, - "grad_norm": 0.1250829734778908, + "grad_norm": 0.21280062488260573, "learning_rate": 2.2529328126774496e-07, - "loss": 0.5511, + "loss": 0.557, "step": 525 }, { "epoch": 0.950316169828365, - "grad_norm": 0.12347363002885929, + "grad_norm": 0.19466678941893495, "learning_rate": 2.1005944394242694e-07, - "loss": 0.5555, + "loss": 0.5604, "step": 526 }, { "epoch": 0.952122854561879, - "grad_norm": 0.14522071992485233, + "grad_norm": 0.21086407350506756, "learning_rate": 1.9535520453756506e-07, - "loss": 0.5556, + "loss": 0.562, "step": 527 }, { "epoch": 0.9539295392953929, - "grad_norm": 0.12689680120062793, + "grad_norm": 0.30098595384873267, "learning_rate": 1.8118108958229795e-07, - "loss": 0.5536, + "loss": 0.5596, "step": 528 }, { "epoch": 0.955736224028907, - "grad_norm": 0.12751446868474398, + "grad_norm": 0.19293881839582736, "learning_rate": 1.6753760662307217e-07, - "loss": 0.56, + "loss": 0.566, "step": 529 }, { "epoch": 0.957542908762421, - "grad_norm": 0.1377057005084481, + "grad_norm": 0.2178986422135447, "learning_rate": 1.5442524420547666e-07, - "loss": 0.5515, + "loss": 0.5576, "step": 530 }, { "epoch": 0.959349593495935, - "grad_norm": 0.12467052220580256, + "grad_norm": 0.2742424638712097, "learning_rate": 1.4184447185674354e-07, - "loss": 0.5642, + "loss": 0.5712, "step": 531 }, { "epoch": 0.9611562782294489, - "grad_norm": 0.12474201474426573, + "grad_norm": 0.20441449476859883, "learning_rate": 1.297957400689398e-07, - "loss": 0.5532, + "loss": 0.5589, "step": 532 }, { "epoch": 0.9629629629629629, - "grad_norm": 0.1378124489625079, + "grad_norm": 0.2028488844322694, "learning_rate": 1.1827948028283353e-07, - "loss": 0.5578, + "loss": 0.564, "step": 533 }, { "epoch": 0.964769647696477, - "grad_norm": 0.12241633221352105, + "grad_norm": 0.19641796792115376, "learning_rate": 1.0729610487243968e-07, - "loss": 0.554, + "loss": 0.56, "step": 534 }, { "epoch": 0.966576332429991, - "grad_norm": 0.12161233750054955, + "grad_norm": 0.1961644503643991, "learning_rate": 9.684600713026847e-08, - "loss": 0.5549, + "loss": 0.5613, "step": 535 }, { "epoch": 0.9683830171635049, - "grad_norm": 0.12789778109710662, + "grad_norm": 0.19348594352650886, "learning_rate": 8.692956125322837e-08, - "loss": 0.5637, + "loss": 0.5693, "step": 536 }, { "epoch": 0.9701897018970189, - "grad_norm": 0.12936381253059515, + "grad_norm": 0.26854880925929886, "learning_rate": 7.754712232922678e-08, - "loss": 0.5699, + "loss": 0.5761, "step": 537 }, { "epoch": 0.971996386630533, - "grad_norm": 0.14000238842546384, + "grad_norm": 0.20691432526444847, "learning_rate": 6.869902632447022e-08, - "loss": 0.5537, + "loss": 0.5591, "step": 538 }, { "epoch": 0.973803071364047, - "grad_norm": 0.12479645822662797, + "grad_norm": 0.18552138088036305, "learning_rate": 6.038559007141397e-08, - "loss": 0.5509, + "loss": 0.5564, "step": 539 }, { "epoch": 0.975609756097561, - "grad_norm": 0.12949980004311576, + "grad_norm": 0.19379985533649824, "learning_rate": 5.260711125743445e-08, - "loss": 0.5512, + "loss": 0.5575, "step": 540 }, { "epoch": 0.9774164408310749, - "grad_norm": 0.12362590360829781, + "grad_norm": 0.1915865386485388, "learning_rate": 4.536386841415774e-08, - "loss": 0.56, + "loss": 0.5667, "step": 541 }, { "epoch": 0.979223125564589, - "grad_norm": 0.1318525107421328, + "grad_norm": 0.2671463373041046, "learning_rate": 3.8656120907489266e-08, - "loss": 0.5524, + "loss": 0.5582, "step": 542 }, { "epoch": 0.981029810298103, - "grad_norm": 0.12341218300550963, + "grad_norm": 0.1885606186425532, "learning_rate": 3.24841089283312e-08, - "loss": 0.5661, + "loss": 0.5718, "step": 543 }, { "epoch": 0.982836495031617, - "grad_norm": 0.1313814849485639, + "grad_norm": 0.2328688039271822, "learning_rate": 2.684805348397268e-08, - "loss": 0.5653, + "loss": 0.5714, "step": 544 }, { "epoch": 0.9846431797651309, - "grad_norm": 0.1246027162262715, + "grad_norm": 0.1940328082205833, "learning_rate": 2.1748156390186125e-08, - "loss": 0.5679, + "loss": 0.5744, "step": 545 }, { "epoch": 0.986449864498645, - "grad_norm": 0.12930758713143098, + "grad_norm": 0.20170122816300584, "learning_rate": 1.718460026399138e-08, - "loss": 0.5575, + "loss": 0.5632, "step": 546 }, { "epoch": 0.988256549232159, - "grad_norm": 0.12478369737661307, + "grad_norm": 0.19578981489849026, "learning_rate": 1.3157548517124251e-08, - "loss": 0.5697, + "loss": 0.5759, "step": 547 }, { "epoch": 0.990063233965673, - "grad_norm": 0.1311592043703811, + "grad_norm": 0.24051446818881467, "learning_rate": 9.66714535017954e-09, - "loss": 0.5666, + "loss": 0.5724, "step": 548 }, { "epoch": 0.991869918699187, - "grad_norm": 0.12850976653433435, + "grad_norm": 0.18847855702618435, "learning_rate": 6.7135157474501604e-09, - "loss": 0.5611, + "loss": 0.5674, "step": 549 }, { "epoch": 0.993676603432701, - "grad_norm": 0.1316930656860681, + "grad_norm": 0.2042654954909486, "learning_rate": 4.29676547244906e-09, - "loss": 0.5511, + "loss": 0.5569, "step": 550 }, { "epoch": 0.995483288166215, - "grad_norm": 0.12869152675188197, + "grad_norm": 0.1884805767867668, "learning_rate": 2.416981064127244e-09, - "loss": 0.57, + "loss": 0.5764, "step": 551 }, { "epoch": 0.997289972899729, - "grad_norm": 0.13230005405798148, + "grad_norm": 0.25949043814500883, "learning_rate": 1.074229833769591e-09, - "loss": 0.5523, + "loss": 0.5574, "step": 552 }, { "epoch": 0.999096657633243, - "grad_norm": 0.13151662587114868, + "grad_norm": 0.21033436815555134, "learning_rate": 2.6855986258844666e-10, - "loss": 0.5559, + "loss": 0.5612, "step": 553 } ],